Maxwell Render 2 out

[cgbeige]

So now people can finally stop saying Maxwell is slow. 3 minutes for a glossy refraction render that is almost ready to go:

http://www.3eige.com/cgtalk/mxspeed.jpg

Considering I just waited an hour for an AO pass from MR, I think it's safe to say this thing is fast.

[terencemace]

cgbeige, that is a great image! nice work. and just over 3 minutes----what are your system specs?

thanks

T

[cgbeige]

hehe - that's just the benchwell.com benchmark scene. My system is an 8-core/16-thread Nehalem Mac Pro 2.66 GHz with 24GB of RAM and OS X 10.6.1.

Using the old Maxwell benchmark can be misleading though since renders are much closer to final early on now in 2.0

I've got a few portfolio pieces on the backburner that I'm going to rerender though

[cgbeige]

I'm working on something to comp with a piece of sushi. 1 minute rendered caustic:

http://www.3eige.com/cgtalk/maxwell1minute.jpg

Scene setup time: 5 minutes. Default water and the basic sky/sun. I realize that the top is not right - that's getting cropped out later. It's just to show how fast this thing is when it matters. No photons vs light intensity, no guessing, no AO passes, no fakes, no waiting 5 minutes for prepasses and you can adjust lights while it renders.

It's not without its bugs though.

[cgbeige]

OK - this is maybe the last time I'll bump this thread but people have to see what a properly integrated 3D engine that corresponds to the rendering looks like. Check this out:

http://www.3eige.com/cgtalk/maxwellglrender.jpg

or in video:

http://vimeo.com/6758893

it's almost crazy

[SreckoM]

Hope that they will soon offer demo version, so until then, freely bump this thread whenever you need to :)

[cgbeige]

ya, they will - the download for paying users is just the demo and you copy your license over. It's just that their servers were buckling from all the upgrade downloads so it will be a few days for the demo

[SheepFactory]

Is maxwell usable for animations now?

[cgbeige]

well it's still very lacking for pass control and it doesn't do volumetrics (which is a problem for VRay too) but it's fast enough for archviz walkthroughs now. I don't know if it has flicker problems though since it doesn't do lightmap caching since everything is relative to the camera. I only do still stuff so I'm not the best judge of animation needs

[cgbeige]

5-blade bokeh, caustics, 600,000 polys, motion blur and all 32-bit channels - one minute and fifteen seconds:

http://www.3eige.com/cgtalk/maxwell2render.jpg

I've been using Mental Ray for this last project and forgot that Maxwell gets light that subtle with no work.

[Kel Solaar]

The work they did to increase the speed seems to be amazing :)

KS

[eikonoklastes]

So now people can finally stop saying Maxwell is slow. 3 minutes for a glossy refraction render that is almost ready to go:

http://www.3eige.com/cgtalk/mxspeed.jpg

Considering I just waited an hour for an AO pass from MR, I think it's safe to say this thing is fast.You waited an hour to get an AO from MR in this scene? You're either rendering at abnormally high resolutions (in which case you should probably mention that) or you're doing something wrong.

[inguatu]

still looks like noise is a major issue with Maxwell 2. Good thing Vray is finally out for Maya and of course there's always the ol standby of MR.

[inguatu]

You waited an hour to get an AO from MR in this scene? You're either rendering at abnormally high resolutions (in which case you should probably mention that) or you're doing something wrong.

Yes, meet cgbeige, who consistantly loves to troll and bash MR threads with all things Maxwell and Mac. There's a lot of great work being done with MR, unfortunately there's a few people who go out of their way to TRY and prove that wrong despite the fact they really don't know how to use MR properly. ;)

[cgbeige]

noise isn't a major issue - those renders are after a minute of use and in five more minutes, they would be noise free. If you spent the last two months putting up with the shit that MR has put me through, you'd abuse it too. The reason my AO render pass took so long is due to the renderer suddenly deciding to use one core of a total 16 for no reason, not personal incompetence.

here's a 5.5 minute render (ignore the lazy collision between the water and sushi - just look at the render):

http://www.3eige.com/cgtalk/maxwell5minute.jpg

if someone says that MR could produce that in five and a half minutes, then they are lying.

[cgbeige]

You waited an hour to get an AO from MR in this scene? You're either rendering at abnormally high resolutions (in which case you should probably mention that) or you're doing something wrong.

no - I waited an hour to get an AO pass from a different scene. I was saying that MR is considered "fast" when you have to jump through pass hoops to get final renders, which are much faster now in Maxwell. How long do you think it would take to get a caustic rendering like that motion blurred sushi in water from Mental Ray? It would take at least 4 hours.

I'm not a Maxwell fanboy - I bought VRay for Maya recently too so I'd never have to use MR again. If others here didn't work for companies, they'd likely be doing the same. The only people I see around here who actually like MR are the ones who write books about it so writing about bug updates pays their rent. My favourite error lately is this:

Error: Scene /Volumes/HOME_RAID/WORK/SPAFAX/enRoute/FIELD_TESTING_SPREADS/FIELD_TESTING3_city/scenes/w_wshoef1.mb failed to render.

That render succeeded actually. It seems MR can't even brag about when it gets it right.

[JohnPetrucci]

wow beautiful fast renders!

does it render fur?

[cgbeige]

no - it doesn't do vertex shaders. It's not really good for animation since it's a lot like VRay - it doesn't render particles, fur, etc. It's really more for archviz.

[Jozvex]

Just thought I'd add some points like Maxwell 2 also:

* Supports IES and Eulumdat light profiles
* Does 'thin SSS' for single poly objects like leaves or rice paper
* Can render thousands upon thousands of lights in one scene
* Multilight now lets you adjust the light colours as well as the intensity
* Is just way better than Maxwell 1.7

:thumbsup:

[eikonoklastes]

no - I waited an hour to get an AO pass from a different scene. I was saying that MR is considered "fast" when you have to jump through pass hoops to get final renders, which are much faster now in Maxwell. How long do you think it would take to get a caustic rendering like that motion blurred sushi in water from Mental Ray? It would take at least 4 hours.

I'm not a Maxwell fanboy - I bought VRay for Maya recently too so I'd never have to use MR again. If others here didn't work for companies, they'd likely be doing the same. The only people I see around here who actually like MR are the ones who write books about it so writing about bug updates pays their rent. My favourite error lately is this:

Error: Scene /Volumes/HOME_RAID/WORK/SPAFAX/enRoute/FIELD_TESTING_SPREADS/FIELD_TESTING3_city/scenes/w_wshoef1.mb failed to render.

That render succeeded actually. It seems MR can't even brag about when it gets it right.Well, the pass system in 2009 seems to have critical issues that slow down renders inexplicably. This is more an Autodesk issue than an MR one.

Motion blur and caustics are not something I'd consider using in production yet with MