PDA

View Full Version : Maxwell Render 2.0


Pages : [1] 2

Gadzooks
05-11-2009, 12:25 PM
Next limit is now in the process of revealing some of the new features on the next release of Maxwell 2.0

http://www.maxwellrender.com/version2/

AJ
05-11-2009, 02:16 PM
It's great that it's getting faster & smoother... But 2 hours for that untextured interior to render at 1000x550 is still utter madness to me.

ienrdna
05-11-2009, 02:29 PM
New pics without the noise look cg to me.

Stimpy
05-11-2009, 02:43 PM
New pics without the noise look cg to me.

just add some noise in comp.
voila.

mister3d
05-11-2009, 02:52 PM
just add some noise in comp.
voila.

No... it's not about the noise, they really look more cg. Looks like they sacrificed some quality for the sake of speed. And I would say it's a doubteful decision because it lost a lot in terms of a realism.
ALL pictures look way less realistic, and I would say they look more like vray renders now. What's the point in maxwell then.

Sandr0
05-11-2009, 03:10 PM
Yup, definitely. Looks like there's lot more light bouncing going on in older renders, causing bright highlights here and there and different tones.

lo
05-11-2009, 03:16 PM
it's no secret that adding noise to a crappy render will make it look more realistic. Not better, just more believable as photography.

mister3d
05-11-2009, 03:22 PM
Yup, definitely. Looks like there's lot more light bouncing going on in older renders, causing bright highlights here and there and different tones.

I guess it's about shaders, but who knows.

it's no secret that adding noise to a crappy render will make it look more realistic. Not better, just more believable as photography.

I think it's not so, and maxwell was realistic not because of noise. Or at least if you show some examples where noise added some realism.

R10k
05-11-2009, 03:23 PM
ALL pictures look way less realistic, and I would say they look more like vray renders now. What's the point in maxwell then.

The grain on the 1.7 renders diffuse the level of contrast in places. Obviously as well, adding grain to a render will often make it look less perfect. So, because of those two things, a couple of the v2 renders do look more CG, because the contrast doesn't appear as subtle, and the renders look far too clean. I don't think it's more than that. A number of those renders don't look that crash-hot to start with, so perhaps they were simply chosen to show the amount of grain that's reduced.

Whether Maxwell is capable of more than Vray, I have no idea. But, I don't think those renders are the best ones to answer that question.

Magnus3D
05-11-2009, 03:29 PM
You cannot judge the performance and quality based on those few images alone. Be patient and wait for all the other reveals and until it's actually finished, then make a fair judgement of it's quality and performance again. Now you guys are only speculating and guessing.

/ Magnus

DanielWray
05-11-2009, 03:38 PM
I won't make a decision based on those images, i mean i'm not overly impressed, but then again it's just demonstrating speed and noise reduction.

I'm going to wait and see what else they have to offer, to be honest i won't be buying it, i don't use any software that supports it and i have lux render (although it's not really upto maxwells speeds). But still, i look forward too what they have to offer :)

TAVO
05-11-2009, 03:40 PM
The Hardware specs in those images are mainly high performance, so you still need that to have a decent image in decent time.

AJ
05-11-2009, 03:40 PM
Now you guys are only speculating and guessing.Yes but if we weren't doing that then we'd be doing stupid things, like work.

alexyork
05-11-2009, 03:50 PM
2 hours for that clay interior render? no thanks...

Stimpy
05-11-2009, 03:57 PM
Now you guys are only speculating and guessing.

you must be new here..

mister3d
05-11-2009, 04:04 PM
The grain on the 1.7 renders diffuse the level of contrast in places. Obviously as well, adding grain to a render will often make it look less perfect. So, because of those two things, a couple of the v2 renders do look more CG, because the contrast doesn't appear as subtle, and the renders look far too clean. I don't think it's more than that. A number of those renders don't look that crash-hot to start with, so perhaps they were simply chosen to show the amount of grain that's reduced.

Whether Maxwell is capable of more than Vray, I have no idea. But, I don't think those renders are the best ones to answer that question.

I really want to believe this, but it's funny that the level of contrast changes the look so dramatically. Those subtle gradations look more convincing. Though yes I kow that th noise changes the brightness of elements. So they should place 2 images with the same level of noise to see whether there's a difference in versions.

You cannot judge the performance and quality based on those few images alone. Be patient and wait for all the other reveals and until it's actually finished, then make a fair judgement of it's quality and performance again. Now you guys are only speculating and guessing.

/ Magnus
Not quite, as they provided the images... but you're right, maybe there will be other more fair comparisons.

R10k
05-11-2009, 04:06 PM
I really want to believe this, but it's funny that the level of contrast changes the look so dramatically.

That's contrast for you :)

alexyork
05-11-2009, 04:07 PM
speculation...?

the dev team must know what they're doing, right? they wrote this software.

if they think 2 hours to render a simple white room with minimal detail, no textures and 1 sunlight at low res is acceptable then that says a great deal about this software and its potential, or lack thereof.

JML
05-11-2009, 05:02 PM
Yup, definitely. Looks like there's lot more light bouncing going on in older renders, causing bright highlights here and there and different tones.

yes, for example in the render 3, the blue light from the sky does not reach the floor anymore,
looks like less bounces..
I agree with others, the images from 1.7 looks better, and other renderers can achieve the same quality with better render times..
2 hours on a i7 for image 4, ouch!

ZacD
05-11-2009, 05:21 PM
you must be new here..

ROFL thats what the internet is for, guessing and speculating

vlad
05-11-2009, 07:18 PM
speculation...?

the dev team must know what they're doing, right? they wrote this software.

if they think 2 hours to render a simple white room with minimal detail, no textures and 1 sunlight at low res is acceptable then that says a great deal about this software and its potential, or lack thereof.

It seems to appeal to people who dont like tweaking render settings/lighting/materials and/or have lots of spare time to render and/or have access to a sizable renderfarm. And there seems to be a market for that. Altho I'm no advocate of unbiased renderers in their present state, there most probably will come a time when tech will allow them to be viable solutions.

Bubbaloo
05-11-2009, 07:56 PM
The renders presented are not beauty renders, but renders that are made to show noise clearance in areas that used to take longer to clear up in Version 1. The 2x - 6x speed increases shown on the site are impressive in themselves, but couple that with the rapidly growing speed of hardware and you must ask yourself in a few years, is there really a need for interpolation in rendering? Rendering will be more of a scientific simulation mixed with art, than pictures made from spare parts as biased renderers are doing now. In fact Maxwell is doing just that even now (scientific simulation). It's way ahead of it's time and waiting for the world of computer technology to catch up. That's my take on it anyway. Opinions vary.

ienrdna
05-11-2009, 11:14 PM
The renders presented are not beauty renders, but renders that are made to show noise clearance in areas that used to take longer to clear up in Version 1. The 2x - 6x speed increases shown on the site are impressive in themselves, but couple that with the rapidly growing speed of hardware and you must ask yourself in a few years, is there really a need for interpolation in rendering? Rendering will be more of a scientific simulation mixed with art, than pictures made from spare parts as biased renderers are doing now. In fact Maxwell is doing just that even now (scientific simulation). It's way ahead of it's time and waiting for the world of computer technology to catch up. That's my take on it anyway. Opinions vary.


For a still image? Sure. But for feature? NPR rendering? Compositing?

Mr.Tyler Durden
05-11-2009, 11:36 PM
The renders presented are not beauty renders, but renders that are made to show noise clearance in areas that used to take longer to clear up in Version 1. The 2x - 6x speed increases shown on the site are impressive in themselves, but couple that with the rapidly growing speed of hardware and you must ask yourself in a few years, is there really a need for interpolation in rendering? Rendering will be more of a scientific simulation mixed with art, than pictures made from spare parts as biased renderers are doing now. In fact Maxwell is doing just that even now (scientific simulation). It's way ahead of it's time and waiting for the world of computer technology to catch up. That's my take on it anyway. Opinions vary.

I second that for 100 percent.

all biased renderers had their source of dev some years ago, where hardware wasn't capable and insufficient. They surely have their right of existence even nowadays.

Today, to be honest, I still am surprised (and was since the early betas of maxwell) about the level of realism and quality reached.

cheers

Tyler

DanielWray
05-11-2009, 11:36 PM
I think maxwell defintley has it's place in the market. Perhaps it does need a little kick up the rear in terms of speed, like all unbiased render engines, but i wouldn't think it'd be an easy task keeping a render ubiased and impleneting speed up hacks etc.

For arch viz work, or product display, where your maybe showing off a few diffrent still images and you wan't to show to the client(s) a nearly true life like representation then you probably couldn't go wrong with maxwell, i mean let's face it, not al CG people (from artists to Arch viz, to product designers) understand or are capable of producing true to life renders using hand made light setups etc etc. So to have software like this, even if it's currently a little slow, is surely a great thing.

It would be nice to have some sort of plugin that uses a very low setting with low samples to roughly preview a scene before you render though.. perhaps cuastics or intels new hardware will allow this.

mister3d
05-12-2009, 03:08 AM
There is no doubt that the future is for renderers like maxwell.

Jozvex
05-12-2009, 03:13 AM
if they think 2 hours to render a simple white room with minimal detail, no textures and 1 sunlight at low res is acceptable then that says a great deal about this software and its potential, or lack thereof.

I think if any artist decides to use Maxwell to render such a scene it's more of a reflection on them, not the software! Completely diffuse white scenes are always going to take ages without a bounce cutoff or some interpolation because the light just keeps on bouncing for ages and you can clearly see unresolved areas. Stick textures in that scene and it would look far, far better and in less time. It's just a noise resolving demo!

CHRiTTeR
05-12-2009, 09:13 AM
Its weird that this is supposed to be a 'physical correct' and 'unbiased' renderer but everytime they change the code to make it faster or 'enhance' the engine it gives a difference in lighting and shading...

Conclusion: not so verry unbiased and physical correct if you ask me ;)

The first (and verry buggy) verison of maxwell gave much more realistic results then later versions.

Now you guys are only speculating and guessing.

/ Magnus

How is this speculating and guessing?!
We have comparable A-B examples of the same scenes. posted by the devs themselves.

Sandr0
05-12-2009, 10:28 AM
There is no doubt that the future is for renderers like maxwell.

I vote for more flexible, hands on tools. "Simulation" approach pretty much takes all fun and challenge out of lighting/rendering/shading and can only get you so far.

my 2 cents...

Stimpy
05-12-2009, 10:50 AM
I vote for more flexible, hands on tools. "Simulation" approach pretty much takes all fun and challenge out of lighting/rendering/shading and can only get you so far.

my 2 cents...

ye god forbid architects actually want correct results in their renderings...

alexyork
05-12-2009, 10:59 AM
I think if any artist decides to use Maxwell to render such a scene it's more of a reflection on them, not the software! Completely diffuse white scenes are always going to take ages without a bounce cutoff or some interpolation because the light just keeps on bouncing for ages and you can clearly see unresolved areas. Stick textures in that scene and it would look far, far better and in less time. It's just a noise resolving demo!

i'm not complaining about the quality of the image. yes these are indeed just noise-resolving demos. but 2 hours to render that scene can't possibly be practical in any real-world application.

i've used maxwell a couple of times in the past and I can tell you, once you press render you better pray your client comes back saying "perfect, it's finished", because if they don't (and when do they ever?), asking you to move a light or add something, you're looking at a complete re-render taking many hours. that can cost you your reputation and future business.

it seems that the vast majority of people using these so-called "unbiased" renderers (although as many have pointed out this is not quite the case) are bedroom artists producing personal work or studios with a whopping great render farm and time to burn doing nice tests. there's a reason for that. it's just not useable in day-to-day commercial work.

what studio would take on a renderer (or any tool for that matter) that would force them to wait for, say, 9 hours for a render to look acceptable when they can use an alternative that will achieve the same quality in a fraction of the time? especially in the current climate when fast turnarounds are commonplace and cost is hugely important.

yes, in the future, when we're all running a farm of 8-cpu octocore machines something like this might be feasible. but that's a long way down the line.

i also don't know why everyone is falling over themselves for maxwell's apparent amazing realism. you can achieve the same level of realism with any of the other renderers if you know what you're doing and have a strong foundation in photography and the way light works in nature/reality. and guess what, your stuff won't take 2 days to render either.

just my opinions and experiences based on trying to use this thing in production.

alexyork
05-12-2009, 11:04 AM
ye god forbid architects actually want correct results in their renderings...

it's about balance. they want realism but inevitably someone (usually the client) will request that you do something that will be completely unrealistic. an obvious and common example is making an interior as bright as the exterior in a sunlit day shot. yes, photographers often blend exposures to achieve this, but the result still often jars with the viewer. or they ask that you make an area in shadow much brighter, making the contrast in the image unrealistic or just unattractive. or they ask you to under expose the BG, whatever it might be. It's very rare in my experience that you ever produce your ideal render based almost entirely on photographic principles and the client is happy with it as-is...

mister3d
05-12-2009, 11:05 AM
I vote for more flexible, hands on tools. "Simulation" approach pretty much takes all fun and challenge out of lighting/rendering/shading and can only get you so far.

my 2 cents...

Not "all" fun. There will be another techniques used like in a real-life lighting with GI. Anyway I meant the algorithms that are used in maxwell, especially the shaders.

Magnus3D
05-12-2009, 11:09 AM
Just because you guys don't like Maxwell and the whole 'unbiased' approach to rendering does not mean it's useless and that no one else can use it besides those so called "bedroom artists". That's plain stupidity to say such a thing, such a comment is based upon nothing. There are many of us Maxwell users who earn a good living with many satisfied clients, and far from everyone owns a renderfarm, i'm myself only using one quadcore machine to produce all my work with and i had no need for a farm.

Please don't just bash this just because you can and you feel like it, because it's not Vray or mental ray. Yes it's different and it's slower than any other technology out there, but Maxwell has a very strong userbase who are both dedicated and professional. We are no second class citicens just because we enjoy another renderengine and tech.

/ Magnus

mister3d
05-12-2009, 11:30 AM
i also don't know why everyone is falling over themselves for maxwell's apparent amazing realism. you can achieve the same level of realism with any of the other renderers if you know what you're doing and have a strong foundation in photography and the way light works in nature/reality. and guess what, your stuff won't take 2 days to render either.


We ran a test here with a simple BALL IN A BOX, lit with one area light. it took me 2 hours to replicate the look of a maxwell render. There were 3 guys testing in, one in vray (me), the other in mental ray and the third in maxwell. So this showed obviously that maxwell produces far more realistic results. I think it's about their shader algorithms. It's not that simple as you think.

playmesumch00ns
05-12-2009, 11:33 AM
Maxwell is still capable of producing absolute crap in the hands of someone who doesn't know what they're doing, same as any other renderer.

I'm curious about the contrast differences between the versions as well. Since they touted themselves as being the most physically accurate solution on the market before they obviously got their maths wrong. Or was it right before and is wrong now?

DanielWray
05-12-2009, 11:38 AM
I vote for more flexible, hands on tools. "Simulation" approach pretty much takes all fun and challenge out of lighting/rendering/shading and can only get you so far.

my 2 cents...

Just use another renderer than, you know each to there own and everything. While i agree with you, i love having full control and i personally like to make everything in an image, otherwise it does feel a little like cheating.

But there are instances where maxwell is needed and unbiased renderers alike, let's face it though, you won't be using this to make a feature film or short animation so why ask for a more flexible tool when it's main job isn't aimed at that specific job in the first place.

Jozvex
05-12-2009, 11:43 AM
i'm not complaining about the quality of the image. yes these are indeed just noise-resolving demos. but 2 hours to render that scene can't possibly be practical in any real-world application.

Fortunately I don't know of anyone asking for unbiased occlusion-pass-esque renders. It's not a reliable indication of speed because it's a specifically slow and difficult to resolve setup. It's an impractical result born from an impractical task.

i've used maxwell a couple of times in the past and I can tell you, once you press render you better pray your client comes back saying "perfect,

Instead of praying, I think most of the 80+ companies on the client list (I'd link to it but I think it's flash based, it's on the main website though if you'd like to see it) have adopted practical workflows based on the tool they've chosen to use. A couple of times may have been enough for you, but some of us have been using it for years quite successfully.

it seems that the vast majority of people using these so-called "unbiased" renderers (although as many have pointed out this is not quite the case) are bedroom artists producing personal work or studios with a whopping great render farm and time to burn doing nice tests. there's a reason for that. it's just not useable in day-to-day commercial work.

Honestly... it's unbiased. Unbiased doesn't mean the divine powers up above gave the developers *the* ultimate quantum level equations. Of course they can improve upon them with newer versions. As for the bedroom artists thing, again I wish I could link you to the clients list.

i also don't know why everyone is falling over themselves for maxwell's apparent amazing realism. you can achieve the same level of realism with any of the other renderers if you know what you're doing and have a strong foundation in photography and the way light works in nature/reality. and guess what, your stuff won't take 2 days to render either.

Can you sit a real world photographer in front of mental ray, vray or what have you and honestly expect them to light and render a scene well, unless they've had many hours of training on raytracing/GI/occlusion/IBL/antialiasing/etc and the different benefits of lambert/phong/blinn/anisotropic/etc shaders? 'Cause you can sit one down in front of Maxwell and they pretty much just get it. Quite a number of the userbase come from traditional backgrounds and love Maxwell because you can go "oh, for an indoor scene I'd normally use 100 watt lightbulbs and ISO of 400" and those values plug straight into it.

just my opinions and experiences based on trying to use this thing in production.

Same here. :wavey:

P.S. I'm not trying to be uber-defensive I'm just saying, people use it because it does actually work and do what it claims to.

dmeyer
05-12-2009, 11:47 AM
i'm not complaining about the quality of the image. yes these are indeed just noise-resolving demos. but 2 hours to render that scene can't possibly be practical in any real-world application.

i've used maxwell a couple of times in the past and I can tell you, once you press render you better pray your client comes back saying "perfect, it's finished", because if they don't (and when do they ever?), asking you to move a light or add something, you're looking at a complete re-render taking many hours. that can cost you your reputation and future business.

it seems that the vast majority of people using these so-called "unbiased" renderers (although as many have pointed out this is not quite the case) are bedroom artists producing personal work or studios with a whopping great render farm and time to burn doing nice tests. there's a reason for that. it's just not useable in day-to-day commercial work.

what studio would take on a renderer (or any tool for that matter) that would force them to wait for, say, 9 hours for a render to look acceptable when they can use an alternative that will achieve the same quality in a fraction of the time? especially in the current climate when fast turnarounds are commonplace and cost is hugely important.

yes, in the future, when we're all running a farm of 8-cpu octocore machines something like this might be feasible. but that's a long way down the line.

i also don't know why everyone is falling over themselves for maxwell's apparent amazing realism. you can achieve the same level of realism with any of the other renderers if you know what you're doing and have a strong foundation in photography and the way light works in nature/reality. and guess what, your stuff won't take 2 days to render either.

just my opinions and experiences based on trying to use this thing in production.

Anyone taking 9 hours per frame in Maxwell is either rendering billboards or doesn't know how to optimize. Yes there are some things to avoid in Maxwell to keep your render times down, but that is just like any other renderer.

We've been using it for 2 years, rendering animation at 2k. I have yet to see a frame take more than 2 hours.

alexyork
05-12-2009, 12:00 PM
perhaps things have changed in the maxwell camp over the last few years then, and you can expect a high quality render in more practical times, but my experience with it (about a year ago was the last time i think) was that it was unusable in production unless your client has 0 render-based comments. (and yes I did dig deep into how to use it properly i wasn't just messing around).

again, that's just my experience. i'm glad you folks who are using it in production are happy with it. but why don't the dev team post up some actual proper production examples along with rendertimes? i still have my doubts about what actual benefit this software gives you.

Sandr0
05-12-2009, 12:10 PM
ye god forbid architects actually want correct results in their renderings...
Your point is?
If by correct results you mean correct looking images, then it's absolutely possible to do so without unbiased engines and long rendertimes.
If by correct results you mean some sort of simulation data, then it has more to do with CAD and science world, not necessarily archviz artists.

@mister3d

Well, i'm not competent enough to talk about algorithms behind maxwell shaders.
It's true that faking stuff doesn't make sense unless you have to. It's also true that there's no need to do additional computing unless necessary. As computers get powerful enough there will be less reasons to fake/simplify shaders and stuff. That's kind of obvious, but it has nothing to do with maxwell or similar renderers :)

@DanielWray (http://forums.cgsociety.org/member.php?u=207548)

Yes, I agree. Maxwell has it's specific market, and it's cool at what it does. I'm not maxwell customer and don't ask to turn it into more flexible tool - it was just my response to comment saying that future is for unbiased renderengines.

Jozvex
05-12-2009, 12:18 PM
why don't the dev team post up some actual proper production examples along with rendertimes?

Heh, they're doing the whole 'slow reveal' thing. Releasing renders showing particular new features/improvements one at a time. Apparently it was quite hard to find renders *only* displaying speedups and not any of the other new stuff.

i still have my doubts about what actual benefit this software gives you.

That's cool, and hey it may not benefit you at all. It's not the only renderer I'm using (also using mental ray and 3Delight quite happily) but when I need it I love it. The plugins have certainly come a long way since you last used Maxwell I'm sure, I use the Maya integration 95% of the time over Maxwell Studio.

I admit it, I'm a 4th level fanboy on the Maxwell tester team. Step back or I may accidentally render you unconscious.. over a few hours, but the lighting will be lovely.

alexyork
05-12-2009, 12:23 PM
Step back or I may accidentally render you unconscious.. over a few hours, but the lighting will be lovely.

heh that's a quality line :)

mister3d
05-12-2009, 12:26 PM
t was just my response to comment saying that future is for unbiased renderengines.

I didn't mean the future is for unbiased renderers, but that the future is for such algorithms of materials currently not many renderers support, and maxwell is one of those. That's what excites me about rendering future.

Sandr0
05-12-2009, 12:33 PM
@mister3d

Well, I believe maxwell, vray, mentalray e.t.c. shaders are actually based on same papers and algorithms. It's just that maxwell doesn't cut corners.

Again, my 2 cents. Perhaps someone more experienced in shader magic can shed some light on this topic =)

peace :)

Magnus3D
05-12-2009, 12:44 PM
Other engines have started implementing similar shadertypes, the bsdf brdf btdf bssrdf family of shaders for a more realistic approach to shaders in their respective engines. Although the core of the renderengines still remain untouched they're more of a halfassed attempt at reaching similar shaderquality as you get with Maxwell. Yes in theory it's often based around the same or similar algorithms for the b*df group of shaders in other engines but in practice they're far from close to the real thing you get inside Maxwell. It even differs alot between other 'unbiased' engines which are much more tightly built around the same group of b*df shaders.

/ Magnus

DanielWray
05-12-2009, 12:49 PM
I don't know if this is down to the shaders, but alot of renders from maxwell and other unbiased render engines is a sense of atomsphere in the image, i don't see that very often in other renders from other engines.

It's sort of hard to explain, i'm not sure if that's down to the shaders or the lighting models that they use.

mister3d
05-12-2009, 01:00 PM
@mister3d

Well, I believe maxwell, vray, mentalray e.t.c. shaders are actually based on same papers and algorithms. It's just that maxwell doesn't cut corners.

Again, my 2 cents. Perhaps someone more experienced in shader magic can shed some light on this topic =)

peace :)

AFAIK, neither Vray or mental ray engines perform calculations in spectral range.

BigPixolin
05-12-2009, 01:17 PM
The people who I know personally who prefer maxwell cannot use another renderer like MR,vray, or Brazil to save their lives.
They are so used to the renderer doing everything for them they have absolutly no concept of how lights work in the real world and how that can transfer among all renderers.
If it isn't a preset light emitter forget it.

DanielWray
05-12-2009, 01:23 PM
Well that may be for the people that you know, but that holds no ground for other people.

I'm using luxrenderer atm, which is, i guess similair to maxwell, yet i understand how light works, and how it should be portayed in certain circumstances and how it interacts with items etc etc...

Bubbaloo
05-12-2009, 01:37 PM
The people who I know personally who prefer maxwell cannot use another renderer like MR,vray, or Brazil to save their lives.
They are so used to the renderer doing everything for them they have absolutly no concept of how lights work in the real world and how that can transfer among all renderers.
If it isn't a preset light emitter forget it.

That's funny because most of the people I know who use Maxwell used mental ray or vray before, and after they found Maxwell, they didn't use them anymore... :)

BigPixolin
05-12-2009, 01:51 PM
That's funny because most of the people I know who use Maxwell used mental ray or vray before, and after they found Maxwell, they didn't use them anymore... :)

Different strokes for different folks.
Just curious if they do more stills than animation?

Bubbaloo
05-12-2009, 02:01 PM
Stills = Maxwell
Animation = mental ray

...at least for now...

Venkman
05-12-2009, 02:14 PM
I don't know if this is down to the shaders, but alot of renders from maxwell and other unbiased render engines is a sense of atomsphere in the image, i don't see that very often in other renders from other engines.

It's sort of hard to explain, i'm not sure if that's down to the shaders or the lighting models that they use.

I've noticed it, too. They have a little something more to them, don't they? I think a faster Maxwell Render is a good thing. Bring those renders down to earth, and then buy a fast dual quad nehalem machine... ooooooo.... sound nice.

Jozvex
05-12-2009, 02:23 PM
The people who I know personally who prefer maxwell cannot use another renderer like MR,vray, or Brazil to save their lives.
They are so used to the renderer doing everything for them they have absolutly no concept of how lights work in the real world and how that can transfer among all renderers.
If it isn't a preset light emitter forget it.

How awful that must be, concentrating on producing the art instead of whether your dmap shadows are too low resolution or not.

I'm not sure I understand the real life part though, everything about Maxwell's lights is based on real life! In mental ray you put in a value of 5 for a light's intensity, 5 nothing-in-particulars that come from a point in space or basic area light shape. In Maxwell you model a basic object to be the light source and choose to use watts/lumens/candelas/etc just as you would have in reality.

Of course I'm simplifying what is available in both renderers but you get the idea. Having automatic depth of field, motion blur, GI, etc that just plain works based on what you'd see from a real camera is just sooo much more predictable and easy. Yes, easy, and easy is good. By the way, you can render passes in Maxwell and passes in another renderer if you like, to get more flexibility. Real life photographers seem to have good flexibility too though to an extent.

AJ
05-12-2009, 02:38 PM
Having automatic depth of field, motion blur, GI, etc that just plain works based on what you'd see from a real camera is just sooo much more predictable and easy.It's funny you should sat that, I was showing a photographer how I set up shots in V-ray & when I showed him I could alter the depth of field without altering the 'aperture' he nearly shit himself with excitement.

playmesumch00ns
05-12-2009, 02:38 PM
@mister3d

Well, I believe maxwell, vray, mentalray e.t.c. shaders are actually based on same papers and algorithms. It's just that maxwell doesn't cut corners.

Again, my 2 cents. Perhaps someone more experienced in shader magic can shed some light on this topic =)

peace :)

It's all much of a muchness. The only difference is whether they're designed from the ground up to do biased or unbiased rendering. Maxwell, fry, indigo etc all fall into the latter camp. You can do unbiased rendering in mental ray and vray using their path tracing shaders, but they don't support spectral rendering out of the box as far as I know

playmesumch00ns
05-12-2009, 02:41 PM
Other engines have started implementing similar shadertypes, the bsdf brdf btdf bssrdf family of shaders for a more realistic approach to shaders in their respective engines. Although the core of the renderengines still remain untouched they're more of a halfassed attempt at reaching similar shaderquality as you get with Maxwell. Yes in theory it's often based around the same or similar algorithms for the b*df group of shaders in other engines but in practice they're far from close to the real thing you get inside Maxwell. It even differs alot between other 'unbiased' engines which are much more tightly built around the same group of b*df shaders.

/ Magnus

That's really not true at all. Maxwell's shaders really aren't based on reality any more than mental ray's are. That's why the best renders you see coming out of Maxwell are the ones that use very simple materials - glass, metals etc with measured reflectance data. Making realistic-looking organic materials still takes artistic talent.

What's different is in the way Maxwell calculates light transport, i.e. calculating the way that light bounces around the scene. The 'unbiased' term refers to the fact that the algorithms will, given an infinite amount of time, converge to the 'correct' result, whereas algorithms like photon mapping or irradiance caching (final gathering) will never generate a mathematically correct result.

Of course, 'correct' assumes you have set up your shader parameters correctly, which as I've already mentioned is still prone to human error.

BigPixolin
05-12-2009, 02:48 PM
How awful that must be, concentrating on producing the art instead of whether your dmap shadows are too low resolution or not. .

Not things like that.
Things that are in the real world that you HAVE to know in order to create it in MR.
Things you have to know in order to be a professional photographer.
It is just happening in maxwell and they don't know why. Exposure settings, light temputures, size of area lights, whitepoints etc......

This is not true for everyone just the people who I speak to personally.
We have had this conversation amongst ourselves when I give them jobs that require animation.
They fully understand and comment themselves how Maxwell has blinded them to seeing the underlying mechanics involved with lights in general wheather they are CG or real life.


I'm not sure I understand the real life part though, everything about Maxwell's lights is based on real life! In mental ray you put in a value of 5 for a light's intensity, 5 nothing-in-particulars that come from a point in space or basic area light shape. In Maxwell you model a basic object to be the light source and choose to use watts/lumens/candelas/etc just as you would have in reality.
See above

Of course I'm simplifying what is available in both renderers but you get the idea. Having automatic depth of field, motion blur, GI, etc that just plain works based on what you'd see from a real camera is just sooo much more predictable and easy. Yes, easy, and easy is good. By the way, you can render passes in Maxwell and passes in another renderer if you like, to get more flexibility. Real life photographers seem to have good flexibility too though to an extent.

Yes easy is good. But to master the art and have full contol over it in any situation one must know what is happening under the hood.
A real life camera automaticlly picks up motion blur and depth of feild. A real photograpehr knows why and can control it at will to his advantage. A ametuar photographer has no clue what its going on and it just happens. Same thing here.
Again I do not intend to generlize people. Just the ones I talk to personally.

Jozvex
05-12-2009, 02:48 PM
It's funny you should sat that, I was showing a photographer how I set up shots in V-ray & when I showed him I could alter the depth of field without altering the 'aperture' he nearly shit himself with excitement.

I bet! :p

Haha, of course there's a time and place for everything. You can actually do a depth pass in Maxwell (an unbiased one even, you should see it, good for measuring the distance to Mars etc) to add DoF in post later instead.

It's late for me and I can see my Sarcasm warning light is already flashing so I'd best go.

No hard feelings everyone, Maxwell is just better than all of you.

AJ
05-12-2009, 02:51 PM
Maxwell is just better than all of you.
Excellent!

BigPixolin
05-12-2009, 02:54 PM
Stills = Maxwell
Animation = mental ray

...at least for now...

Theres the difference we do mostly if not 100% animation.
I too look forward to the day hardware/software enables maxwell to be used more in animation.

Bubbaloo
05-12-2009, 03:05 PM
Theres the difference we do mostly if not 100% animation.
I too look forward to the day hardware/software enables maxwell to be used more in animation.

I hear ya. The day is getting closer!

CHRiTTeR
05-12-2009, 04:31 PM
We ran a test here with a simple BALL IN A BOX, lit with one area light. it took me 2 hours to replicate the look of a maxwell render. There were 3 guys testing in, one in vray (me), the other in mental ray and the third in maxwell. So this showed obviously that maxwell produces far more realistic results. I think it's about their shader algorithms. It's not that simple as you think.

says maybe more about who's doing the test :)

BTW, maxwell does cut corners too and ppl who keep calling pathtracing-only renderengines unbiased and think that other big renderers like vray, mental ray, etc cant render unbiased dont know what they are talking about and its funny to see how they try to sound like they know everything about it.

Maxwell isnt much easier to use compared to other renderers either. Maybe when it first came out, it was a bit easier (except if you want to create your own shader).
Also, you need to actually model the lightsource excactly to have a correct/realistic light simulation. In vray, mental ray, brazil, final render,... you just load an ies file and voila, done.
Physical cameras are included in every major rendering software these days (with much more control).
Mental ray and vray can also do path tracing. Biased or unbiased.

The main reason why different high quality renderengines give different results is not the shading (if you know what you're doing), its the tonemapping and has nothing to do with the rendering tech because tonemapping is a post process.
And thats also why they have a different 'atmosphere'.
But thats also the case with real cameras, they all use different tonemapping.
And that can be easily adjusted with photoshop anyway.

AFAIK, neither Vray or mental ray engines perform calculations in spectral range.

so?

That's really not true at all. Maxwell's shaders really aren't based on reality any more than mental ray's are. That's why the best renders you see coming out of Maxwell are the ones that use very simple materials - glass, metals etc with measured reflectance data. Making realistic-looking organic materials still takes artistic talent.

What's different is in the way Maxwell calculates light transport, i.e. calculating the way that light bounces around the scene. The 'unbiased' term refers to the fact that the algorithms will, given an infinite amount of time, converge to the 'correct' result, whereas algorithms like photon mapping or irradiance caching (final gathering) will never generate a mathematically correct result.

Of course, 'correct' assumes you have set up your shader parameters correctly, which as I've already mentioned is still prone to human error.

Exaclty, but its perfectly possibly to use 'biased' techniques to produce an image that looks exactly the same as an 'unbiased' version to the human eye (and beyond).
Especially in that flat shaded example :p

mister3d
05-12-2009, 06:14 PM
CHRiTTeR

I knew you would say this. If you want, here's the two images from this test, the maxwell version and the vray version which I matched without postwork. As you can see the scene is pretty simple, so I guess it won't be difficult for you to replicate. Try for yourself and see how default vray shaders will work. Also if you can, save the scene file so I could look at the settings. I'm really interested, if your initial vray test (without crazy tweaks) will produce something similar.
And yet those differ a bit, as i didn't go too crazy, and no post.

http://img2.imageshack.us/img2/1773/maxwellfz3.th.jpg (http://img2.imageshack.us/my.php?image=maxwellfz3.jpg)

http://img2.imageshack.us/img2/8110/ballanimationvraynewtg6.th.jpg (http://img2.imageshack.us/my.php?image=ballanimationvraynewtg6.jpg)

I bet you will see that vray by default isn't working the same way as maxwell. If you want though, try to create the scene, render it and use tonemapping in photoshop to create the same effect. But please show the original. :) And yet, I bet the vray, or mental ray results will look less realistic than maxwell, which didn't need any tweaks in photoshop or whatever.
The qyestion is how will you achieve the correct tonemapping for such a simple scene without references? I don't see your point.

I'm not sure if you refer to me about "unbiased renderers", but I didn't say anything about it and it doesn't inerest me.
Do you really believe that maxwell produces the same results initially as other renderers in terms of realism?


so?

I presume that' s what makes maxwell renderers different, not it's GI.

DanielWray
05-12-2009, 06:33 PM
The above to images too me, differ in one major way and that is the uniformness (is that a word?) of the diffuse lighing, the maxwell render seems to have inconsistant lighing across the whole area, by inconsistant i mean that the light levels vary quite alot over the surfce area, this is very clear (to me) in the small thumbnail view, you can see the variation on the grey surface, the other render seems a little more balanced across the surface, with very little too none variation in lighing, that is just how i see the image though, i could be wrong.

However, looking at both of the images, they clearly have there postives and negatives.

John Keates
05-12-2009, 06:34 PM
Mental Ray CAN do spectral rendering by the way, but I'm not sure how it compares to Maxwells. However, if you activate chromatic abberation etc to fully take advantage of the spectral rendering in Maxwell, render times shoot up (from what I remember of the demo).

I like Next Limits determination to keep their renderer as real as possible but the render times are still a bit much.

mister3d
05-12-2009, 06:39 PM
The above to images too me, differ in one major way and that is the uniformness (is that a word?) of the diffuse lighing, the maxwell render seems to have inconsistant lighing across the whole area, by inconsistant i mean that the light levels vary quite alot over the surfce area, this is very clear (to me) in the small thumbnail view, you can see the variation on the grey surface, the other render seems a little more balanced across the surface, with very little too none variation in lighing, that is just how i see the image though, i could be wrong.

However, looking at both of the images, they clearly have there postives and negatives.

Well, it's NOT out of the box vray render, it's tweaked to match! :) And I mean, vray won't render the same as maxwell even in such a simple scene. And I believe it's about shader algorithms, but I may be wrong, so maybe Chritter will prove otherwise.
Yet, in this test I missed using the vraycamera and vray path tracing, so maybe Chritter will show better results.

CHRiTTeR
05-12-2009, 06:41 PM
id like to do some quick tries, but keep in mind that if i cant do it that doesnt mean its impossible :)
And the scene does look quite simple indeed.

I can already see the shaders arent identical and overal brightness is also different.
That shouldnt be difficult to correct.


Can u post both scenes? The maxwell one and the vray one?

mister3d
05-12-2009, 06:42 PM
Double post

mister3d
05-12-2009, 06:45 PM
id like to do some quick tries, but keep in mind that if i cant do it it doesnt mean its impossible :)


Can u post both scenes? The maxwell one and the vray one?

I will ask the guy if he still has them. I just don't have it anymore.
I'm in no way try to confront you in a competition manner, I'm really interested if maxwell is all about tonemapping.
Bear in mind, I used different tweaks like layered shaders, changing GI saturation and contrast etc to try to reach what maxwell did. But right from thr start, vray version looked way different than what it's now. That's the point.

That what bugs me in those comparisons between the version 1.7 and 2: the 2 version looks like "vray", whereas 1.7 looks like something which is almost not posible in vray (I just didn't see such renders from vray without compositing).

lo
05-12-2009, 06:55 PM
your vray scene looks like it's missing reflective GI caustics.

mister3d
05-12-2009, 07:01 PM
your vray scene looks like it's missing reflective GI caustics.

Yes, it's not perfect. Yet the shadows are blurred and Gi is a photon map, not path-tracing.

CHRiTTeR
05-12-2009, 07:11 PM
I bet you will see that vray by default isn't working the same way as maxwell. If you want though, try to create the scene, render it and use tonemapping in photoshop to create the same effect. But please show the original. :) And yet, I bet the vray, or mental ray results will look less realistic than maxwell, which didn't need any tweaks in photoshop or whatever.
The qyestion is how will you achieve the correct tonemapping for such a simple scene without references? I don't see your point.

I'm not sure if you refer to me about "unbiased renderers", but I didn't say anything about it and it doesn't inerest me.
Do you really believe that maxwell produces the same results initially as other renderers in terms of realism?


Off course they dont use the same default configuration. Vray even doesnt have gi on by default. :)

I think its possible without even touching photoshop.

It doesnt mean if one looks more pleasing or natural that it actually is the most real version.
It is perfectly possible that the more fake looking version is actually the most correct/realistic version.

This test doesnt show which result is more realistic. Its just a good test to show if you can get the same result with another renderer.

mister3d
05-12-2009, 07:15 PM
Off course they dont use the same default configuration. Vray even doesnt have gi on by default. :)

I think its possible without even touching photoshop.

It doesnt mean if one looks more pleasing or natural that it actually is the most real version.
It is perfectly possible that the more fake looking version is actually the most correct/realistic version.

This test doesnt show which result is more realistic. Its just a good test to show if you can get the same result with another renderer.

:) Ok, ok, let's hope the scene is still there that you try to work from the scrath with vray and see for yourself what I mean. But why do you deny the fact that maxwell does more realistic renders than most raytracers out of-the-box?
I think that what looks more realistic is more realistic, isn't it? Especially if we talk about results with simple setups like one area light and several surfaces?

DanielWray
05-12-2009, 07:20 PM
Instead of testing with a scene that you have no real image to compare with, why not find an image that is relativly simple, perhaps a vase or jar and render a scene in maxwell, vray, mentral ray and i'll use luxrender and compare the real image?

pluMmet
05-12-2009, 07:48 PM
Everybody seems to be arguing and i don't want to join in really but I just wanted to say that Unbiased renders like Fry and Maxwell's images look far superior to me.

That's my bottom line. For someone to say that a studios won't use it because it takes too long makes me laugh... Studios don't make photo real movies with any renderer... It Takes Too Long :lightbulb

CHRiTTeR
05-12-2009, 07:49 PM
:) Ok, ok, let's hope the scene is still there that you try to work from the scrath with vray and see for yourself what I mean. But why do you deny the fact that maxwell does more realistic renders than most raytracers out of-the-box?
I think that what looks more realistic is more realistic, isn't it? Especially if we talk about results with simple setups like one area light and several surfaces?

I just dont agree.
Vray also renders realistic out of the box, everything you need is there... you only need to turn some stuff on, which u also need to do in maxwell (skylights, lights, camera, ....) so its not different. Its not like maxwell makes all the shaders and materials for you and places+configures all the lights and cameras automaticly. You still need to do all that manually. The only difference is that it has gi turned on by default. Thats not a huge difference.
And you can always save/make vray render presets and just load/reuse those if you want.

I dont agree on what 'looks' realistic 'is' realistic. Its pretty easy to render a realistic 'looking' image that has uncorrect shadows etc... But it doesnt really matter in this test and lets not start arguing about that.
The focus of this test isnt realism perse, its more about making a vray render look as close as possible or identicall to the maxwell one.

filestyleCGI
05-12-2009, 08:15 PM
Everybody seems to be arguing and i don't want to join in really but I just wanted to say that Unbiased renders like Fry and Maxwell's images look far superior to me.

That's my bottom line. For someone to say that a studios won't use it because it takes too long makes me laugh... Studios don't make photo real movies with any renderer... It Takes Too Long :lightbulb

Cosign

And to mr alexyork (http://forums.cgsociety.org/member.php?u=40641), regarding exactly that point.
I work in the cgi for print field and use maxwell only for photorealistic renders.
We work for a couple "big players" in the ad industry, where there are always corrections and client wishes.

However its all about not sticking your head in the ground and going stupid saying stuff like:

Oh well, all my other render engines can render the same picture in 3 minutes instead of 50...good for you, still maxwell renders look way better comparing a to b. Keep in mind that great artists can do amazing work with crap. were talking technicalities here.

So you ask, how does this work on a day-to-day basis and how can you work with clients.

Easy. We do low res monitor screen renderings with MultiLight (talk about changing a light...) and meet with our client to see what is possible (or send a jpg) and once they are happy you can go render fullres for print and do all the teeny weeny stuff in post...

I'm not for flaming or anything, I just think its great when users with high postcount give prejudiced, and stuck in the old days attitude the forehand.

Oh, and if youre interested, my clients tell me they have never ever seen their products represented so great in cgi.


I do think its nice to trick a little as you have to do with biased GI renderers like Mental Ray, Vray etc. I also see amazing images done with these engines. Fine if you want to light your interior by re-rendering 100 times moving stuff back and forth an inch untill you get a real nice look. Okay, What I do is setup 2 or perhaps 3 different light setups, render multilight and probably my 4th rendering with be my final.

Because it looks good form the start.

Sorry for being a little pissed, but I can't stand the ignorance.
I will upgrade to 2.0 and I'm sure I will atleast close to very happy.

Not to mention the community is great, Heya guys I see some of you here (Its Joris ;))

cant wait for 2.0. haters go home and try rendering some dispersion ;)

Mihai75
05-12-2009, 10:20 PM
Why do these discussions always appear, like the 'biased' guys somehow feel threatened and have to explain to the unbiased guys why it's a bad thing to use an unbiased render, they can do the same with X renderer much faster. Do you feel the deep CG knowledge you have accumulated during many sleepless nights will somehow be invalidated when a newbie photographer say, can create a realistic CG image, without knowing what a photon map is, or an AA filter or any of the other 30+ settings you have to know and understand in a biased renderer? Settings that have absolutely no connection to how light 'works' in the real world, that are purely there from a CG necessity because of how the renderer itself works.

When you count the time it took you to finish an image, do you only count the render time? Or do you also count the time it took you to tweak those 30+ settings that vary from scene to scene, to avoid any artifacts, flickering, have perfect AA and so on. What is more valuable, your time or a machines time? Even if you are an experienced user, you still have to spend time to tweak those settings for every scene that you work on, instead of working on something else. You want to say for a scene where you need accurate GI, indirect reflected caustics (and accurate ones, not just a blurry mess), nice looking glossy reflections, you just adjust the settings for that scene in 5 minutes, hit the render button and you get a render at 3000x3000 in 1 hour?

Biased/Unbiased, it's just another tool, that should be appreciated for what they allow you to do and both allow you to do some pretty amazing things. Personally, if I spend a lot of time to create a nice model, spend time again to create good textures, I don't mind longer render times if I know each time that the work I put into the scene will be represented in a very realistic way. It doesn't make me a better artist because I know the intricacies of how 4 types of Final Gathering settings work. But, I can appreciate that this approach serves a purpose as well, and in the end I can appreciate the work of all the people that actually develop these tools without which we would have to go outside more and face the shockingly cruel reality:scream:

Zendar
05-12-2009, 11:50 PM
By the time I've read thru this topic my Maxwell work is already at 90%...

What I mean is, some of you have no trouble adding countless posts that pretty much say 'my dad is better than yours' arguing over something that is a simple tool that you MAY or not choose. You are wasting hours of time and yet you complain that some of us don't mind pressing render button on Maxwell and go do different things while it's working.


I've read this topic and at some points it's almost a reflection of my 5 year old with his mates having a chat in a sandbox.

That being said. I'm going to swim for a bit as thanx to Maxwell (and I love it) I have time to stretch my sore muscles. :scream: :thumbsup:

PS. I'm using Maxwell on my mates system only for four days and honestly it' as simple as people say. Just press render and go do different things. I really like that. Not because I'm lazy or incapable (even tho I'm very fresh to 3D) but because I prefer to do art instead of converting myself into a light engineer.

Jozvex
05-13-2009, 12:29 AM
Here's an animated gif comparison of a scene using only direct illumination in both Maxwell and mental ray:

http://www.jozvex.com/images/lambertMentalMaxwell.gif

Both have an equivalently sized, flat area lightsource and I just raised the exposure on the Maxwell image to get it to match close enough to the mental ray one. They are similar but not the same. I'm not sure what that means exactly, I just thought I'd show it. It could be a shader difference or the light may emit differently from Maxwell's plane geometry light, not sure. Also, this is using the actual 'lambert mode' function the Maxwell material has, if you turn that off it looks quite a bit more different still.

BTW, maxwell does cut corners too and ppl who keep calling pathtracing-only renderengines unbiased and think that other big renderers like vray, mental ray, etc cant render unbiased dont know what they are talking about and its funny to see how they try to sound like they know everything about it.

Path tracing in mental ray (at least via the 3D apps) is completely hopeless! :surprised PPT in VRay is much better obviously, but even then you can't render a terribly big image before you start running out of paths to clear it up, or at least that's how it was when I last checked.

Maxwell isnt much easier to use compared to other renderers either. Maybe when it first came out, it was a bit easier (except if you want to create your own shader).

You don't think it's easier? Fair enough. :scream: I find using just the one material node over the countless mental ray ones to be way simpler. The mia/architectural material has like 5 times the buttons and settings of the Maxwell one. Not to mention all the rendering options and linear worklflow setups and light settings.

Of course this is all coming from a desire to be photorealistic, clearly having 500 settings is much better when you want to start being more stylistic etc.

Also, you need to actually model the lightsource excactly to have a correct/realistic light simulation. In vray, mental ray, brazil, final render,... you just load an ies file and voila, done.

IES lights are cool for sure. I wouldn't necessarily say they create a better simulation. Shadows cast by the lights aren't based on the IES profile as far as I know, neither are photons for GI.

Physical cameras are included in every major rendering software these days (with much more control).

I suppose, all the settings are just sort of arbitrary though, at least in mental ray for Maya. Without lights having the right sort of intensities and the scale of your scene actually affecting the render it's not that useful.

Haha I can't wait 'til they post the next feature update. You'd think speed increases would get people excited and positive.

dagon1978
05-13-2009, 01:13 AM
Here's an animated gif comparison of a scene using only direct illumination in both Maxwell and mental ray:

http://www.jozvex.com/images/lambertMentalMaxwell.gif

Both have an equivalently sized, flat area lightsource and I just raised the exposure on the Maxwell image to get it to match close enough to the mental ray one. They are similar but not the same. I'm not sure what that means exactly, I just thought I'd show it. It could be a shader difference or the light may emit differently from Maxwell's plane geometry light, not sure. Also, this is using the actual 'lambert mode' function the Maxwell material has, if you turn that off it looks quite a bit more different still.


uh uh terminator artifacts in maxwell? :p

vlad
05-13-2009, 01:20 AM
...
That being said. I'm going to swim for a bit as thanx to Maxwell (and I love it) I have time to stretch my sore muscles. :scream: :thumbsup:
...

Try that in a production environment with crazy tight deadlines :scream: :thumbsup: :surprised

Jozvex
05-13-2009, 02:14 AM
Matteo! Shhh I thought I paid you not to mention that!!

Yeah I think all unbiased renderers have terminator issues. Luckily texturing and more complex geometry covers it up usually.

BigPixolin
05-13-2009, 02:28 AM
Here's an animated gif comparison of a scene using only direct illumination in both Maxwell and mental ray:

http://www.jozvex.com/images/lambertMentalMaxwell.gif

Both have an equivalently sized, flat area lightsource and I just raised the exposure on the Maxwell image to get it to match close enough to the mental ray one. They are similar but not the same. I'm not sure what that means exactly, I just thought I'd show it. It could be a shader difference or the light may emit differently from Maxwell's plane geometry light, not sure. Also, this is using the actual 'lambert mode' function the Maxwell material has, if you turn that off it looks quite a bit more different still.


Looks like the maxwell image had vignette in it?

Jozvex
05-13-2009, 02:41 AM
You're right! :banghead:

Curses!!

Airflow
05-13-2009, 03:12 AM
Can somone point out the terminator issues cos I prefer the maxwell render and cant see any probs.

3dtutorial
05-13-2009, 03:14 AM
Can somone point out the terminator issues cos I prefer the maxwell render and cant see any probs.

Look at the stairstepping artifacts in the terminator.

Cheers,

J

Airflow
05-13-2009, 03:22 AM
Must be my monitor, but to me the mentalray render looks noisy and the maxwell one looks smooth. need my eyes checked. :deal:

ZacD
05-13-2009, 03:25 AM
Must be my monitor, but to me the mentalray render looks noisy and the maxwell one looks smooth. need my eyes checked. :deal:

The shadow in mR has more noise, but the render settings and times were not posted.

BigPixolin
05-13-2009, 03:45 AM
Must be my monitor, but to me the mentalray render looks noisy and the maxwell one looks smooth. need my eyes checked. :deal:

There is also depth of field blurring things in the Maxwell image.
Possibly a light glow/glare in the maxwell image aswell.

Probably best to strip all lense effects for any real comparision.

Jozvex
05-13-2009, 05:12 AM
I'll do a better one now! :p

There isn't any glare/glow but yes I did just leave the area shadows at their default sampling level for mental ray, I didn't think it would get quite this picky. The terminator issue isn't too bad here because the sphere is quite dense, but with lower polygon counts it can be ugly.

I'll edit this post in a few minutes with a new image. And no the Maxwell one didn't take hours to render haha.

mister3d
05-13-2009, 05:27 AM
Jozvex, this comparison doesn't show any benefits of maxwell. There must be comething more complex with metal, dielectrics, glass... at least something that would clearly show "it's maxwell" and then people would try to make it in mental or vray.

Jozvex
05-13-2009, 06:10 AM
Here we go for round 2:

http://www.jozvex.com/images/renderCompareMentalMaxwell.gif

mental ray: The area light has 48 samples with quadratic falloff. I used a gamma node to ungamma the colour of 0.5 grey. Photographic lens shader is on 2.2 gamma with no black crush and 0.8 highlight burn simply because those values are what Maxwell uses but of course it's probably not equivalent. Rendering samples 0, 2 with a contrast of 0.05. Mitchell filtering with a side of fries. Rendertime 25 seconds.

Maxwell 'lambert': Same as before (0.5 grey with lambert mode switched on) but this time I turned off the vignetting and upped the fStop to 16 to try and get rid of some DoF. Rendertime 3.5 minutes. Yes yes, clearly much longer than 25 seconds but with complex scenes/lighting you start to see a better tradeoff.

Maxwell standard: Only differences are that I turned off the lambert switch on the shader so it now uses the full Maxwell shading model with a roughness of 99 (being the roughest). Rendertime was 4 minutes to reach the same sample level.

Mister3D, yeah this certainly doesn't show any of Maxwell's strengths but this test was stemmed from people discussing if the shading model was different or not. The 'Maxwell standard' image I rendered is using the shading model that gives you all the different shading possibilities for glass/metal/cloth/etc apart from SSS and thin film coatings which you add separately.

sandykoufax
05-13-2009, 06:21 AM
Interesting. So many replies with just some images in there. I didn't expected this situation, haha.

Samo
05-13-2009, 06:26 AM
is there really a need for interpolation in rendering?
That's probably what Maxwell is doing behind the scenes, some dirty interpolation tricks to reduce noise.

Jozvex
05-13-2009, 06:58 AM
sandykoufax! Hi, :)

Yeah it's all a bit of a non-event really haha. Maxwell 2 will be much faster, that's what we should take from this thread. It certainly makes me happy! :bounce:

I was just about to type out the ultimate Maxwell equation that unlocks the key to awesome, but Next Limit might send an unbiased ninja after me so I'd better not. :shrug:

Magnus3D
05-13-2009, 07:37 AM
Now when there's a Vray for Maya thread i wonder if we Maxwell users should slaughter that thread like you guys wrecked this thread so nicely, i mean.. it would only be fair. :)

/ Magnus

CHRiTTeR
05-13-2009, 07:49 AM
Here's an animated gif comparison of a scene using only direct illumination in both Maxwell and mental ray:

http://www.jozvex.com/images/lambertMentalMaxwell.gif

Both have an equivalently sized, flat area lightsource and I just raised the exposure on the Maxwell image to get it to match close enough to the mental ray one. They are similar but not the same. I'm not sure what that means exactly, I just thought I'd show it. It could be a shader difference or the light may emit differently from Maxwell's plane geometry light, not sure. Also, this is using the actual 'lambert mode' function the Maxwell material has, if you turn that off it looks quite a bit more different still.

Remove maxwell's depth of field and vignetting (or add it in mental ray) and they look the same.


Path tracing in mental ray (at least via the 3D apps) is completely hopeless! :surprised PPT in VRay is much better obviously, but even then you can't render a terribly big image before you start running out of paths to clear it up, or at least that's how it was when I last checked.

True, (dont know much about mental ray though),
Path tracing in vray was more developed to be used to have good secondary GI and not so much as a pure pathtracing solution (although you can use it for that).
If you use the fully adaptive brute force thingy (aka the universal settings in vray) you get the same results as you would wtih path tracing, but much faster (because its fully adaptive).
Also the next version of vray will have bi-directional path tracing. Which looks verry promessing and fast.


You don't think it's easier? Fair enough. :scream: I find using just the one material node over the countless mental ray ones to be way simpler. The mia/architectural material has like 5 times the buttons and settings of the Maxwell one. Not to mention all the rendering options and linear worklflow setups and light settings.

Well its not easier or more difficult, its just more confusing because you got more options available, which most you can leave at default so that requires no extra work.

Of course this is all coming from a desire to be photorealistic, clearly having 500 settings is much better when you want to start being more stylistic etc.

Again, a 'reaslistic' render is quite relative. Because there are tons of realworld cameras who give verry different results in color and brightness/contrast also.

If 'realistic' means physicall correct behaviour of light, then both renderers do this.
Although maxwell seems to give different results with every new releasy, so for me its obvious that they are changing something in regards with the behaviour of light and shading which cant be correct because this simply doesnt happen in real life.
I also can see how this can give problems in production when you need to re-render an old image.

Aslo you now have VERRY cool enviroment fog in vray, so thats kind of more realistic than maxwell, because we have fog and dust everywhere in the real world. :)


IES lights are cool for sure. I wouldn't necessarily say they create a better simulation. Shadows cast by the lights aren't based on the IES profile as far as I know, neither are photons for GI.

If there is info about the light's shape in the IES file, it will be used by vray (dont know about mental ray) and you'll get correct area shadows and everything.


I suppose, all the settings are just sort of arbitrary though, at least in mental ray for Maya. Without lights having the right sort of intensities and the scale of your scene actually affecting the render it's not that useful.

Im a max user so i dont know about maya, but thats not really because the rendering tech cant do it, its more about the options not being available in your modeling/animation package's UI. Has nothing to do with mental ray.

Haha I can't wait 'til they post the next feature update. You'd think speed increases would get people excited and positive.

Well, they would be intresting if they did it without resulting in different looking images. ;)

khendar
05-13-2009, 07:49 AM
This is starting to descend into "X software vs Y software" territory. Play nice please.

mister3d
05-13-2009, 08:01 AM
This is starting to descend into "X software vs Y software" territory. Play nice please.

Not really . The point was to prove that maxwell shader algorithms differ from other renderers, not if it's better or not, but whether it's different.
But yeah, maybe uit's time to move into shader and lighting section. :blush: Let's create a topic there, guys?

CHRiTTeR
05-13-2009, 08:11 AM
oop, didnt notice round2, sorry :blush:

Looks like maxwell one has some extra bouncelight coming from the spere on the groundplane.

Are you sure GI is on in mental ray with fair quality settings?


Can u post the scene files so other ppl can have a go at it?

That's probably what Maxwell is doing behind the scenes, some dirty interpolation tricks to reduce noise.

Yes, i noticed that too. Shadows arent that crisp and the GI light looks more 'smeared/spread'.
Looks like they are using larger samples, which results in a biased solution by the way.


Although i do fully realise that its not abnormal when an image rendered by using pathtracing techniques does change brightness in certain areas quite a lot over time. Even if you compare a mid-noisy image with the final noise-free rendering. But to be honnest, they should've not used such confusing examples if that is the case here.
They could've perfectly showed camparable noise free final images and just mention the rendertimes. Then it would show there is no difference in imagequality, only in speed. Thats the whole point of those examples in the first place, right? So why didnt they do that? :)
That in combination with the 'interpolated' look is why i have strong doubts.
Cant be sure of course until the final product is out and everyone can test it for themselves :)
This i just the impression i get by looking at their examples, and discussing that is what these forums are for right?

mister3d
05-13-2009, 08:38 AM
They could've perfectly showed camparable noise free final images and just mention the rendertimes. Then it would show there is no difference in imagequality, only in speed. Thats the whole point of those examples in the first place, right? So why didnt they do that? :)

They were tired to wait.

stallion151
05-13-2009, 09:01 AM
unbiased just means it doesn't round up the maths to the nearest decimal or cap the light bounces, people think it means god is literally drawing the image for you and then cry foul if an improvement seems to have been made...i think its safe to say that Maxwell isn't perfect, its trying to be but improvements are inevitable and welcome.
I've only just started playing around with Maxwell and find it very liberating, I have a powerful machine and results are very quick. I can still use Maya while its rendering in a separate window.

Stop you effn naggin. If you want traditional lighting go back to lighting with single directionals and spots, because its just a tool.
Don't see this immature bickering in the building industry, this wheelbarrow is better than this one. Like BigPixolin said "Different strokes for different folks"
Use it, don't use it, nobody cares or loves you.

Magnus3D
05-13-2009, 09:10 AM
Sure Maxwell is not good for everything, even i know that and i agree. But is Vray and mental ray perfect for everthing ? no they're not. But people here don't wanna hear that their renderengine sucks in one way or another. They just wanna trash other engines which they don't use themselves and ignore other people's opinions.

/ Magnus

CHRiTTeR
05-13-2009, 09:25 AM
Its not about arguing which one is 'best', because that is quite relative.

Its about the fact that many ppl seem to think maxwell gives more realistic results then you could get with other renderers. Or that it is easy to use.
Which is not really true. It was easier to use when it just came out though, thats true, but now most other renderers have physicall cameras and reinhard tonemapping too and 1 click gi with quality presets. Not so difficult at all...

If it doesnt bother you, good for you, but why is it then that when someone says something 'negative' about it you experience it as 'naggin'?

Forums are there to share experience and opinions and hopefully learn something out of it.
You already know everything and think theres no point in discussing things, good for you. What are you doing here?

But fact is that someone must be right, right?
And its about trying to try to work things out, share thoughts and hopefully find out which arguments are true and which not.

I do realise i probably sound like an ass a lot of times, but im just not affraid to say my thoughts and if someone can correct me with some solid arguments then thats perfectly fine. Im always eager to learn!

Sure Maxwell is not good for everything, even i know that and i agree. But is Vray and mental ray perfect for everthing ? no they're not. But people here don't wanna hear that their renderengine sucks in one way or another. They just wanna trash other engines which they don't use themselves and ignore other people's opinions.

/ Magnus

I agree. For example in a spectral renderer it is easier to implement things like dispersive effects, but its not impossible to do in another 'standard' renderer and thats more something for the developpers to worry about and not the users.

And i cant think of something that cant be dont in another 'standard' renderer. Except for correct thin film coating (for now) which is not THAT of an important effect that has to be 100% accurate (there are exceptions yes), but its still a valid point yes. :)
I dont know if this could be done in a 'standard' raytracer or if its just a matter of implementation.
*edit: actually it is possible but as far as i know its not implemented as an easy parameter 'YET' (and you have to do it the hard/manual way).

If you know any more things, pls do share :)
(As long its not that 'photorealistic' marketting crap)

pluMmet
05-13-2009, 09:58 AM
Its not about arguing which one is 'best', because that is quite relative.

Its about the fact that many ppl seem to think maxwell gives more realistic results then you could get with other renderers. Or that it is easy to use.
Which is not really true.

I've always thought that Vray renders look "dead" to me while Fry and Maxwell looked real...

I thought maybe I hadn't given Vray a chance in a while so I went to Chaos Groups web site to look at some renders and they don't have a gallery. So instead of seeing the best that Vray has to offer I would be reduced to pouring over the Chaos Group forums to try to find the needle in the hay stack.

Looking "good" and looking "real" are two different things. I've hardly seen any Vray render that looked real (to me of course.)

As for render times I've always heard that once a scene gets big and complicated the unbiased is faster then any of the others with all of their features turned on.

As far as the unbiased being easier to use... I guess it's just how some people are wired? Maxwell was my first unbiased and it made as much sense to me as taking a picture IRL.

Skin is a harder thing for unbiased ATM (more so for Maxwell then Fry) while MR has a great shader for it. Non the less MR Fast skin takes many tweaks just in that one shader. Sure you get better at it over time but that is one thing you can give up when you learn unbiased.

So there is the strong opinion I actually thin that unbiased look real and the rest as a rule don't (without serious tweaking and in certin situations where lighiting is set up just for that one shot.)

CHRiTTeR
05-13-2009, 10:12 AM
Yes, but that is like me going to the maxwell forum in search of quality and seeing unrealistic lighting because they gave a lightbulb the wrong colortemperature for example. Or they show a render of a 'golden' statue but the shading looks more like copper... It still depends on the users.

High and photolike quality is fairly easy to achieve these days with 'standard' raytracers.

I agree there are many/some disapointing pics in the vray forums. But i think theres also more then enough verry realistic examples.

I dont know if you have any experience of rendering in a production enviroment where everything needs to go fast then its normal you give up some quality for speed, because the possibility is there and have full control there.
That doesnt mean the same scene cant be rendered much more realistic and detailed. Its just that ppl configure the time/quality balance to give results that are 'acceptable/good enough' for the client.
Most ppl dont render caustics, that doesnt mean its impossible. Its just because it isnt 'that' important for the client he doesnt want to invest money in the time needed to calculate caustics 'non-rendering' ppl will barely notice.

mister3d
05-13-2009, 10:17 AM
This talk won't lead anywhere until people will run some tests with maxwell and other's shaders and lighting. The best way is to setup the scene in let's say mental ray, and using the same parameters test it in vray and maxwell. And then compare the results. Then you will see if maxwell does more realistic results or not. But I guess maxwell versions could have gone through some optimizations so maybe the version 1.0 would reveal the most distinct difference.

rivoli
05-13-2009, 10:18 AM
I thought maybe I hadn't given Vray a chance in a while so I went to Chaos Groups web site to look at some renders and they don't have a gallery.


actually they do have a gallery:

http://www.chaosgroup.com/en/2/galleries.html

Bercon
05-13-2009, 10:26 AM
pluMmet, pretty off topic but here is the VRay gallery: http://www.chaosgroup.com/en/2/galleries.html

ronaldomiranda
05-13-2009, 11:32 AM
This discussion, what is more realistic, is very subjective; since everyone see things very diferently. Just look at the STILL images sections of this forum, everyone sees differents angles of 1 same image posted, day by day...

(I meant the visual perception of realism, and not the technical stuff discussed before ;) )

Besides that... i doubt anyone in here is capable of see an image and tell in which render was done! take the best pictures of all galleries: Modo, Vray, Mental Ray, Fry, Maxwell!

its very subjective!

(my 2 cents)

MasonDoran
05-13-2009, 11:33 AM
hey guys, if you want make speculations....at least go to the MR developer forums instead of CGTalk to get your first hand information:

:::::::::::::::::::::::::::

http://www.maxwellrender.com/forum/viewtopic.php?f=3&t=31365&start=60


:::::::::::::::::::::::

In short, the differances you see in the rendering are because 2.0 claims to be more accurate (bugfixes from 1.0). It is still absolutely unbiased without any shortcuts or cutting corners. So the 1.x version was less accurate ? Actual quality of the renders are moot in this point, as it is only a benchmark demonstration of speed and noise.

thev
05-13-2009, 12:38 PM
In short, the differances you see in the rendering are because 2.0 claims to be more accurate (bugfixes from 1.0). It is still absolutely unbiased without any shortcuts or cutting corners. So the 1.x version was less accurate ? Actual quality of the renders are moot in this point, as it is only a benchmark demonstration of speed and noise.So it was unbiased and ultra-accurate before, and it is unbiased and ultra-accurate now, but the results are obviously different. Then apparently one of the results is wrong. Can you tell which is really the "correct" one? If it is the new version, then all images done with the old one are "wrong"...

Bubbaloo
05-13-2009, 12:46 PM
Ok, it's time to "unsubscribe" this topic...

Magnus3D
05-13-2009, 12:51 PM
That's right Bubba, haven't you learnt anything from this tread.. the conclusion is now that Maxwell sucks! and Vray and mental ray and Fryrender rules. So now we all Maxwell nerds better crawl back under our stones and hide again.

/ Magnus

mister3d
05-13-2009, 12:56 PM
It's just nobody has the balls to test as I did.
http://img410.imageshack.us/img410/8110/ballanimationvraynewtg6.th.jpg (http://img410.imageshack.us/my.php?image=ballanimationvraynewtg6.jpg)

dmeyer
05-13-2009, 12:58 PM
It's just nobody has the balls to test as I did.
http://img410.imageshack.us/img410/8110/ballanimationvraynewtg6.th.jpg (http://img410.imageshack.us/my.php?image=ballanimationvraynewtg6.jpg)

Har Har. Pun intended? :)

DanielWray
05-13-2009, 01:10 PM
I think the ball scene may be a good way to test which renderer can produce the most noise/ error free render.

Personally i think a scene needs to be rendered that has a comparable photo, that way you can see which one is the most physically correct and which render can produce true to life images, in the quickest and easiest way possible. After all, those Vray renders were utterly amazing in the gallery, it is just as capable of Maxwell..

ThirdEye
05-13-2009, 01:12 PM
Can't believe we're fighting over lambert spheres with a plane and a light source :surprised

wombat778
05-13-2009, 02:15 PM
Personally i think a scene needs to be rendered that has a comparable photo, that way you can see which one is the most physically correct and which render can produce true to life images, in the quickest and easiest way possible. After all, those Vray renders were utterly amazing in the gallery, it is just as capable of Maxwell..

Here is an old post in the Maxwell forum that has a Cornell box reference photo, along with a Maxwell render for comparison. Maybe its worth trying to duplicate it in Vray...

http://www.maxwellrender.com/forum/viewtopic.php?t=14254

vlad
05-13-2009, 02:21 PM
I've always thought that Vray renders look "dead" to me while Fry and Maxwell looked real...

I thought maybe I hadn't given Vray a chance in a while so I went to Chaos Groups web site to look at some renders and they don't have a gallery. So instead of seeing the best that Vray has to offer I would be reduced to pouring over the Chaos Group forums to try to find the needle in the hay stack.

Looking "good" and looking "real" are two different things. I've hardly seen any Vray render that looked real (to me of course.)

...

Go to CGarchitect's gallery and have a look around. You'll find plenty of bad Maxwell renders and plenty of good Vray renders. And vice-versa. Maxwell is absolutely no assurance that you'll get good results. Neither is Vray or any other piece of software.
In any case, check out this Vray user : http://area.autodesk.com/index.php/stories_bts/bts_detail/publications_by_alex_roman/

CHRiTTeR
05-13-2009, 02:25 PM
In short, the differances you see in the rendering are because 2.0 claims to be more accurate (bugfixes from 1.0). It is still absolutely unbiased without any shortcuts or cutting corners. So the 1.x version was less accurate ? Actual quality of the renders are moot in this point, as it is only a benchmark demonstration of speed and noise.

Haha, sorry but thats just hilarious.


So it was unbiased and ultra-accurate before, and it is unbiased and ultra-accurate now, but the results are obviously different. Then apparently one of the results is wrong. Can you tell which is really the "correct" one? If it is the new version, then all images done with the old one are "wrong"...

Indeed, but the 'interpolated' look and speed improvements make me think its more like the other way around.

Still id like to see a better (noise free images) comparison between the versions

TAVO
05-13-2009, 02:26 PM
In any case, check out this Vray user : http://area.autodesk.com/index.php/stories_bts/bts_detail/publications_by_alex_roman/

That guy should got an award by his Post work, i bet those images are heavily retouched.

The ChaosGroup Gallery is very nice though.

pluMmet
05-13-2009, 02:34 PM
Thanks for the Vray gallery links. I'm on an iTouch now and for the next few hours but I will check them out when I get to a nice big color corrected monitor ;)

mister3d
05-13-2009, 02:39 PM
Here is an old post in the Maxwell forum that has a Cornell box reference photo, along with a Maxwell render for comparison. Maybe its worth trying to duplicate it in Vray...

http://www.maxwellrender.com/forum/viewtopic.php?t=14254

Great, I will prepare the scene for both mental and vray to test.

pluMmet
05-13-2009, 04:27 PM
Great, I will prepare the scene for both mental and vray to test.
A sphere a cylinder, two boxes, a light and 2 hours.

How's it coming?

JML
05-13-2009, 04:48 PM
I've always thought that Vray renders look "dead" to me while Fry and Maxwell looked real...
that's just hilarious.

A sphere a cylinder, two boxes, a light and 2 hours.
How's it coming?
grow up


This thread reminds me of childish mac-pc wars..

dagon1978
05-13-2009, 05:03 PM
Matteo! Shhh I thought I paid you not to mention that!!

Yeah I think all unbiased renderers have terminator issues. Luckily texturing and more complex geometry covers it up usually.
eheh i know i know Jared... just kidding :beer:
looking forward to see what's coming on, i'm ready to buy my next workstation, maybe this is the time to try maxwell again


So it was unbiased and ultra-accurate before, and it is unbiased and ultra-accurate now, but the results are obviously different. Then apparently one of the results is wrong. Can you tell which is really the "correct" one? If it is the new version, then all images done with the old one are "wrong"...
:scream: point taken vlado :thumbsup:

pluMmet
05-13-2009, 05:16 PM
that's just hilarious.


grow up


This thread reminds me of childish mac-pc wars..
People say that the set up time in unbiased is not faster....

Why is this not a valid question?

And why can't Vray renders look lifeless to me?

mister3d
05-13-2009, 05:30 PM
A sphere a cylinder, two boxes, a light and 2 hours.

How's it coming?

I was busy with another stuff. It will be ready very soon. I think within an hour.

JML
05-13-2009, 05:37 PM
And why can't Vray renders look lifeless to me?
That was hilarious because you were generalizing that all vray renders look lifeless.
If they ALL look lifeless to you then I would suggest changing carreers.

Some renders are good and some bad, just like in maxwell.
Maxwell brings perfect lighting but I saw some users trashing it with unrealistic textures,or
with the sun at weird positions for example.

I'm not a vray fan but the work it creates and speed it does it at is really good.
there is a reason why it's chosen by most right now.

I own maxwell but I cannot use it because of heavy render times. Same at work.
But maybe later it will get better.

pluMmet
05-13-2009, 05:50 PM
It was a generalazation and I'm glad that's how you took it.

I made it clear that I haven't made a habit of searching Vray renders.

I haven't been shopping for an external rendered for a while.

How all that could be funny is beyond me?

:( my iTouch will not let me spell rendered with r at the end

JML
05-13-2009, 06:03 PM
How all that could be funny is beyond me?
I was making fun of your first comment. if you still don't get it,
I don't have time to draw you a picture.

I made it clear that I haven't made a habit of searching Vray renders. but yet you say that All vray renders are lifeless.

I haven't been shopping for an external rendered for a while.
you should look around other renderers to know which one better fits your need,
instead of being a fanatic to only 1.

alexyork
05-13-2009, 06:05 PM
i wasn't going to reply to this thread after all the ridiculous insults flying around (some of which were clearly directed at me) but it seems it's extremely difficult to actually ask valid questions about a tool without the fanboy userbase jumping in and throwing insults back at you, when all you wanted was a proper answer...there are plenty of valid questions throughout this thread and a disproportionate amount of proper answers. my observations were based entirely on experience, not conjecture...

all i care about is how efficient my renderer is. i.e. how quickly i can get a result i'm happy with. i can get the same quality as a maxwell render out of scanline if i want, or vray or mentlray or MSPaint for that matter. all it comes down to is which tool allows you to get the best results in the most efficient manner. that might be quick setup times or quick rendertimes or both in many cases. again, all that matters is the end result and how efficiently you managed to get to it. who honestly cares one iota what the tool is? i have never understool this fanboy attitude to software... why are people so defensive over this stuff?

how do I (and I would guess many if not most others) choose a renderer?:

1) price (null point really as most of them aren't too pricey these days, or even free)
2) set-up times. (how quickly/easily you can prepare your scene, materials/lights etc)
3) rendertimes
4) required post-work (to compensate for reduced quality/realism, if applicable).
5) compatibility with other facilities/tools/studios etc.

2-4 (the most important things) can be basically summed up as "efficiency".

here's an example:

if renderer A is super-easy and quick at set-up stage but takes ages to render (in my experience maxwell is a good example of this) and renderer B takes longer at set-up but renders much quicker (Vray is quite a good example of this), then you have a difficult choice to make, because both allow you to reach the same goal in a similar time. but if render C is quicker at both then why would you choose either A or B? (Mental Ray is a good example of this). If you can get to your end result more efficiently, what reason would there be to choose an alternative? Perhaps compatibility with other facilities/studios or price. Again, this is my own opinion based on testing on real-world projects. i've simply found that with mental ray I can get to my desired end result more quickly than with maxwell, that's all... no fanboyism here, no conjecture.

i'm always in the market for a new renderer, just like many others. as soon as something comes along that will allow me to get to my end result more efficiently then you can bet I'll be all over it. it doesn't exactly help that when you ask people about their experiences with a package you get back a stream of fanboy garbage... that's ironically one big reason why i stopped using the nextlimit forums and eventually dumped maxwell.

i've seen raw renders that have come out of scanline that trump raw maxwell renders for quality and realism, and of course speed. point? it's down to the artist. that is all. and all that (should) matter when it comes to choosing the tool is its efficiency.

saying something like all vray renders look lifeless says absolutely nothing about vray. all it says it those particular artists are producing lifeless images. that might be because they're not properly up to speed with vray or they're not too hot in the post department, or... any other reason that is entirely to do with the artist not the software.

but thanks to those who took the time to post tests and examples and actually did come back with useful information.

CHRiTTeR
05-13-2009, 06:40 PM
i dont think its fanboyism thats going on here, but its more is a typical case of internet-miscommunication. :D



if renderer A is super-easy and quick at set-up stage but takes ages to render (in my experience maxwell is a good example of this) and renderer B takes longer at set-up but renders much quicker (Vray is quite a good example of this), then you have a difficult choice to make, because both allow you to reach the same goal in a similar time. but if render C is quicker at both then why would you choose either A or B? (Mental Ray is a good example of this). If you can get to your end result more efficiently, what reason would there be to choose an alternative? Perhaps compatibility with other facilities/studios or price. Again, this is my own opinion based on testing on real-world projects. i've simply found that with mental ray I can get to my desired end result more quickly than with maxwell, that's all... no fanboyism here, no conjecture.

But i really dont understand how you can manage to have long setup times in vray while you say its fast in maxwell. And what confuses me even more is that u find mental ray easier to setup while its verry simular to how vray works (actually, in max, they totally copied vray's way to setup a renderer).
Or are you using mental ray in maya or XSI (or something else except max)? Could be faster there, i dont know, i have no experience with those apps.

Just out of curiousity... Can you share with us what slows your setup times down in vray?

mister3d
05-13-2009, 06:44 PM
Ok, here's the scenes for both vray and mental ray. It's for max 2009 or higher.

http://www.sendspace.com/file/sureum

I set up the cameras for both, but I couldn't find film size in mental ray settings (and couldn't before).

If anyone knows how to resolve this issue, please elaborate.

Ok, so now anyone claiming that other renderers are capable of producing the same quality are able to prove it. Please also include your scene file with your test, so anyone could be certain it's not photoshop. Maxwell users are also welcome as the kings of the show.

That's the reference information about the scene http://www.maxwellrender.com/forum/viewtopic.php?t=14254

http://img216.imageshack.us/img216/2392/boxtessceney.jpg (http://img216.imageshack.us/my.php?image=boxtessceney.jpg)

wombat778
05-13-2009, 06:54 PM
But i really dont understand how you can manage to have long setup times in vray while you say its fast in maxwell.

I think a major point thats been missed is that the amount of setup time will vary greatly with someone's familiarity with the renderer. For example, personally I am very familiar with Mental Ray, and have never used V-Ray. I can guarantee that my setup time in Mental Ray would be substantially lower than my setup time in V-Ray, simply because in V-ray I would have to spend time just figuring out everything. This says absolutely nothing about the relative difficulty of setup by someone who has no familarity with either package.

Where I suspect Maxwell wins out it is in setup by someone with less experience with other renderers. It took me a long time of Mental Ray use to understand photons/final gather/primary and secondary rays/search radii/path tracing/filters/bounces/etc. Now that I have all that knowledge, it takes about 5 minutes for me to set up a Mental Ray scene for good GI. But, if I did not know all that (and did not already have a good sense of what values to use for the hundreds of settings), I suspect it would take me a very long time to get something good out of Mental Ray. For someone new to rendering, I have to imagine that Maxwell would be much faster to learn.

mister3d
05-13-2009, 06:59 PM
Hehe,
Tonemapping is in vray's color mapping dialogue (in vray tab of the render window).
Choose reinhard and u can set the burning and gamma values there.

D75 doesnt stand for the cameratype. It stands for a white balance at 7500K.
For 6500K u have to choose D65

Thanks for your effort to model the scene and providing us with it.
But if i can do a request, i'd like to see some maxwell versions too. Just to show that if anyone cant exaxtly replicate things like colortone and glossyness isnt because of the renderer but because its pretty hard to match those things by eye.

Great! I fixed the scenes. I misprinted, sure it's not camera, haha, just tired a little.

CHRiTTeR
05-13-2009, 07:02 PM
I set up the cameras for both, but I couldn't find film size in mental ray settings (and couldn't before), as well as "burn" function in vray settings using the vray physical camera. I set up the whitebalance of vraycamera to D75, though the test camera is D70 (but there's no preset for D70). The mental ray whitebalance is set to 6500k. I'm not sure how to set to 6500k in vray camera as there are no presets for this and only an rgb entry.

If anyone knows how to resolve these issues, please elaborate.

Ok, so now anyone claiming that other renderers are capable of producing the same quality are able to prove it. Please also include your scene file with your test, so anyone could be certain it's not photoshop. Maxwell users are also welcome as the kings of the show.


Hehe,
Tonemapping is in vray's color mapping dialogue (in vray tab of the render window).
Choose reinhard and u can set the burning and gamma values there.

D75 doesnt stand for the cameratype. It stands for a white balance at 7500K.
For 6500K u have to choose D65

Thanks for your effort to model the scene and providing us with it.
But if i can do a request, i'd like to see some maxwell versions too. Just to show that if anyone cant exaxtly replicate things like colortone and glossyness isnt because of the renderer but because its pretty hard to match those things by eye.

I think a major point thats been missed is that the amount of setup time will vary greatly with someone's familiarity with the renderer. For example, personally I am very familiar with Mental Ray, and have never used V-Ray. I can guarantee that my setup time in Mental Ray would be substantially lower than my setup time in V-Ray, simply because in V-ray I would have to spend time just figuring out everything. This says absolutely nothing about the relative difficulty of setup by someone who has no familarity with either package.

Where I suspect Maxwell wins out it is in setup by someone with less experience with other renderers. It took me a long time of Mental Ray use to understand photons/final gather/primary and secondary rays/search radii/path tracing/filters/bounces/etc. Now that I have all that knowledge, it takes about 5 minutes for me to set up a Mental Ray scene for good GI. But, if I did not know all that (and did not already have a good sense of what values to use for the hundreds of settings), I suspect it would take me a very long time to get something good out of Mental Ray. For someone new to rendering, I have to imagine that Maxwell would be much faster to learn.

exactly! but they shouldnt say its faster when in fact they dont really know :)

mister3d
05-13-2009, 07:03 PM
This was done with V-Ray way back in 2006; it can be done a lot easier now, since V-Ray has efficient mesh lights (while a self-illumination material was used for this rendering back then). Note that the image also has camera distortion to better match the original photograph.

http://www.spot3d.com/vray/images/stuff/vray_cornellLife_noAlpha.png

Best regards,
Vlado


Just curious, what is the best way to work with a mesh light (vraylightmlt?) for this test?

thev
05-13-2009, 07:03 PM
This was done with V-Ray way back in 2006; it can be done a lot easier now, since V-Ray has efficient mesh lights (while a self-illumination material was used for this rendering back then). Note that the image also has camera distortion to better match the original photograph.

http://www.spot3d.com/vray/images/stuff/vray_cornellLife_noAlpha.png

Best regards,
Vlado

CHRiTTeR
05-13-2009, 07:05 PM
Hehe, i remember that one and was silently waiting/hoping for someone to post it because i couldnt seem to find it using google.

CHRiTTeR
05-13-2009, 07:20 PM
Just curious, what is the best way to work with a mesh light (vraylightmlt?) for this test?

Vray has meshlights since SP3.
This image was rendered using an older version which didnt have meshlights yet, so thats why the caustics and shadows are a bit different.

I wonder vlado, if you still have the scene somewhere and, it doesnt take too much time (im sure you're verry bussy)... could u rerender this scene using SP3?

mister3d
05-13-2009, 07:20 PM
Vray has meshlights since SP3.
This image was rendered using an older version which didnt have meshlights yet, so thats why the caustics and shadows are a bit different.


Meshlights= vraylightmtl?
But what kind of GI in vray will work besT with it? Path tracing? This already shows that maxwell is initially more suited to work with mesh lights, as neither vray nor mental ray work well with meshlights with their adaptive GI afaik.

thev
05-13-2009, 07:26 PM
Meshlights= vraylightmtl?Nope. While the visual results are exactly the same, mesh lights are much more efficient to calculate than a simple vraylightmtl material. This is because, with a mesh light, V-Ray actually knows that it is a light and can sample it explicitly. With a vraylightmtl materail, the illumination is just a side effect of the random GI sampling, and thus takes longer to calculate.

But what kind of GI in vray will work besT with it?All GI methods of V-Ray work with mesh lights, including photon mapping.

Path tracing? This already shows that maxwell is initially more suited to work with mesh lights, as neither vray nor mental ray work well with meshlights with their adaptive GI afaik.Have you actually tried the mesh lights with SP3 of V-Ray?

Best regards,
Vlado

thev
05-13-2009, 07:29 PM
I wonder vlado, if you still have the scene somewhere and, it doesnt take too much time (im sure you're verry bussy)... could u rerender this scene using SP3?I have it, yes. I'll give it a try...

Best regards,
Vlado

CHRiTTeR
05-13-2009, 07:30 PM
Meshlights= vraylightmtl?
But what kind of GI in vray will work besT with it? Path tracing? This already shows that maxwell is initially more suited to work with mesh lights, as neither vray nor mental ray work well with meshlights with their adaptive GI afaik.

That was the case with older vray versions, but the newer SP3 has an option to have meshlight that works as a direct light.

http://www.spot3d.com/vray/help/150SP1/vray_light_mtl.htm
http://www.spot3d.com/vray/help/150SP1/light_params.htm

Im not 100% sure but i seem to remember seeing vray images that used light cache (pathtracing) before maxwell was out or even announced. Not verry important, but still a spicy little detail.

mister3d
05-13-2009, 07:42 PM
Thev, Chritter, thank you for the explanation! No... I didn't know about that as I didn't follow the improvements. This definitely adds points to Vray in terms of a realistic simulation.

MasonDoran
05-13-2009, 09:30 PM
So it was unbiased and ultra-accurate before, and it is unbiased and ultra-accurate now, but the results are obviously different. Then apparently one of the results is wrong. Can you tell which is really the "correct" one? If it is the new version, then all images done with the old one are "wrong"...


the link was provided....copy paste now:

fabrizio_S wrote:
Image 4: the big left door of the forniture has more light details in maxwell 1.7 (I mean the highlight on the frame).
Or above thedrawers...the light (shadows and bright areas) are much more convincing in maxwell 1.7. Maxwell 2 seems to be non
so accurate. And the blue tint of the sky? in maxwell 2 is less than maxwell 1.7.

Actually, there is no single change in light propagation. But, you're right about the difference in mentioned points.
I'm happy to say the differences are all about the huge improvement/fixes in our BRDF model. Together with the
help of your valuable feedback, we detected a few weak and inaccurate points our roughness model has in v1 and
we drastically improved its response to light in v2 in the name of making this engine more accurate than ever.
In the following days/weeks we will be proud to present these improvements to you in a better way.

Tora_2097
05-13-2009, 10:11 PM
Guys,

The whole point of this agression is beyond me a great many people seem to be heavily agitated by the MXW 2.0 announcement for some reason.
I do happen to know a bit about those renderings and moreover I believe I am one the very few people in this thread who can actually claim to be up to date with both Vray and Maxwell workflows from a production POV... meaning its not just speculation and ranting.

MXW will be rewritten from the ground up and saw modifcations is all areas, including tonemapping and BSDF models-there will be further announcements covering this matter in detail in the future. This is of course natural since development continues and you gather experience and know-how. Every render engine -or even software for that matter- sees updates claiming it can do things either more accurate now or faster.
So to a certain extent it could be regarded as valid if you say that renderings done in the old engine are less correct than new ones since the old engine had/has shortcomings that were impossible to tackle in the old code. It will probably be the same when MXW 3.0 comes out that it can do things again better than 2.0.
There is still only one thing that is physically correct and that is mother nature. :)
MXW was and is unbiased in a sense that it does not interpolate in any sense (hence the long rendertimes) and it is for sure the engine which is the most able to closely mimic realistic light/surface interactions.

That is what it was designed to do.

Still you need an experienced artist to use its full potential. There are a million crap renders for every software out there be it Vray, Mental Ray or MXW. You need to have the same knowhow of how light works in reality how to define surface properties in a realistic manner etc. If you feed it BS you get BS out. It certainly saves you time concerning plain rendersettings however.

Which brings me to my next point:
Use your tools in a way that you can exploit the strong parts of that software. You would not want to render a 2000frame HD animation in MXW- it is not built to do that (although in theory you could)
So I would choose a different tool in this case- at work we use Vray, it is hard to beat when you have lots of renderings at hand and need great flexibility. Productviz or -to some extent- interiorviz for example can greatly benefit from Maxwells tools and workflow however (multilight anyone?). And it does not come as a surprise that other engines incorporated aspects of MXW into their software like physical cam and/or skysystem (although Next Limit did not invent all the underlying maths to do that, they at least built a working system with it).

When talking about "production readiness" one has to say that it really depends on the task at hand. In my case -99% archviz- we use it seldomly because it is still not quite able to output the desired amount of frames in a daily turnaround- although the speedpanelty is a lot less than what most people expect, this is especially the case as scenecomplexity rises- a white room does not scratch its potential.
A lot of the clients does not even want to have it completly photoreal however (or have their very own version of reality they live in :)) , there is always cheating and often major postwork going on-and animations are for now out of question- that is why Vray is our primary tool for production work.
We mostly use Maxwell for portfolio pieces when time is of little concern and quality does matter above everything else. Plus we adapted a lot of the MXW methodology and build quite a lot of Vray shaders in a " maxwell sense".
Lastly I do not think spheres on a plane shows anything meaningful and recreating that cornell box shot is useless as well since tweaking settings until you can mimic something it pointless to me. This shot is not challenging for the engines and a more interesting thing to see would be a new scene rendered independently from one another. I can mimic that cornellbox in any renderengine once I have a reference image that I can inspect all day long.

Regards,

Benjamin

P.S.

vlad
05-14-2009, 12:12 AM
...
There is still only one thing that is physically correct and that is mother nature. :)
...

That's quite a bold claim.
Source please.

R10k
05-14-2009, 01:59 AM
That's quite a bold claim.
Source please.

*blinks*

Um... the outside world? :curious:

ZacD
05-14-2009, 02:16 AM
*blinks*

Um... the outside world? :curious:

Does such a place exist?

R10k
05-14-2009, 02:22 AM
LOL... good question ZacD. I've only heard about it in myth and legend ;)

Seriously though, I hope the above comment by vlad is a joke.

vlad
05-14-2009, 03:52 AM
LOL... good question ZacD. I've only heard about it in myth and legend ;)

Seriously though, I hope the above comment by vlad is a joke.

No man, I'm totally serious, and when you consider the discrepancies between Mother Nature 1.0 and Mother Nature 1.75, there are some interesting parallels here...

R10k
05-14-2009, 04:09 AM
I'm relieved you're considering those discrepancies :)

Jozvex
05-14-2009, 05:29 AM
Hi, just to answer your questions CHRiTTeR:

Looks like maxwell one has some extra bouncelight coming from the spere on the groundplane. Are you sure GI is on in mental ray with fair quality settings?

There's no GI in either renderer's image, you can set Maxwell to direct light only which I what I've done for that scene.

Can u post the scene files so other ppl can have a go at it?

Sure why not, here it is as a Maya ASCII and FBX file:

Maya ASCII (from 2008 sp1 but edited for any version) (http://www.jozvex.com/freestuff/Scenes/basicLambertsMaya.ma)
FBX scene (http://www.jozvex.com/freestuff/Scenes/basicLambertsFBX.fbx)

The Maya scene may complain that you don't have Maxwell if you.. don't have Maxwell. I left it set up for mental ray though. All I do for the Maxwell version, is hide the light and unhide the light geometry, change the shaders to the Maxwell lambert (already one made) and obviously select a different renderer.

Speaking of the 'real world', wouldn't it be great if you were walking somewhere and suddenly there was a blotchy/noisy/artifacty area with bad GI... it would make all our jobs much easier because we could say those things occured in nature!

My new ferrari looks like plastic because the dealer used the wrong shader! :scream:

mister3d
05-14-2009, 07:42 AM
Lastly I do not think spheres on a plane shows anything meaningful and recreating that cornell box shot is useless as well since tweaking settings until you can mimic something it pointless to me. This shot is not challenging for the engines and a more interesting thing to see would be a new scene rendered independently from one another. I can mimic that cornellbox in any renderengine once I have a reference image that I can inspect all day long.


This is partially true, because you can get most renderers (not all) to look the same if you have a reference. The point with balls was that in maxwell it was achieved out-of-the-box. But well, who knows, maybe with using correct camera settings the results were more similar.
The box example has many camera and light parameters. Still, we don't have exact shader parameters, so yes, it's a bit of guessing.

The point was to prove to Chritter that Maxwell does make even simple shaders and lighting more convincing, otherwise it wouldn't make sense to use it. This is my point.

CHRiTTeR
05-14-2009, 08:54 AM
You know, maybe its not a bad idea if developers included some more high quality presets in other renderers. Dont forget to have one called 'unbiased' though, lol ;)

First i thought it might be great if they just set the default settings to high quality, but then everybody will start complaining its not fast enough (although still faster than a pure path tracer in many cases) and that some other renderer can render much faster (but less accurate) etc.

Now the default settings are kind of balanced between speed and quality (in vray at least, mental ray is more set at speed by default in max). Which probably gives ppl a bad 'first impression' that its faster but cant do high quality.

Maybe renderers should offer an option to choose between a 'basic' and 'advanced' interface.
Because although there are a lot of options, most of them you dont need if you want to achieve 'physicall correct' results. But its nice to have them because then we can play to get more 'stylish' renders (like boosting the contrast of the GI light, or using a non physical correct light fallof, etc...).
So for those who dont require full control and just want physicall correctness, they can choose for a slimmed down (less confusing) interface with only basic controls.
And for those who like to have control over every little detail possible (like me) they can have an advanced interface with all the tweakable parameters.
Both should offer access to the presets off course.


I dont 'hate' maxwell's way of rendering or anything. I just dont like how next limit keeps on marketting it like its the only possible way to have unbiased solutions (verry untrue) and that other renderers dont use physicall correct calculations (verry untrue).
If you have to use untrue arguments to sell your product then something is probably wrong.

It was indeed easier to use, but thats not the case anymore, although it still is easier to learn because there are no confusing parameters and thats great for ppl who are new to rendering and have to learn the basics, but once you get more picky about getting the result you have pictured it in your head, you'll need more control.

Most ppl throwing the word 'unbiased around dont really understand what it stands for.
They think that it means that it works like a real camera, and because of that no other renderer can get even close to its quality. Which is verry untrue.
Yes, many other renderers take lots of shortcuts, but that doesnt mean this always influences the quality and accuracy like they want you to believe. It 'can' influence quality, but not in all cases. For example, if you want to render it fast, just use settings that are biased, if you want quality, just disable it and use more brute force sampling for example.

Fact is, they cant think of any better big argument to sell their product. It is (was) a 'new' technologie and offered a workflow to do things in another and easier way that didnt require lots of knowledge about rendering. While other renderers didnt offer this easy-to-use approach (like a physicall camera, input of actual light energy etc) at the time. But that is/was more an interface problem then a limitation of the software.

I do like fryrender though, they dont throw so much crap around and tend to keep their promesses. ;)

I suggest if you want to do or are doing a lot of rendering for a living you take a good look around on the net. There are plenty of unbiased (neutral) websites who talk about rendering and have verry intresting info about differences in render engines and methods.
Also dont believe the guy who tries to sell you something to much, especially the last few years its really becomming ridiculous how much misleading info is thrown around to take advantage of ppl who arent well informed and trust everything to much simply because they dont know. Believe me, I worked in the graphics/publishing world for a long enough period, i know what im talking about.

mister3d
05-14-2009, 09:32 AM
You know, maybe its not a bad idea if developers included some more high quality presets (including an 'unbiased' one) in other renderers.

First i thought it might be great if they just set the default settings to high quality, but then everybody will start complaining its not fast enough (although still faster than a pure path tracer in many cases) and that some other renderer can render much faster (but less accurate) etc.

Now the default settings are kind of balanced between speed and quality (in vray at least, mental ray is more set at speed by default in max). Which probably gives ppl a bad 'first impression' that its faster but cant do high quality.

Maybe renderers should offer an option to choose between a 'basic' and 'advanced' interface.
Because although there are a lot of options, most of them you dont need if you want to achieve 'physicall correct' results. But its nice to have them because then we can play to get more 'stylish' renders (like boosting the contrast of the GI light, or using a non physical correct light fallof, etc...).
So for those who dont require full control and just want physicall correctness, they can choose for a slimmed down (less confusing) interface with only basic controls.
And for those who like to have control over every little detail possible (like me) they can have an advanced interface with all the tweakable parameters.
Both should offer access to the presets off course.

Well, those additional controls will be redundant within 5-10 years with the increasing horsepower. All interpolations, sudivisions will be hidden controls. Though those who need to render animations will still use them, such as hardcore mental ray users who need fast motion blur and animated glossy interpolations.
I'm not sure why you are saying mental ray gives an impression of a worse quality? :rolleyes: Vray has a "basic" interface, and mental ray has a more "advanced", but those are aimed to different audiences. Vray is very good to those who are new to rendering and need to render architectural vizualisation. Mental ray is for a more seasoned users who know what they want from it.
Both vray and mental ray have their "physical" settings listed first, and then there are additional possibilities listed last. But yeah, there is no automatic DOF, etc, but those require real camera knowledge anyway.
Anyway, it's too early to provide such an interface. Maybe when we have 16-core processors this will make sense. I guess vray will move towards maxwell anyway, looking at it's last papers about volume caustics etc.

pluMmet
05-14-2009, 09:47 AM
I'm constantly reading on geek.com of some new way to get processors to 1 thz.

Also according to geek.com the reason they don't give us more advanced processor speeds is they stand to make too much money giving us small advances as they do.

I also read in Popular Mechanics in the mid 90's that Intel had a 1 thz processor they were giving tot he government by 2001 and a 10 thz processor by 2004.

You hear all sorts of neat stuff on cable and magazines. I saw that they can make cars that don't rust by superheating the parts after they are made and on discovery channel they had a simple process to make normal glass bullet proof.

Non of these things ever come out...


Anyway we shouldn't need to work about optimizing renderers. Eventually though the future will actually be here and processors will be fast enough to not worry about it.

mister3d
05-14-2009, 11:05 AM
CHRiTTeR, so your point is:
neither mental ray, nor vray do not differ in shader algorithms as well as other options from maxwell. So all those realistic images from maxwell is solely a speculation of developers?

CHRiTTeR
05-14-2009, 11:50 AM
Well, those additional controls will be redundant within 5-10 years with the increasing horsepower. All interpolations, sudivisions will be hidden controls. Though those who need to render animations will still use them, such as hardcore mental ray users who need fast motion blur and animated glossy interpolations.
I'm not sure why you are saying mental ray gives an impression of a worse quality? :rolleyes: Vray has a "basic" interface, and mental ray has a more "advanced", but those are aimed to different audiences. Vray is very good to those who are new to rendering and need to render architectural vizualisation. Mental ray is for a more seasoned users who know what they want from it.
Both vray and mental ray have their "physical" settings listed first, and then there are additional possibilities listed last. But yeah, there is no automatic DOF, etc, but those require real camera knowledge anyway.
Anyway, it's too early to provide such an interface. Maybe when we have 16-core processors this will make sense. I guess vray will move towards maxwell anyway, looking at it's last papers about volume caustics etc.

Im not talking about interpolation or subdivisions
Although subdivisions and samples are not verry different of maxwell's way of 'sampling' ;) ), just do more sampling to get a better result, its the same basic principle.
So no difference there.

Im am talking about how strong the GI should be, how contrasty the GI solution needs to be (without influencing the direct light) in a certain area or how sharp or blurry some shadow should be or how much light some objects should reflect or influence GI etc...
These settings will always be needed for ppl who have specific ideas in their head and dont just want everything to look 'real', many times 'real' is just boring because we experience 'real' all the time. You can get real, but if you want to achieve a certain quality/style that is impossible to get in the real world (unrealistic) or want the render to be done in a way the attention gets more directed toward a certain part of the image these controls are verry welcome and needed (and im not only talking about a case of 'just place some extra lights there').

If 'realistic' is really what we want and desire, then why is it that movies with heavy post production and color grading look so much more intresting than untouched 'raw' footage?

Where did i say mental ray gives worse quality?
This is a good example of what is going wrong with threads like this.
I said that Mental Ray's DEFAULT settings in 3dstudio max are not based on quality, but on speed. Thats all. Its perfectly possible to make an amazing, uber-quality render with mental ray, but most likely not with the default settings. :)
Vray isnt less capable then mental ray, its just configured at higher quality by default and has a cleaner interface (again, thats something personal). That may give the impression its 'simpler' at first glance but in fact, under the hood they are not so verry different and both offer enough features to achieve what you need. Except they use slightly different techniques and differently optimised maths to do the same thing and use different organised UI's to control the stuff.

Vray has a physicall 'automatic' camera too. That does have exposure, automatic dof and all that stuff.
Mental ray has its photographic exposure... dont know about automatic realistc dof though... but thats just a matter of implementing the maths and voila. Its not because the engine cant do it.
Sorry if this sounds rude, but you really shouldnt be talking about stuff you dont know for sure. Have you actually checked what vray can do before taking that conclusion?
Maxwell doesnt do 'real camera' renderings either, because that would be verry complicated and heavy thing to do. That would require you to code how every ray gets influenced by every part of the lens one by one, clearly thats not the case and would make it even slower, a lot slower. ;)
As far as i know you cant even do lens distortion (for now), but you can in vray/mental ray.
Does that mean its impossible in maxwell? No, its just not there.
Just like you couldnt do isotropic SSS in maxwell until not so long ago. :)

I wouldnt say vray is moving towards maxwell. Vray 2 will have bi directional path tracing yes, and maxwell is a path tracer. But as i mentioned before, vray had its lightcache (path tracing) even before maxwell was out or announced. So following your conclusion its the other way around and Maxwell is going the vray route? Clearly thats not the case.

Volumetric caustics are an existing physical phenomena. Its not because maxwell could do it first that vray is copying them.
Thats like saying, vray can render reflections and when maxwell came out, they could render reflections too, so maxwell is copying vray. No, again thats not the case. Reflections are required to make realistic images, so thats why they are there, just like volume caustics.
They are things that happen in real life so they are needed if you want to add realism.
It would be copying if they were calculated exactly in the same way and/or used the same interface. Which is clearly not the case.

CHRiTTeR
05-14-2009, 11:53 AM
CHRiTTeR, so your point is:
neither mental ray, nor vray do not differ in shader algorithms as well as other options from maxwell. So all those realistic images from maxwell is solely a speculation of developers?

No what im saying is that their algorithms can result in the same result.

To give a verry simplifying example.

One may do 2 + 2 = 4
While the other does 1 + 1 + 1 + 1 = 4
Or one may even do 2ē = 4
And a verry badly optimised one will do (6 * 5 )ē / (5ē * 9) = 4

Different approaches, same result :)

Another example:
You have a simple ground plane with a gradient texture on it.
One renderer will calculate the color by tracing a lot of rays pixel by pixel while another renderer is smart enough to see it can just check the 2 colors at the beginning and end of the gradient and interpolate those values from beginning to ending to recreate a gradient in less time. Now, again this is a verry simplifying example, but its good enough to give you a basic idea i think.

mister3d
05-14-2009, 12:25 PM
Where did i say mental ray gives worse quality?
This is a good example of what is going wrong with threads like this.
I said that Mental Ray's DEFAULT settings in 3dstudio max are not based on quality, but on speed. Thats all. Its perfectly possible to make an amazing, uber-quality render with mental ray, but most likely not with the default settings. :)
Vray isnt less capable then mental ray, its just configured at higher quality by default and has a cleaner interface (again, thats something personal). That may give the impression its 'simpler' at first glance but in fact, under the hood they are not so verry different and both offer enough features to achieve what you need. Except they use slightly different techniques and differently optimised maths to do the same thing and use different organised UI's to control the stuff.

Saying that vray isn't less capable than mental ray is not correctly formulated. Yes, for some tasks one will win, another will crawl. Yet saying they are very similar is not correct, as there are a lot of features missing in vray, such as:
luma shaders
glossy interpolations suitable for animations
rapid mblur
advanced shader trees, such as mental ray has, mental ray wins with its shaders for complex tasks
many features of arch&design shader
some production shaders support



Vray has a physicall 'automatic' camera too. That does have exposure, automatic dof and all that stuff.
Mental ray has its photographic exposure... dont know about automatic realistc dof though... but thats just a matter of implementing the maths and voila. Its not because the engine cant do it.

I know about vrayphysical camera, moreover I know most features of vray (not new ones of the sp3 though). It's strange you ask this as the scenes I put yesterday have both mental ray ph.exposure set and vray physical camera, so perhaps you just forgot.
Mental ray shaders provide more flexibility and choices, and that's undeniable.


Sorry if this sounds rude, but you really shouldnt be talking about stuff you dont know for sure. Have you actually checked what vray can do before taking that conclusion?


Of course, I did tests for both engines, where one wins, another loses. And those gaps are pretty evident, and differ much for different situations.


Maxwell doesnt do 'real camera' renderings either, because that would be verry complicated and heavy thing to do. That would require you to code how every ray gets influenced by every part of the lens one by one, clearly thats not the case and would make it even slower, a lot slower. ;)
As far as i know you cant even do lens distortion (for now), but you can in vray/mental ray.
Does that mean its impossible in maxwell? No, its just not there.
Just like you couldnt do isotropic SSS in maxwell until not so long ago. :)

Well, lens distortion is just a joke, if you mean a barrel distortion. Still, I'm not a shader expert so I can't really guess what makes maxwell renders look more realistic, but as shader algorithms differ, I think they are the reason.

No what im saying is that their algorithms can result in the same result.

Of course, but as long their algorithms differ, you shouldn't make such statements because they are misleading and you didn't test both engines to claim that.

CHRiTTeR
05-14-2009, 01:04 PM
Saying that vray isn't less capable than mental ray is not correctly formulated. Yes, for some tasks one will win, another will crawl. Yet saying they are very similar is not correct, as there are a lot of features missing in vray, such as:
luma shaders
glossy interpolations suitable for animations
rapid mblur
advanced shader trees, such as mental ray has, mental ray wins with its shaders for complex tasks
many features of arch&design shader
some production shaders support

true, but theres also features that vray has, but arent in mental ray (unless you make a shader for it, but thats also possible with vray).

I dont see how complex shader trees are automaticly a good thing, more a bad thing if you ask me as it makes calculations more complicated.

So first you say maxwell shaders are the best, but yet they are verry simple.
And now you say that mental ray has better shaders then vray because they are more complex.
make up your mind.

I know about vrayphysical camera, moreover I know most features of vray (not new ones of the sp3 though). It's strange you ask this as the scenes I put yesterday have both mental ray ph.exposure set and vray physical camera, so perhaps you just forgot.
Mental ray shaders provide more flexibility and choices, and that's undeniable.

Then why did you say you cant do automatic dof like in maxwell?
I haven had the occasion yet to play with that scene, i even didnt download it yet.
As Vlado already posted a quite convincing render and is going to rerender it in SP3, so no need for me to spend time recreating the shaders etc... because i know he'll do it better then me.
Also, i dont have access to vray here as i quit my job recently.

Well, lens distortion is just a joke, if you mean a barrel distortion. Still, I'm not a shader expert so I can't really guess what makes maxwell renders look more realistic, but as shader algorithms differ, I think they are the reason.

In terms of realistic shader. Im sure no maxwell render out there is correct either. You need a verry high res scanned/modeled (volumetric) version at a molecular level of the enviroment for that. Also shading should be applied at a molecular level.

All im saying is that vray can produce the same results as maxwell, but faster (in almost all cases). Sure the numbers you need to input on non physicall based values will probably be not the same. But thats not important at all, because even in your uber-holly maxwell shaders the roughness value is not a phycally based value (unlike things like ior ,but again, this is also correct in vray).

mister3d
05-14-2009, 01:32 PM
true, but theres also features that vray has, but arent in mental ray (unless you make a shader for it, but thats also possible with vray).

I dont see how complex shader trees are good thing, more a bad thing if you ask me as it makes calculations more complicated.

It's not about how complex they are, but more possibilities for a demanding user. But as long as vray starts supporting the arch&design shader, maybe in the future it will be supported completely, giving vray users more options. There are some fun additions like controlling specular intensity separately from refl through shader etc. Still, I doublt vray will adopt all the shaders of mental and its strong sides. So those are just different animals, for now.


Then why did you say you can do automatic dof like in maxwell?

You got me wrong: I don't know how DOF is implemented in maxwell, but it sounds like it's on by default, whereas it isn't neither in vray nor mental. That's what I meant by automatic DOF, must be my missplelling, I should say "turned on by default".


In terms of realistic shader. Im sure no maxwell render out there is correct either. You need a verry high res scanned/modeled (volumetric) version at a molecular level of the enviroment for that. Also shading should be applied at a molecular level.

All im saying is that vray can produce the same results as maxwell, but faster (in almost all cases). Sure the non physicall based values you need to input will probably be not the same. But thats not important at all, because even in your uber-holly maxwell shaders the roughness value is not a phycally based value (unlike things like ior ,but again, this is also correct in vray).

Maybe not "correct", and many make fun of it, but as "Maxwell Render™ can fully capture all light interactions between all elements in a scene no matter how complex they are. All lighting calculations are performed using spectral information and high dynamic range data."
- this stands for something. You are in denial of it. Once again, I haven't seen such realistic images prior to maxwell. Yes it gives some more realistic results which still can be killed by a user. But there's something special about maxwell. I saw renders and thought "this looks really realistic and stands out", and often it was maxwell.

ZacD
05-14-2009, 01:35 PM
I'm constantly reading on geek.com of some new way to get processors to 1 thz.

Also according to geek.com the reason they don't give us more advanced processor speeds is they stand to make too much money giving us small advances as they do.

I also read in Popular Mechanics in the mid 90's that Intel had a 1 thz processor they were giving tot he government by 2001 and a 10 thz processor by 2004.

You hear all sorts of neat stuff on cable and magazines. I saw that they can make cars that don't rust by superheating the parts after they are made and on discovery channel they had a simple process to make normal glass bullet proof.

Non of these things ever come out...


Anyway we shouldn't need to work about optimizing renderers. Eventually though the future will actually be here and processors will be fast enough to not worry about it.

Super computers and computer clusters maybe, but this is the future right now http://download.intel.com/pressroom/kits/research/40G_modulator_presentation.pdf

CHRiTTeR
05-14-2009, 01:47 PM
It's not about how complex they are, but more possibilities for a demanding user. But as long as vray starts supporting the arch&design shader, maybe in the future it will be supported completely, giving vray users more options. There are some fun additions like controlling specular intensity separately from refl through shader etc. Still, I doublt vray will adopt all the shaders of mental and its strong sides. So those are just different animals, for now.

Funny thing is, the arch & design material is just a rip off of the standard vray material.


You got me wrong: I don't know how DOF is implemented in maxwell, but it sounds like it's on by default, whereas it isn't neither in vray nor mental. That's what I meant by automatic DOF, must be my missplelling, I should say "turned on by default".

ok, thats true, but is that really important? It takes less then 10 seconds to click the on button.

Maybe not "correct", and many make fun of it, but as "Maxwell Render™ can fully capture all light interactions between all elements in a scene no matter how complex they are. All lighting calculations are performed using spectral information and high dynamic range data."
- this stands for something. You are in denial of it. Once again, I haven't seen such realistic images prior to maxwell. Yes it gives some more realistic results which still can be killed by a user. But there's something special about maxwell. I saw renders and thought "this looks really realistic and stands out", and often it was maxwell.

Well, one thing is sure, at least 1 of us is in denial :)

Seriously... I asked you before about spectral calculations. You dont know what they are or how they are different to vray's calculation yet you seem convinced you can use those as an convincing argument.
With some explanation you clearly copy pasted from somewhere but doesnt tell me anything. Just some fancy words.
what if i ask you if vray uses or can use spectral calculations? im sure you dont know for sure either and probably have to google it or something.

Thus so far, for you its just a flashing word you captured somewhere in their advertisements and you instantly get the impression that its the best thing ever since cold beer. Yet, you cant say why?

Thats like saying detergent A is better than detergent B, because the commercial says so!
It uses ultra blue crystals! What do the crystals do? I dont know, but is has them!!! yaaaay!

Every time i expressed my opinion i tried to explain why with enough facts to back it up in an understandable way i think, but the only thing you keep repeating is 'it just looks more realistic'.

You clearly dont have a basic understanding about the differences between pathtracing and brute force qmc and yet you seem to be convinced they cant give simular results?

But there's something special about maxwell. I saw renders and thought "this looks really realistic and stands out", and often it was maxwell.

Often... not always ;)

BigPixolin
05-14-2009, 01:58 PM
Seriously... I asked you before about spectral calculations. You dont know what they are or how they are different to vray's calculation yet you seem convinced you can use those as an convincing argument.
With some explanation you clearly copy pasted from somewhere but doesnt tell me anything. Just some fancy words.
what if i ask you if vray uses or can use spectral calculations? im sure you dont know for sure either and probably have to google it or something.

Thus so far, for you its just a flashing word you captured somewhere in their advertisements and you instantly get the impression that its the best thing ever since cold beer. Yet, you cant say why?


We had a guy here that used to do the exact same thing.
At first he was saying how Maxwell calculates everything in infrared. Which just baffled me. I then found out he was talking about spectral calculations.
Seems their marketing engine is getting exactly who it needs to get.

mister3d
05-14-2009, 02:10 PM
Yet vray is now working on support for this rip-off shader tree. If you are convinced that vray shaders provide the same flexibility and options as mental ray ones, good for you. I gave you enough reasons why it isn't so.
I don't use path tracing for obvious reasons, but I know the difference, I'm not sure how this concerns our discussion.
Yes I don't know what the benefits the spectral rendering provides and I trust my eye here, claiming there is something different about maxwell renders. You seem either incapable to prove otherwise whether it gives a benefit of not, your statements are "you can achieve the same results in vray as in maxwell", ignoring the differences in the engines.
As far as I know vray does not use spectral calculations for now. That's why you yet to see believable diamonds in either of them without faking through rgb refraction inputs.
Your argument that "you can create realistic image in vray etc" has nothing to do with maxwell, that's the main problem, as you didn't prove whether it has considerable differences in its algorithms. Why do you think people are waiting for hours for maxwell frames? :)
I think this dicsussion makes not much sense already as neither you, nor me use maxwell, so it just has not any ground based on from both sides. The best proof is experience with examples, but I can't compare those engines as I don't have maxwell.My point there IS difference between maxwell and vray renders, and it's not just unbiased approach of maxwell or talented users.
I created a test scene just for this reason, bacause such speculations and guesses can continue forever. Yet it's not a perfect testing example.

CHRiTTeR
05-14-2009, 02:17 PM
dont try that inverse psychology crap on me, lol ;)

pluMmet
05-14-2009, 08:19 PM
Super computers and computer clusters maybe, but this is the future right now http://download.intel.com/pressroom/kits/research/40G_modulator_presentation.pdf

Not even close :cool:


Graphene Multiplier (http://www.geek.com/articles/chips/mit-unveils-graphene-chip-could-lead-to-1000-gigahertz-processors-20090326/) takes the noise out of a chip which takes away the heat which allows you to make it faster.

Of course this is just the latest in a long line of things I've seen on geek.com that allow for a 1 thz chip. Non of which ever come out.

Bubbaloo
05-14-2009, 09:12 PM
I don't know what the benefits the spectral rendering provides

This was done a while ago with Maxwell. It can be faked in Vray or Mental Ray, but with Maxwell, I used a silicon dioxide ior file.

http://i186.photobucket.com/albums/x267/Bubbaloo1974/Diamond_sm.png

mister3d
05-14-2009, 09:47 PM
Bubbaloo, that looks really good: very natural transitions of dispersion, I doubt you can make it look as well in mental or vray.
And I must say it's a real pain when you need to make diamonds for a viz and all those wonderful engines can't do this :argh: . And those fakes don't look as convincing at all.
I understand why developers don't include such an option for now, but it's a long time when people need this badly.

Magnus3D
05-14-2009, 10:03 PM
That's a pretty one you made there Bubba :) not sure i seen it before.

/ Magnus

Bubbaloo
05-14-2009, 10:04 PM
I doubt you can make it look as well in mental or vray

Oh, yes you can... just not as easily. That was the point of my post. Maxwell's ease of use is a trade-off for it's longer render times. By the time you set this up in Vray or mental ray and then did the post work, your Maxwell render would probably be done. With version 2 arriving soon, that trade-off will become less, though...

BTW, thanks for the compliment!

BigPixolin
05-14-2009, 10:19 PM
Tradeoff for setup times only applies to still images. Also I don't want to sound like a a$$ but if you don't think you can get that image from MR or vray then you really should not be commenting on this.

vlad
05-14-2009, 10:42 PM
Bubbaloo, that looks really good: very natural transitions of dispersion, I doubt you can make it look as well in mental or vray.
And I must say it's a real pain when you need to make diamonds for a viz and all those wonderful engines can't do this :argh: . And those fakes don't look as convincing at all.
I understand why developers don't include such an option for now, but it's a long time when people need this badly.

So I guess NextLimit must be very grateful to jewellers then :beer:
What I'd like to see now is Maxwell tackling fake diamonds

simonenastasi
05-15-2009, 05:00 AM
Tradeoff for setup times only applies to still images. Also I don't want to sound like a a$$ but if you don't think you can get that image from MR or vray then you really should not be commenting on this....or Brazil. In Brazil (since ver.1) it's as easy as turn on "dispersion"...

JorgeIvanovich
05-15-2009, 05:45 AM
Funny thread, bottomline Maxwell itīslow like hell and anyone using this in production with todays deadlines, have more balls than Dirty Harry.

Vray itīs good until you charge the scene a lot,then mental ray becomes a better option.

I use Vray for arch,Mental for more demanding character stuff involving displace and sss and 3delight for maya for really heavy hair scenes.

We use vray in some comercials in the studio but vray always have issues with memory you have to do several things to keep the engine happy.Mental ray memory managment itīs better.Wish Mental have point based occ and better hair render times.Hair in mental take ages.And in maya with Fast dont look good.

For me 3delight for maya itīs the best,wish know some about shading code to use more.

I see a lot of nice Maxwell renders,but all i see can be done in vray without big render times.

This itīs done with 3delight a small hair test ,200000 curves no post effect, just geometry

1920*1080..in 3 minutes,this itīs what i call a very good optimized engine.I donīt want fancy things that take ages, i want a engine that let me cheat reality, not just seat my lazy ass to wait renders.

http://img41.imageshack.us/img41/4669/hairp.th.jpg (http://img41.imageshack.us/my.php?image=hairp.jpg)

mister3d
05-15-2009, 08:53 AM
Tradeoff for setup times only applies to still images. Also I don't want to sound like a a$$ but if you don't think you can get that image from MR or vray then you really should not be commenting on this.

I made fake dispersion in both renderers. You seem to misunderstand what I said: you can get close to it, but only to an extent. They have their limitations.

Zemmuonne, I know, but will you use it just for diamonds? I mean this should be present in every renderer long time ago.

CHRiTTeR
05-15-2009, 08:54 AM
Oh, yes you can... just not as easily. That was the point of my post. Maxwell's ease of use is a trade-off for it's longer render times. By the time you set this up in Vray or mental ray and then did the post work, your Maxwell render would probably be done. With version 2 arriving soon, that trade-off will become less, though...

BTW, thanks for the compliment!


Yes it can be just as easy.
I already mentioned way back in this thread that dispersive effects are easier to implement in a spectral renderer for the developper.
That doesnt mean its impossible to implement in a renderer that doesnt use spectral calculations. And once its implemented its just as easy as putting an extra slider in the UI.

The reason why its not there in vray (or mental ray) is because it isnt (wasnt) on the top of their development list.
Have a look at brazil. It has dispersion since version 1 that was released waayyyyyy back ;)
(Although i must admit im not sure if it was based on physicall correct calculations or just a quick 'hack', but i think to recall it is/was physicall correct). :)

Just like maxwell doesnt have enviromental fog or clouds.

Really, we are going in circles here. ;)


I dont agree, configuring vray is quite fast. It certainly doesnt take hours.
(again, i dont have much experience in mental ray in this, so i cant speak for mental ray)

mister3d
05-15-2009, 08:56 AM
Yes it can be just as easy.
I already mentioned way back in this thread that dispersive effects are easier to implement in a spectral renderer for the developper.
That doesnt mean its impossible to implement in a renderer that doesnt use spectral calculations. And once its implemented its just as easy as putting an extra slider in the UI.

The reason why its not there in vray (or mental ray) is because it isnt (wasnt) on the top of their development list.


Yes, and it would be nice, don't you think? And here comes your own reason why vray will move to a maxwell path in a way. At least by implementing spectral rendering. And then I think it will beat maxwell.
It's pretty evident that vray started implementing things that were just not feasible couple of years ago in terms of a rendering speed for users. That's pretty interesting and maybe will change the vray paradigm to a more sophicticated, and therefore more complex renderer. This is really promising.

CHRiTTeR
05-15-2009, 09:01 AM
This itīs done with 3delight a small hair test.

we're only talking about different raytracers here (mainly maxwell, as this is a maxwell topic).
Its already complicated and confusing enough for most ppl it think, lets not start a rasteriser vs raytracer discussion also. That's going to be a big confusing mess. :)

CHRiTTeR
05-15-2009, 09:22 AM
Yes, and it would be nice, don't you think? And here comes your own reason why vray will move to a maxwell path in a way. At least by implementing spectral rendering. And then I think it will beat maxwell.

I think i clearly explained that this isnt the case.
I could also tell you maxwell is copying vray or mental ray by using a simular argument.
They are not copying eachother. They are both renderers, so off course they share simular features. But they do things in a different way and have their own vision and philosophy about things and they chaos group are verry good at optimising things to get the most out of it while keeping possible future additions in mind.
But what this discussion is about is that vray-like renderers can give you simular results and are also physicall correct (thats just a matter of using physicall correct math) + almost always faster and arent that hard to use anymore.

BTW, brazil doesnt use spectral calculations and still it can do dispersion.
Just because an effect displays all (visible) colors of the spectrum doesnt automaticly mean it uses or needs spectral calculations.

See, this is the problem. You dont know what it actually means, yet you keep using as an argument in the hopes to be right because you are brainwashed yo believe so, but it actually proves you dont know what you're talking about.

mister3d
05-15-2009, 09:37 AM
Where did I say they are copying each other? I said vray moves to a more physically correct renderer.
Ok, you made me to read about this spectral rendering.
"In computer graphics, spectral rendering is where a scene's light transport is modeled with real wavelengths. This process is typically a lot slower than traditional rendering, which renders the scene in its red, green, and blue components and then overlays the images. Spectral rendering is often used in ray tracing or photon mapping to more accurately simulate the scene, often for comparison with an actual photograph to test the rendering algorithm (as in a Cornell Box) or to simulate different portions of the electromagnetic spectrum for the purpose of scientific work. The images simulated are not necessarily more realistic appearing; however, when compared to a real image pixel for pixel, the result is often much closer.

Spectral rendering can also simulate light sources and objects more effectively, as the light's emission spectrum can be used to release photons at a particular wavelength in proportion to the spectrum. Objects' spectral reflectance curves can similarly be used to reflect certain portions of the spectrum more accurately.

As an example, certain properties of tomatoes make them appear differently under sunlight than under fluorescent light. Using the blackbody radiation equations to simulate sunlight or the emission spectrum of a fluorescent bulb in combination with the tomato's spectral reflectance curve, more accurate images of each scenario can be produced."

Well, if it means nothing to you, then :argh: .

In other words, if vray calculates rgb and getting the abstract 3d space, maxwell calculates light wavelenghts from infrared to ultraviolet, right?I'm sure that's what makes maxwell renders look more convincing.

CHRiTTeR
05-15-2009, 10:12 AM
Where did I say they are copying each other? I said vray moves to a more physically correct renderer.

No you didnt say that, ok my bad.
You said vray is more moving towards maxwell.
I can easyly say the same about maxwell using the same arguments.
Ok, now what? What is the point of mentioning this?


Ok, you made me to read about this spectral rendering.

Finally!
Now i wonder, how long have you been defending/justifying your investment by using these arguments and not knowing what it really did (until now).

"In computer graphics, spectral rendering is where a scene's light transport is modeled with real wavelengths. This process is typically a lot slower than traditional rendering, which renders the scene in its red, green, and blue components and then overlays the images. Spectral rendering is often used in ray tracing or photon mapping to more accurately simulate the scene, often for comparison with an actual photograph to test the rendering algorithm (as in a Cornell Box) or to simulate different portions of the electromagnetic spectrum for the purpose of scientific work. The images simulated are not necessarily more realistic appearing; however, when compared to a real image pixel for pixel, the result is often much closer.

Especially notice how much they use the word 'often'.
And you need to compare pixel by pixel to notice really minamal difference.
Now, look at the thread's name, click on the link in the first post and look at the examples.
Clearly(!) Something is not right. There's more then just a bit of minimal 'pixel by pixel' difference going on. And this is both using the same 'infamous automatic realistic spectral physical renderer'. Not even a different one!
So while spectral calculations can have the advantages your quoted source mention, this is clearly not the case in maxwell when they market and sell it by making you believe so.
Thats why i mentioned earlier i dont hate its tech.

As an example, certain properties of tomatoes make them appear differently under sunlight than under fluorescent light. Using the blackbody radiation equations to simulate sunlight or the emission spectrum of a fluorescent bulb in combination with the tomato's spectral reflectance curve, more accurate images of each scenario can be produced."

That is if you do the shading of your tomatoe right and for that you'd need a measured brdf.
Now, like many times says before. It comes all down to the artist.
Just because you're technology can do things like this 'better' in theory its fairly safe to say that in practice the difference between a maxwell render and a vray render and their effect on tomatoes will give comparable differences because of human inaccuracy and user input!

This is all nice and well when you got some tech at home to measure exact and precise properties of tomatoes and their surrounding and the enviroment (floating dust particles etc) and the lighting. Input that in your renderer, but really do you have such (probably complicated and expensive) equipment?
And even then, all this has to be at molecular level. 1 slight error and results arent correct anymore.

But even more important (where it all starts), the shape/model of you lights have to perfectly match the real thing (not just some sphere or plane) and you need to remodel every detail that influences the light direction (scratches, screw,...). If thats already wrong to begin with all other effort is useless in terms of achieving exact results.

Suddenly not so easy anymore heh?

I all mentioned this before btw. :)

Now, is it really worth to wait so much longer because it offers a really minimal difference in realism that you'll never notice anyway because of user (model/volume and material/shader/texture) inaccuracy?

mister3d
05-15-2009, 10:55 AM
Yes, those maxwell examples look suspicious. But i'm not after an "absolutely correct" idea, but about closer one.
And that's true, maxwell will move towards vray with time, otherwise they will stuck with their "unbiased" rigid idea. They just needed to make an impression of something completely new.
The point is that maxwell arrived before its time, it doesn't mean it's unusable. It's just too slow for now.
CHRiTTeR, I respect your level of knowledge, but untill we run some clever tests it's all speculation, you know. Yes, maybe, maybe, but noone can say for sure.
I didn't say that maxwell is precise, but it may be closer to a more accurate renderer, why not?
Of course nobody is talking about precise measured reflectance for all kinds of object, but isn't maxwell uses more complex, or measuared BRDF's? I don't know, so I'm asking. For example if you make approximations of BRDF's for most common materials, it may work.

CHRiTTeR
05-15-2009, 11:47 AM
Yes, those maxwell examples look suspicious. But i'm not after an "absolutely correct" idea, but about closer one.

I understand, but really there are so much factor influencing its accuracy that its not really needed.
I can tell you quite surely the same level of 'accuracy' you get in the end on your screen is achievable with vray or mental ray.

And that's true, maxwell will move towards vray with time, otherwise they will stuck with their "unbiased" rigid idea.

I think they are more 'moving' towards eachother, but they will implement features differently.

They just needed to make an impression of something completely new.
The point is that maxwell arrived before its time, it doesn't mean it's unusable. It's just too slow for now.

Yes, its technologie is kind of incredible if you think of it. But also renderers we are more used to like vray are verry impressive.
Its like flying with an airplane when you're going on a vacation. Ppl dont realise how incredible the fact is they are flying and can get around the world in no-time. Because they are used to it, it doesnt seem impressive, but it still is.

CHRiTTeR, I respect your level of knowledge, but untill we run some clever tests it's all speculation, you know. Yes, maybe, maybe, but noone can say for sure.

I dont know much. Theres faaaaaar more skilled and smarter ppl on these forums then me (and maybe/probably they think im stupid).
I just have few intrests and like to do some research and try to reason using facts.
And i always check different sources (what does A say, then what does B say and then check/compare that using neutral (unbiased) sources) before taking conclusions.
Its not that timeconsuming or big effort now we all have internet. But its still kind of tricky to find good and reliable sources, yes.

Especially because i worked in publishing for some time i know you NEVER can trust slogans or commercials. Do your research before you wanna buy something thats fairly important and/or expensive. Dont listen to the guy who is getting your money. :)

I think its pretty clear whats going on here.
They said it was ultrarealistic and unbiased before.
Now, same scene, different results and its still ultrarealistic and unbiased.
Now as far as i know, the laws of physics didnt change right? So somewhere theres something not right.

Although i noticed in the white shaded/textureless interior with the chairs. That the sunlight is falling at a slightly different angle. So maybe theres something extra in the lighting?

But then on the other hand, by the looks of the images the newer ones 'look' far less realistic and do indeed 'seem' to have some kind of interpolation going on. Also, i still believe the verry first version of maxwell gave best result (but it was also the slowest one).
Because of this i get the impression they are sacrificing quality for speed.

But i am carefull though, i mentioned i wanted to see the final (noise free) results and compare those. And i am thinking what other factors may influence this change in quality/result without actually changing precission and physicall laws, but cant think of one.
So thats im telling here int the hopes to get a clear answer?

I didn't say that maxwell is precise, but it may be closer to a more accurate renderer, why not?
Of course nobody is talking about precise measured reflectance for all kinds of object, but isn't maxwell uses more complex, or measuared BRDF's? I don't know, so I'm asking. For example if you make approximations of BRDF's for most common materials, it may work.

Well the accuracy gets lost by factors like the inaccurate textures and models anyway, so its not really worth the effort to invest time and money at such accurate tracing of light.
Until we get perfectly captured models and textures/material that will really take advantage of this kind of precision. Probably in the form of verry high resolution voxels, where every voxel has its own property. Or like verry verry dense and little particles.
But to model something at these resolutions is probably impossible and scans seem to be a more acceptable method. This is not something an average architect can do easyly, hehe.

Im still curious what the changes are that result in such different results though.
And if it turns out to be something i may have overlooked or dont know, i wont feel stupid really. Because i just place my thoughts here and no one else seems to know either... :)

The shaders in maxwel still relly a lot on guessing. Yes you can use measured .ior files. But where can I find .ior files of a certain less-popular material i need to visualise for example?
The next vray will also have measured materials, but here again, you still need to find and have the files containing all the measured data before you can use it.
So most materials will still be made by guessing values, just like in any other popular renderer. So no matter how accurate the calculations using these factors/inputs are, if 1 input isnt right/correct/realist, the final result wont be right/correct/realistic.
We'll have some sort of huge library of measured materials over time, that would be sweet, but then again there's the model, textures...

mister3d
05-15-2009, 01:06 PM
I understand, but really there are so much factor influencing its accuracy that its not really needed.
I can tell you quite surely the same level of 'accuracy' you get in the end on your screen is achievable with vray or mental ray.

Using references - yes. Out-of-the-box - needs testing along with maxwell.


I think they are more 'moving' towards eachother, but they will implement features differently.


Well yeah, I agree.


But then on the other hand, by the looks of the images the newer ones 'look' far less realistic and do indeed 'seem' to have some kind of interpolation going on. Also, i still believe the verry first version of maxwell gave best result (but it was also the slowest one).
Because of this i get the impression they are sacrificing quality for speed.


This really proves that techniology matters too, not just an artist, as you can see.
I will keep an eye on those noise issues too as they are contradicting.


Well the accuracy gets lost by factors like the inaccurate textures and models anyway, so its not really worth the effort to invest time and money at such accurate tracing of light.
Until we get perfectly captured models and textures/material that will really take advantage of this kind of precision. Probably in the form of verry high resolution voxels, where every voxel has its own property. Or like verry verry dense and little particles.
But to model something at these resolutions is probably impossible and scans seem to be a more acceptable method. This is not something an average architect can do easyly, hehe.

It's not about only how shaders look, but also how light interactions with them, right? So those subtleties you see in maxwell versions may be a good proof that the idea is valid.
It's clear that even measured BRDF's are just rough approximations so it's not possible in any way in the nearest future, but phong's and blinns may become obsolete in the next decade with more advanced models. Just awild guess.



The shaders in maxwel still relly a lot on guessing. Yes you can use measured .ior files. But where can I find .ior files of a certain less-popular material i need to visualise for example?
The next vray will also have measured materials, but here again, you still need to find and have the files containing all the measured data before you can use it.
So most materials will still be made by guessing values, just like in any other popular renderer. So no matter how accurate the calculations using these factors/inputs are, if 1 input isnt right/correct/realist, the final result wont be right/correct/realistic.
We'll have some sort of huge library of measured materials over time, that would be sweet, but then again there's the model, textures...

It would be enough to have something like 20-50 measured BDRF's for an average joe, such as metals, plactics, ceramics, glass, wood, polished wood, rubber, etc.
Well textures is another talk, as well as artistic lighting. Yes those are important, but underestimating lighting and shaders here is not a good idea: just imagine you need to go with scanline again. Yes, you can do somw things there, but not as much at all. Fancy textures won't save all shots in viz.

Bubbaloo
05-15-2009, 01:13 PM
Do you guys have jobs?:eek:

CHRiTTeR
05-15-2009, 01:36 PM
It's not about only how shaders look, but also how light interactions with them, right? So those subtleties you see in maxwell versions may be a good proof that the idea is valid.


Well actually shading just means that -> How light interacts with a material based on its physicall properties and thats what gives them their look. So thats a verry true statement :)

If there are differences in maxwell's shader and vray shaders (you have choice between 3 in vray), which probably is the case, then off course they too will behave differently and give different results. But there is no thing as an 'universall shader' in real life, so there will be things that look more real using maxwell's shading and there will be things that look more real using one of vray's shaders. Again, little differences and both will be probably not exact.

It's clear that even measured BRDF's are just rough approximations so it's not possible in any way in the nearest future, but phong's and blinns may become obsolete in the next decade with more advanced models. Just awild guess.

I dont know, i think its possible to do verry exact messurements though, but maybe (and highly probably) this will result in so much data that its not really usable for this kind of usage.
But im sure the measured date used will result in results that are accurate enough for the human eye and a computer screen. :)

Its possible that we wont need blins and phongs anymore, and would have a huge online library of measured shaders or something yes.


It would be enough to have something like 20-50 measured BDRF's for an average joe, such as metals, plactics, ceramics, glass, wood, polished wood, rubber, etc.


Yes that would help a lot, no doubt. Especially for things like carpaint and plastic.
But here again we have the problem its just an approxiamtion (like you already said), because there are tons of different carpaints and plastics who all behave different to light.



Well textures is another talk, as well as artistic lighting. Yes those are important, but underestimating lighting and shaders here is not a good idea: just imagine you need to go with scanline again. Yes, you can do somw things there, but not as much at all. Fancy textures won't save all shots in viz.

Well its how it all works togetter. One is not more important then another if you want highest possible realism and accuracy.
So you can have the best lighting and/or shading, but if the textures arent good, its useless in terms of getting an accurate and realistic image.
Same thing counts the other way.
You can have the best textures ever, but if the lighting and/or shading isnt right, its useless in terms of getting an accurate and realistic image.

Since we dont have verry accurate and correct textures (lets even forget about shading), the results will be 'wrong' compared to real life no matter how good and accurate your lighting model is.

CHRiTTeR
05-15-2009, 01:47 PM
Do you guys have jobs?:eek:

Not anymore, lol. Does it show? :scream: :blush:

BigPixolin
05-15-2009, 02:44 PM
Do you guys have jobs?:eek:

lol yes they require animation and tight deadlines. Which is why I'd like to see Maxwell 30x faster.

Bubbaloo
05-15-2009, 02:58 PM
Realtime Maxwell, it would be nice!

mhampton
05-15-2009, 02:59 PM
Personally... I would rather spend my time creating an image and thinking "I want to put a blue light over there" and simply put a blue light "over there" and be done with it, rather than spending time fighting with the software/render engine to make that blue light behave like a blue light.

In that instance... it seems that Maxwell wins. I would love to be able to use Maxwell. However, I also work in a design environment that the only thing that would be more intense would be going to war. The client making changes on you 2, 3 or 4 times? Bah... that's nothing... what about the designers you are working with changing something on you no less than 47 times BEFORE the client even ever sees the rendering/animation?

So, for me, 30 minute/1 hour/2 hour+ render times are absolutley out of the question. Especially for a 1200x1000 pixel image (or even smaller)??? What are the render times at 3000x1500 or so? 8 hours? 24 hours? Forget it! And animations... don't even think about it. So, I believe Vray/Mental Ray wins in the render times.


Maybe in 5-10 years as computer speed increases.... render engines might move more to work like Maxwell.

For those that might only have to re-render an image no more than a couple of times, and arn't doing animations.. I can see where Maxwell would be a very good solution, for everything else.. not so much.


For the Maxwell users... Can Maxwell distrubute the rendering of a single image accross multiple render nodes? Or does a single image need to be rendered on a single node?


P.S... I have heard of people who will do a very quick down and dirty rendering of a scene in Maxwell (with lots of noise and such.. low settings, whatever it takes to render it quickly), but then use the result of that as a reference to what the "reality" should be in order to set up their lighting in another engine (vray/mr).

BigPixolin
05-15-2009, 03:08 PM
Realtime Maxwell, it would be nice!

No arguments there!:thumbsup: Realtime maxwell with a more "advanced controls" option and I'd be a happy camper.

BigPixolin
05-15-2009, 03:09 PM
P.S... I have heard of people who will do a very quick down and dirty rendering of a scene in Maxwell (with lots of noise and such.. low settings, whatever it takes to render it quickly), but then use the result of that as a reference to what the "reality" should be in order to set up their lighting in another engine (vray/mr).

Cool idea. I may give that a try. Thanks.

mister3d
05-15-2009, 03:10 PM
P.S... I have heard of people who will do a very quick down and dirty rendering of a scene in Maxwell (with lots of noise and such.. low settings, whatever it takes to render it quickly), but then use the result of that as a reference to what the "reality" should be in order to set up their lighting in another engine (vray/mr).

That proves once again that maxwell provides a more believable result.

CHRiTTeR
05-15-2009, 03:18 PM
Personally... I would rather spend my time creating an image and thinking "I want to put a blue light over there" and simply put a blue light "over there" and be done with it, rather than spending time fighting with the software/render engine to make that blue light behave like a blue light.

In that instance... it seems that Maxwell wins.

i really dont understand that. Because in vray or mental ray or whatever renderder creating a blue light is just as easy.

You always have to do it in 2 steps.
make/model a lightsource and choose a blue color. done.


For me it seems more difficult in maxwell because you have to always use geometry as a lightsource (which you have to model) and you need to open the material editor to apply color and tell it to be a light.
While with other lights all controls for the positioning and properties of the light are kept in the same window for a faster workflow.

CHRiTTeR
05-15-2009, 03:22 PM
P.S... I have heard of people who will do a very quick down and dirty rendering of a scene in Maxwell (with lots of noise and such.. low settings, whatever it takes to render it quickly), but then use the result of that as a reference to what the "reality" should be in order to set up their lighting in another engine (vray/mr).

That proves once again that maxwell provides a more believable result.

No proof. Just more misinformed users.

A verry quick and noisy render in maxwell is absolutely not accurate. Its noisy for a reason.
Just take a few screen grabs every few minutes and observe how dark areas suddenly get bright and how some bright areas suddenly get more dark.
Especially dielectric materials, reflections and caustics.

Not a realiable technique at all.


You can perfectly input watts and limunous power (and other units) into vray lights also.
The only thing i miss there is the ability to input colortemperature in kelvin instead of rgb values, but i used to have a little colorchart for that anyway so its not that big of a problem, but slightly annoying some times yes.

Magnus3D
05-15-2009, 03:28 PM
It looks like you're working fulltime here in the thread to prove that Maxwell does indeed suck and that no one needs it. Who's paying you to do all this Chritter ?

/ Magnus

mister3d
05-15-2009, 03:30 PM
It looks like you're working fulltime here in the thread to prove that Maxwell does indeed suck and that no one needs it. Who's paying you to do all this Chritter ?

/ Magnus

Lol, no, we just want to exhaust nexlimit to give us a free copy of maxwell to test. :D

Bubbaloo
05-15-2009, 03:35 PM
For the Maxwell users... Can Maxwell distrubute the rendering of a single image accross multiple render nodes? Or does a single image need to be rendered on a single node?

Absolutely it can. Maxwell scales very linearly with the number of processors you throw into it. 2 of the same spec computers will clear up a render almost twice as fast as a single. Distributed bucket renders from mental ray just aren't nearly that efficient. Plus, you can stop and resume any rendering, which comes in very handy sometimes.

CHRiTTeR
05-15-2009, 03:37 PM
It looks like you're working fulltime here in the thread to prove that Maxwell does indeed suck and that no one needs it. Who's paying you to do all this Chritter ?

/ Magnus

No one is paying me for this. But I do know enough ppl who payed enough money for maxwell with all it wonderfull promesses some time back and they all feel screwed and deceived now, except for one guy, but he admits that that is because he can use its rendertimes as an excuse to do less work. :)

Plus, you can stop and resume any rendering, which comes in very handy sometimes.

Yes, that is indeed a verry handy feature!

Magnus3D
05-15-2009, 03:41 PM
Alright, but that does not mean you cannot find a equal amount of unhappy users of Vray and mental ray. You will most likely find more of them than you will find unhappy Maxwell users. I'm sure you will come up with some argument to try to prove me wrong on that point too but feel free to do so if it makes you happy.

/ Magnus

CHRiTTeR
05-15-2009, 03:49 PM
Clearly you have no idea about how chaos group treats its customers.
And compare that to how next limit treats them.

Btw, im only answering your question so dont accuse me or anyone else of being payed for this pls

dmeyer
05-15-2009, 03:53 PM
We switched to Maxwell because overall we spend less artist time on renderings. The computer spend more time actually rendering, but with the significant savings in setup / shader tweaking time we are able to move on to other projects as the farm works.

Magnus3D
05-15-2009, 03:54 PM
Oh well, and i can only say the same about you as you have no clue how Next Limit treats us. Don't try to talk about stuff you have no idea about Critter.

I think it's time a admin closed this thread now, it has been going offcourse for way too long. The thread stopped serving it's purpose long long time ago when these guys totally trashed it.

/ Magnus

mister3d
05-15-2009, 03:56 PM
I think it's time a admin closed this thread now, it has been going offcourse for way too long. The thread stopped serving it's purpose long long time ago when these guys totally trashed it.

/ Magnus

Why is it offtopic? We were discussing the noise behaviour of nextlimit examples and some parralel matters about it. Why are you reacting so much? Our discussion follows in a constructive manner, so I really don't get you.

CHRiTTeR
05-15-2009, 03:58 PM
And how would you know that? ;)


I think the topic is slightly sensitive yes, but not off topic.

Ppl have doubt about the shown examples. Thats not off topic.
If you dont like the answers given then dont ask the question.

mhampton
05-15-2009, 04:01 PM
i really dont understand that. Because in vray or mental ray or whatever renderder creating a blue light is just as easy.

You always have to do it in 2 steps.
make/model a lightsource and choose a blue color. done.


Really? It automatically behaves at the right color temperature and intensity? It automatically casts realistic soft shadows? It automatically generates caustics? Can it be polarized? Would a white light automatically break into the full spectrum if sent through a prism? Does it reflect dust in the atmosphere? Does it automatically create volumetrics? Is it affected by gravity? If the light is being animated.. does it slow down as it passes through different mediums? Does it automatically create lens flairs if directed at the camera? And so on and so on.... all the things that real light does in mother nature in abundance.

I'm not saying Maxwell does even a fraction of those things. All I'm saying is that I would rather be able to put a blue light over there.. because artistically I want a blue light over there, and have it behave like a blue light.. than to have to continously "fight" with 3D graphics to get it to behave like a real light.

I like to quote 3D world regarding Craig Zerouni from Digital Domain: "[I've] been coaxing images from computers for over 25 years". Because that's exactly what we do.. we coax images from computers. And anything that let's us do less "coaxing" and more creating... the better.

So I totally disagree that putting a blue light over there.. is a two step process.

Sure.. most of those things can be faked... but I would rather not have to spend the time to fake them. And sure.. it's great to have absolute control of all those things, so that if artistically I don't want the light to cast shadows, I can turn the shadows off. But I would rather have the light start out behaving like a real world light (because this is 90% of the time what I want it to do) and dial it down. Rather than starting with a light that isn't anything like a real world light and have to fake my way up.

Magnus3D
05-15-2009, 04:04 PM
If the same type of witchhunt was started in a thread about Vray or mental ray where we Maxwell users tried to prove it sucks then that thread would be closed fast and we would get a warning for our misbehaviour. It has not happend here, it should be fair.

And it's a long time since the thread was informative in any way, now it's just plain bullshitting and a pissing contest where you guys wanna prove that your engines rule and everything else suck. Too bad you fail to see that.

/ Magnus

mister3d
05-15-2009, 04:11 PM
And it's a long time since the thread was informative in any way, now it's just plain bullshitting and a pissing contest where you guys wanna prove that your engines rule and everything else suck. Too bad you fail to see that.

/ Magnus

This is clearly not true. Nobody said "maxwell sucks", we just argumented about weak and strong points of maxwell and others.

JorgeIvanovich
05-15-2009, 04:33 PM
Like i said funny thread,every user are free to use any engine so discuss wich itīs better of 3 itīs pointless.

I donīt think Maxwell itīs production friendly and there itīs no discusion about that if you live doing this,clients demands a lot of desing changes,you simply canīt wait 5 hs (to be kind with maxwell) for a render.

I donīt see any hubbercool feature in Maxwell,i see some people that use maxwell because it has almost no setup and a big library enought to just drag things and push buttons.

In this business speed and quality itīs the winner,Maxwell can deliver just one,quality.Vray and mental (in capable hands) can do the same and more,faster.

Bubbaloo
05-15-2009, 04:38 PM
I donīt think Maxwell itīs production friendly and there itīs no discusion about that if you live doing this

Maxwell was used in "The Curious Case of Benjamin Button" to create a train station. :cool:

CHRiTTeR
05-15-2009, 04:48 PM
Really?

Yes

It automatically behaves at the right color temperature and intensity?

Yes
Just like in maxwell you need to choose this first offcourse. The computer cant know this.
Although you cant input colortemperature directly (yet), like i already mentioned, but you can use a chart to input corresponding rgb values. Its not like thats much extra work or something, just a little annoying some times.

And you can load ies files (which also automatcly loads the correct color) which gives automatic correct lighting of an existing light.
Can maxwell do this? :)

It automatically casts realistic soft shadows?
Yes

It automatically generates caustics?

Yes, although you need to turn the option on, also not too much work.
But tweaking the caustics can take some extra time, but not too much either.

Can it be polarized?

Dont know enough about polarisation to answer that :)
But probably yes

Would a white light automatically break into the full spectrum if sent through a prism?

If you got the shader to do it, yes. Next version will have one, if im not mistaken.
Brazil already got this for a long time, even before maxwell.
In the visible spectrum at least. No colors you cant see or display like gamma rays, infrared or ultaviolet.

Does it reflect dust in the atmosphere?

Yes, since SP3 you have volumetric enviroment fog which you could use to do dust effects.

Does maxwell have this? :)

Does it automatically create volumetrics?

Yes, if you tell something to be volumetric it will be (like maxwell)

Is it affected by gravity?

Can you explain this pls?

If the light is being animated.. does it slow down as it passes through different mediums?

If you mean if light has actually got speed, like in real life? Then no. Does maxwell do this?
I cant see where this is needed.
If you mean something else, then pls explain.

Does it automatically create lens flairs if directed at the camera?

Not yet, but maxwell's flares are nothing more then a post effect.
So suddenly here the fact that its 'faked' isnt imprtant for you? :)

I'm not saying Maxwell does even a fraction of those things. All I'm saying is that I would rather be able to put a blue light over there.. because artistically I want a blue light over there, and have it behave like a blue light.. than to have to continously "fight" with 3D graphics to get it to behave like a real light.

Well it doesnt mean it cant do it now, that it cant do it in the future, just by adding it.
Thats not the point of the whole discussion!!




I like to quote 3D world regarding Craig Zerouni from Digital Domain: "[I've] been coaxing images from computers for over 25 years". Because that's exactly what we do.. we coax images from computers. And anything that let's us do less "coaxing" and more creating... the better.

So I totally disagree that putting a blue light over there.. is a two step process.

well its 3 words: placing, blue and light.
Light = create a light
placing= position it in the scene
blue = make it a blue light

Sure.. most of those things can be faked... but I would rather not have to spend the time to fake them. And sure.. it's great to have absolute control of all those things, so that if artistically I don't want the light to cast shadows, I can turn the shadows off. But I would rather have the light start out behaving like a real world light (because this is 90% of the time what I want it to do) and dial it down. Rather than starting with a light that isn't anything like a real world light and have to fake my way up.

What do you care if its faked or not, as long the results are correct.
If it can be done much faster by using 'shortcuts' but not less correct then why not do it?
I only care about getting the job done fast enough, and yes its nice if it were actually easier, but its not. This isnt 1990 anymore.

I mean its not like you have to code the light yourself to have inverse square, no thats done automaticly for you.
How it is done? Thats a problem for the devs to worry about which way they choose to calculate things.

You know, maxwell is also faking. Those are no real lightrays flying around in your computer screen, its all maths describing those things.

Vray doesnt fake any more then maxwell does.


It really is the maxwell users here getting more and more off topic, not the other way around.

Bubbaloo
05-15-2009, 04:51 PM
It shouldn't be labeled as faking, but interpolation.

CHRiTTeR
05-15-2009, 04:53 PM
If the same type of witchhunt was started in a thread about Vray or mental ray where we Maxwell users tried to prove it sucks then that thread would be closed fast and we would get a warning for our misbehaviour. It has not happend here, it should be fair.

And it's a long time since the thread was informative in any way, now it's just plain bullshitting and a pissing contest where you guys wanna prove that your engines rule and everything else suck. Too bad you fail to see that.

/ Magnus

We are only questioning the differences shown between V1.7 and V2.0 then maxwell ppl started to get offended by this and they are the ones who started to go more and more off topic. But in all those reactions defending maxwell i havent yet heared anything that explains the difference? Still waiting though...

Magnus3D
05-15-2009, 04:57 PM
It doesn't matter what we say here, you and the other guys will still do your best to prove that we're either lying or that we're wrong. What'the point of discussing if one side thinks it knows everything about everything. It's pointless, just like this whole thread turned out to be.

/ Magnus

CHRiTTeR
05-15-2009, 05:03 PM
It doesn't matter what we say here, you and the other guys will still do your best to prove that we're either lying or that we're wrong. What'the point of discussing if one side thinks it knows everything about everything. It's pointless, just like this whole thread turned out to be.

/ Magnus

well what i agree on is that we do our best to prove what we say and so far only mister3D did the same efford in defence of maxwell.
You are the one who is not adding anything usefull to this thread really.

mister3d
05-15-2009, 05:12 PM
but you can use a chart to input corresponding rgb values. Its not like thats much extra work or something, just a little annoying some times.

And you can load ies files (which also automatcly loads the correct color) which gives automatic correct lighting of an existing light.
Can maxwell do this? :)


AFAIK, the charts usually have the relative to color temperature values, and were created when there was no white balance in 3d. So how will you enter them correclty? Those numbers are treated already with a white balance in them. So using it after is not correct.

Do IES files load color information? I thought they load only light intensity among the caustic light distribution.

CHRiTTeR
05-15-2009, 05:32 PM
AFAIK, the charts usually have the relative to color temperature values, and were created when there was no white balance in 3d. So how will you enter them correclty? Those numbers are treated already with a white balance in them. So using it after is not correct.

Do IES files load color information? I thought they load only light intensity among the caustic light distribution.

Can you give more info about this?

I would reason if the resulting RGB color is the same as the colortemperature it would give the same lightcolor?

U are right, no color information in IES files.
I had to check in IES viewer to be sure and no ies file i checked so far seems to include colorinformation, bummer. Thats really weird :(
I probably had that idea because my ies files contain the colortemperature in their name.
But the vray ies light, does allow input of colortemperature (so it accepts rgb values but also kelvin values) input though.

mister3d
05-15-2009, 05:50 PM
Can you give more info about this?

I would reason if the resulting RGB color is the same as the colortemperature it would give the same lightcolor?

U are right, no color information in IES files.
I had to check in IES viewer to be sure and no ies file i checked so far seems to include colorinformation, bummer. Thats really weird :(
I probably had that idea because my ies files contain the colortemperature in their name.
But the vray ies light, does allow input of colortemperature (so it accepts rgb values but also kelvin values) input though.

Such tables usually list 2 color temperature columns: daylight and indoor. So either you use those numbers with a white balance off, or if you use white balance in your virtual camera you can't use those numbers, as they suppose you won't change the whitebalance and it's white. They say "if your camera would have a daylight balance, then such colors would looks like this". But as long as your camera uses it, those are not correct anymore.
If they have colortemperature in their name, then you can digit it in perhaps.

mhampton
05-15-2009, 05:56 PM
Chritter...

Every one of your yes's.. is followed by a "but"... and although you stated it just once, every one of those "buts", doesn't necissarily follow just turing something on. It's turning it on, setting a lot of settings, render, tweak, render, tweak, render tweak...

That's not a real world light. And yes... maybe "interpolating" is a better term than "faking".. and I don't care if it's "interpolated" or not if the results look good. It's just that right now, there's a lot more tweaking, render, tweaking, render, tweaking render than I would like. And any tool that allows me to skip 20+ steps of tweaking and experimenting to get the results I want.. weither we are talking about lighting, rendering, materials, modeling, motion graphics, fx, compositing, or anything else under the sun.. the better. Why are you arguing against that?

I'm not trumpeting Maxwell... I thought from day one that it's ahead of it's time. The most amazing render engine in the world doesn't do me (or anyone else) any good if it were to take a month to render one image.

And there is certainly a fine line where there is a point of dimishing returns... is a 2 hour render worth a 10% improvement in quality over a 10 minute render?

I know there are a lot of poser haters in the 3D world, and I can model a character from the ground up (although not my strong suit because I don't do it on a daily basis), but if starting with a poser base model (or any stock model that I could buy or own), could shave a day or two off of modeling the character.. hell yea I'm going to use it. Just because I can do something one way, doesn't mean I should.

Everything is just a tool... use the best tools you can to achive the desired results... or ignore certain tools at your on peril.

I've never touched blender... but you better believe I have my eye on it. Same thing with Maxwell... never used it... but I've always had my eye on it.

CHRiTTeR
05-15-2009, 06:01 PM
Ah ok, but if i'm going for a whitebalance @ 6500k and use that table (thus the table is calibrated at 6500k) to get rgb values, is this correct? or not?


But the ies lightsource has a field where you can input the colortemp anyway. :)

mister3d
05-15-2009, 06:01 PM
The people who I know personally who prefer maxwell cannot use another renderer like MR,vray, or Brazil to save their lives.
They are so used to the renderer doing everything for them they have absolutly no concept of how lights work in the real world and how that can transfer among all renderers.
If it isn't a preset light emitter forget it.

Well those must be really strange people. Actually I don't see it as a big problem, as there are just a handful of rules and they work by default in vray and mental in area lights. This just shows that maxwell made them really lazy.

CHRiTTeR
05-15-2009, 06:17 PM
Chritter...

Every one of your yes's.. is followed by a "but"... and although you stated it just once, every one of those "buts", doesn't necissarily follow just turing something on. It's turning it on, setting a lot of settings, render, tweak, render, tweak, render tweak...

No its not always followed by a 'but'. I think thats pretty easy to check for yourself.
If you need to click the on button, i think it was pretty important to mention.
I like to be fair and honnest. Maybe thats something new to you.
also because otherwise you'd undoubtfully used that as an argument that im not explaining and comparing things in a fair way.

And no the tweaking is not as intensive as you are trying to make it sound, not by a long shot.

That's not a real world light. And yes... maybe "interpolating" is a better term than "faking".. and I don't care if it's "interpolated" or not if the results look good. It's just that right now, there's a lot more tweaking, render, tweaking, render, tweaking render than I would like.

Faking and interpolating is not the same thing.

No theres not.

So you say its ok to do a 10hr render because the tweaking takes maybe 1hr on another renderer who can render it in 4hrs?
And even here im being verry generous with the numbers.


And any tool that allows me to skip 20+ steps of tweaking and experimenting to get the results I want.. weither we are talking about lighting, rendering, materials, modeling, motion graphics, fx, compositing, or anything else under the sun.. the better. Why are you arguing against that?

Im not arguing against that, im just saying it doesnt take so much more steps to get comparable results. And its not much more time consuming to get the values right.


I'm not trumpeting Maxwell... I thought from day one that it's ahead of it's time. The most amazing render engine in the world doesn't do me (or anyone else) any good if it were to take a month to render one image.

And there is certainly a fine line where there is a point of dimishing returns... is a 2 hour render worth a 10% improvement in quality over a 10 minute render?

10% is a lot. The difference will be much less then that and compared to a photo is will probably be the same difference.
and you are the one who needs to exagerate here in the hopes to sound convincing. Im just being honnest about everything

I know there are a lot of poser haters in the 3D world, and I can model a character from the ground up (although not my strong suit because I don't do it on a daily basis), but if starting with a poser base model (or any stock model that I could buy or own), could shave a day or two off of modeling the character.. hell yea I'm going to use it. Just because I can do something one way, doesn't mean I should.

I dont like poser either, but the reason why ppl use it is because its actually faster then modeling your own model. So it does save time, this is not the case here.

CHRiTTeR
05-15-2009, 06:18 PM
Well those must be really strange people. Actually I don't see it as a big problem, as there are just a handful of rules and they work by default in vray and mental in area lights. This just shows that maxwell made them really lazy.

Its good for us, though as we can show clients we can do things faster :twisted: :deal:

lazzhar
05-15-2009, 06:37 PM
This is the thread that never ends http://forums.cgsociety.org/images/icons/icon10.gif

mhampton
05-15-2009, 06:57 PM
No its not always followed by a 'but'. I think thats pretty easy to check for yourself.
If you need to click the on button, i think it was pretty important to mention.
I like to be fair and honnest. Maybe thats something new to you.
also because otherwise you'd undoubtfully used that as an argument that im not explaining and comparing things in a fair way.

And no the tweaking is not as intensive as you are trying to make it sound, not by a long shot.

Well... I've got a physical stage light with a blue filter on it right here in my hands... i'm going to go over her and plug it in and turn it on. Apparently in a couple of minutes you can perfectly match what that light is doing (color, intensity, caustics, volumetrics, interference, dust, aborations, etc. etc.) in whatever 3D software you are using... good for you.

mister3d
05-15-2009, 06:58 PM
Its good for us, though as we can show clients we can do things faster :twisted: :deal:

I really can't get this talk "unbiased vs biased", because it's not 2000 year anymore when you had to compose awkward shader trees without many declared functions. Today, when render engines automatically follow energy conservation rules and you have photometric lights chosen as the first ones in max, restricted to inverse-square falloff, you must be really stubborn to make something wrong. Yet those fancy presets for most materials. :argh:

Ah ok, but if i'm going for a whitebalance @ 6500k and use that table (thus the table is calibrated at 6500k) to get rgb values, is this correct? or not?

Not! You see, this table is just a simulation of whitebalance, that's why you have 2 columns of daylight and indoor light. For you to use it with your camera whitabalance you must have only one number of it's tepmerature, right? But you have two, that's why you can't use it with your camera if its whitebalance set differently from white. But if it's set on white, it doesn't make sense. So either you use your camera whitebalance, or use the table numbers without using your camera whitebalance.

CHRiTTeR
05-15-2009, 07:08 PM
Not! You see, this table is just a simulation of whitebalance, that's why you have 2 columns of daylight and indoor light. For you to use it with your camera whitabalance you must have only one number of it's tepmerature, right? But you have two, that's why you can't use it with your camera if its whitebalance set differently from white. But if it's set on white, it doesn't make sense. So either you use your camera whitebalance, or use the table numbers without using your camera whitebalance.

Ok, so using a camera set to have a white whitebalance (255, 255, 255).
Then using the charts 6500k column to input corresponding rgb values in the lights.
Will result in a render as if your camera has a white balance set at 6500k?

This is how i did it in a few cases where i couldnt use max's photometric lights.


Well... I've got a physical stage light with a blue filter on it right here in my hands... i'm going to go over her and plug it in and turn it on. Apparently in a couple of minutes you can perfectly match what that light is doing (color, intensity, caustics, volumetrics, interference, dust, aborations, etc. etc.) in whatever 3D software you are using... good for you.

not with whatever 3D software and the materials need to be setup and everything needs to be modeled. Off course this takes more then a couple of minutes, but it wont take much longer to setup than with maxwell. Especially not if you're going for an 'unbiased' render.
Because most of the time consuming tweaking and testing you are talking about is to get the best time/quality ratio. If you already know you want the best quality possible its not that hard to crank things up.

You are not getting the point at all and keep getting off topic.


This was about the maxwell 1.7 renders looking different then the 2.0 renders.
Even though both are claimed to be ultra accurate and realistic?
Again... how is that possible if both use physicall correct an ultra accurate physicall spectral calculations?

An imprtont 'detail' mentioned while having this argument was that the v2.0 render looks verry interpolated and biased.

JorgeIvanovich
05-15-2009, 07:21 PM
Maxwell was used in "The Curious Case of Benjamin Button" to create a train station. :cool:

One case itīs not production friendly...Jesus

This thread is hilarious

CHRiTTeR
05-15-2009, 08:27 PM
ok i think this will probably get things back on topic.



I think i have found a good condidate to why the diffused lighting/shadows and less blue sky reflections in the interior scenes.... and guess what! its in favour towards maxwell, im sure you didnt see that one comming.

What if they have clouds in their skylight system. Which as in real life makes everything less saturated (less blue skylight) and scatters sunlight around which could result in the more difused shadows and overall gi (thus less contrasty).

Although it doesnt explain why the scene with the wooden blocks, car and hallway (which use HDRI lighting) are have differences in brightness in some areas, but those could verry well be sampling issues (more time needed to get same results) as they are less pronounced differences.




*edit: Check the reflections on the tap. Do i notice some clouds in there?! ;)
Also notice how the 2 scenes who use skylight dont have extra info next to them... :)

*edit2: oops, the cloud-like reflections are actually in the v1.7 version. lollll :blush:
But also notice how the background image is brighter in the V2.0 version of the big hall image. New tonemapper? With loadable presets for existing camera models? (wild guess)

oh well... cant blame me for not trying... Time for bed.

Telemachus
05-15-2009, 10:49 PM
I can't believe I read through every page of this thread. Some of things that have been said here are really disappointing.

There's an old saying; "A bad workman always blames his tools". Whatever tool you are using, you can achieve a perfectly realistic result, if you are skilled in its use. These debates about other renderers not being "accurate" or "mathematically realistic" are moot, unless you are Commander Data from Star Trek and have incredibly good eyesight, you won't be able to tell the difference between a biased and an unbiased render if set up by a skilled user.

An unbiased renderer is easier to use because there is less user input required. This is great especially for non-technical people who don't yet understand the finer points of computer lighting. However, please don't speak of this difference as though it is a burden to those of us who use biased renderers. You only need to learn to use a biased renderer once, not every time you use it. Biased renderers from other manufacturers operate on the same basic principles and are therefore not hard to adapt to, once you have the grasp of one.

Someone in this thread (I forget who) was asking if a biased renderer enabled all sorts of features automatically. The effects listed would require me to click a check box for the most part. Since I don't find clicking my mouse to be a very cerebrally taxing operation, I can live with that particular "disadvantage" of using a biased renderer.

Perhaps I am old-fashioned but I believe it is better to work smarter, not harder. When i'm painting a room, I use a wide roller for the big flat simple areas and a narrow 1-inch brush for the detailed areas around the windows and door frames. It saves time and attention is only given to the areas that need it. I am also less tired and less annoyed after finishing the job. I think this is a pretty good metaphor for why I personally prefer using biased renderers. I like saving time, reducing power-usage (both of the CPU and air-conditioning to cool hot rooms housing computers) and carbon emissions. I like being able to preview a render in seconds using older hardware. I like the smooth result I can get from my renderings by default (dare I say automatically?), which I would never consider a major feature, and hardly deserving of a whole point release for a major piece of software.

It's good that Maxwell Render is improving in terms of speed and output. I'm sure those that use it will appreciate the enhancements. For me, it isn't suited for stills or animation. People who say that throwing more computing power at it will make it more acceptable are missing the point. I could get a hundred people with tiny paint-brushes to paint my room but it is just a waste of resources.

Artistic freedom was also mentioned. Someone said that they would rather devote their time to placing a light than getting involved in the technical details of tweaking parameters to get it to look "right". I think this aspect would be deserving of an entire thread by itself, but i'll just touch on my thoughts about it here. First of all, I don't think there is anything wrong with learning technical terms and adjusting parameters in a field which was itself made most popular by hardcore mathematicians. They provided us with the algorithms and shaders that we so casually and easily manipulate today. There's no reason why we shouldn't try to increase our knowledge of the subject just because we can take a short-cut.

If all you do is set a resolution and click render, drag and drop lights with minimal effort or understanding of the physical under-pinnings, then how are you, in your industry position, any more skilled than a child who could do the exact same thing at home? Oh wait, you're an artist. But art being highly subjective, every man and his dog can be an artist. There has to be some substance and skill involved in it too. Another thing to consider is that once you remove the technical complexity from an occupation, via deliberate design or the use of brute force computation or a combination of the two, you open your chosen field to greater competition, since your occupation is essentially easier to do by someone of lesser skills and experience. A perfect example of this is how IT jobs have gradually been migrating to what are for all intents and purposes third-world countries. Programming is now as simple as dragging and dropping modules together (obviously not all coding is like this but it seems like a growing trend). Of course, simple methods don't necessarily make a brilliant piece of software, but it's close enough for companies wanting to save money by hiring people who have been rapidly trained and who lack the experience of someone with more in-depth knowledge.

Now I know I may have offended some of you with that last paragraph, believe me that it wasn't intentional. It's just that I have seen an awful lot of individuals over the years starting out in 3D who are as obsessed as some of the people in this thread about the technology they use. As if there is some kind of magical software that will make you a better artist, allow you to gain influence and have bikini-clad women hanging off your arm. These people fritter away hours, days, months, collecting every specialist plugin and renderer they can find because someone told them it has some kind of killer feature that'll make all their renderings look photorealistic. I have heard so many things. "V-Ray is the best! It's so easy to use!"... "Maxwell is a NON-BIASED renderer! It's new so it MUST be better!".

It's all basically crap. Ultimately it's the skill of the user that makes or breaks a good piece of artwork. The old masters used their eyes and their brains to learn and understand how lighting worked, enabling them to paint incredibly realistic and beautiful images. They didn't defer that responsibility to some scientist, they saw it as an integral part of their artform. Are those of us who use biased renderers somehow deficient, because we actually bother to learn how to use our tools, and feel like we can rely on our eyes and brains to decide whether or not something looks "right" and not on a "lighting simulator"?

Maxwell, for me, appears to be a "bridge" application, enabling people with knowledge in different fields to easily render realistic 3D scenes. It's not a bad thing for people like that, but it has many deficiencies and the few positives it possesses can be easily countered by a skilled user using traditional renderers. Those wonky renderings of a grey sphere that had stepping artifacts on both versions weren't exactly convincing, although admittedly it's an over-simplistic scene with which to make a judgement.

Sorry for the long post but I got pretty wound up after 16 pages of this stuff :P

vlad
05-16-2009, 02:32 AM
Hey, something to ponder : when hardware will be fast enough for unbiased engines to achieve decent render times, biased engines will render in ... REAL TIME!
:D
OK guys, carry on...

CHRiTTeR
05-16-2009, 10:53 AM
Hey Telemachus, im not going to quote that one, haha, but heres what came in my mind while reading your post anyway.

Many of the non-spectral rendering engines can render unbiased just fine. But they also have the possibility to render using biased solutions/techniques and use shotcuts who result in loss of detail (but are intelligent/smart enough to do it there where its less vissible).

Most ppl using a path tracer dont even let it render until the result is clear of noise. There's always some little amount of noise left. Also note, that its these same users who keep saying they need/require an unbiased render because they desire perfect results.
So they say its important and they really really really need, but they dont really let the render get to that point. Ok yes, there are probably some exception (and now everyone suddenly is going to say he's one of those exceptions).

Most of the 'other renderers' (especially the popular ones) also have support for input based on physicall values (colortemperature, watts, ior, shutterspeed, iso...). This was not so much the case a some years ago, yes. And its when maxwell came out most other renderers where asked to implement this because it is a verry handy possibility, yes.

And as far as i know, theres an equal amount of 'guessing' is needed in maxwell (colors, roughness, etc...).

The biggest difference is just the user interface really. Maxwell choose to 'hide' the ability to tweak parameters (and thus can do less).
While others offer these, not because their engine cant work without these controls, no its because clients asked for them many times to have more control over some result and the time/quality ratio. They also started out with less controll and this stuff got mostly added by request.
Im verry verry verry sure, every one who used maxwell for a reasonable amount of time has wished he had more control over something so it was easier to get desired (maybe unrealistic, but not nessecarely) result. Because they need to do everything like a photographer would to adjust the lighting (and have to model it etc). Isnt it a lot easier if you just had to adjust 1 slider somewhere?

Also you mention artistic choices, but there's not much artistic choices in a render which offer as good as no options to alter the results and will always give the same results. You just have to rely on its baked-in settings.
You can argue its not les artistic from what a photographer does, but it is a lot different.
In real photography you have choice in different lenses, films, cameras etc... who all influence the image's feel and look.

Its perfectly possible (and i hope they will do it) for the 'other' renderers to include an user friendly interface with amazing detail set by default (and an included 'unbiased' preset). Why this isnt the case now is already said.

It is fun to watch the render progress though (until it reaches a certain sampling level that makes updates go to slow).

I just thought id add this.
And i think you said it verry well.

mister3d
05-16-2009, 05:40 PM
Aren't you all tired of putting this argument "it's the artist what matters"? Then go and use scanline, if it's all down to the artist. If those argumenta were addressed to me, then I spent some time learning traditional lighting and photography, and I know their role in cg. Please, this thread is not about what makes good images, it's about what differs maxwell from other renderers.
Until I see a fair comparison, I won't believe that maxwell produces the same results but much slower.

Magnus3D
05-16-2009, 05:46 PM
No, it's not about what differs Maxwell from other renderengines. This thread is about the upcoming Maxwell Render 2.0 release!

/ Magnus

mister3d
05-16-2009, 06:10 PM
No, it's not about what differs Maxwell from other renderengines. This thread is about the upcoming Maxwell Render 2.0 release!

/ Magnus

Yes, you are right, my bad. :)
But putting it into a flame war or what makes a good picture...eh. :rolleyes:
I just wanted to know whether there's something peccular in maxwell that makes its images so convincing (ok, not all, but still).

CHRiTTeR
05-16-2009, 06:19 PM
Aren't you all tired of putting this argument "it's the artist what matters"? Then go and use scanline

Scanline is ok, I still use it from time to time for things like particle passes etc...
Its faster in some cases but also slower when u use/need things like raytracing (gi, glossy reflections, ...).


No, it's not about what differs Maxwell from other renderengines. This thread is about the upcoming Maxwell Render 2.0 release!

/ Magnus

Any new possible suggestions on what makes the renders look so different?

Magnus3D
05-16-2009, 06:33 PM
No, because it has already been said several times here by other Maxwell users. The images posted by Next Limit in the first reveal were ONLY to showcase the speed of 2.0, nothing else. And after that you guys went nuts and speculated wildly about them.

/ Magnus

CHRiTTeR
05-16-2009, 07:01 PM
True, less noise (and more speed) is a good thing. No one is disputing that.

Its just that, again,
maxwell 1.7 is physicall correct, unbiased and ultra precise.
maxwell 2.0 is physicall correct, unbiased and ultra precise.

However, when you go check the images they clearly show different results in the same scene.
How can that be if both are physicall correct, unbiased and ultra precise?

I already suggested clouds.

Now instead of turning the discussion towards that, its you guys who keep on getting off topic and feel insulted because we understand why this verry strange and suspicious.

We tried to explain you why, but you dont like the answer and keep on brabling that we have to shut up.
From the moment some comments are not like you would like them to be, you take it personally that we point out theres something strange about the software you use and you demand the thread to be closed.

So again, any good suggestions? If you think its normal, then pls do say why.

Im off to the movies, i'll check your answer later.

Magnus3D
05-16-2009, 07:26 PM
You are not worthy of a reply, but you will recieve one anyways..

You are still blaming us for derailing this thread, when YOU and some other guys are the one's to blame.

Why the images show different results have also been explained in this thread, if you actually got off your high horse and read our posts instead of ignoring them then you would have known that by now. I guess you will dismiss the explanation and make one up yourself that fits your thoughts instead of the fact from the source.

We have no been insulted or taken comments personally, the only main problem has been this continued witchhunt to pinpoint that Maxwell sucks and that you don't need it and that you can do everything with Vray or mental ray. It seems that no one cares except you yourself.

This thread should have been closed when it hit page two, but the moderators are too slow to react or they simply do not care when the discussion is about how bad Maxwell is.

/ Magnus

mister3d
05-16-2009, 07:55 PM
Scanline is ok, I still use it from time to time for things like particle passes etc...
Its faster in some cases but also slower when u use/need things like raytracing (gi, glossy reflections, ...).


No, it's not just slower, but you can't make many things in it, and glossy reflections\refractions is just one of them.
Underestimating tools often happens here, what may confuse many people, especially those who just start in the cg world.
So when I head "it's not about tools, it's about artists" I immediately think about the artist who uses a modern advanced renderer saying so.
Maxwell certainly has something which neither mental nor vray have.

JorgeIvanovich
05-16-2009, 08:01 PM
This thread should have been closed when it hit page two, but the moderators are too slow to react or they simply do not care when the discussion is about how bad Maxwell is.

/ Magnus

Hi think they are laughing like hell.

You canīt close a thread just because some people have different opinions.

You can simply ignore comments or users and live happy dropping things to Maxwell.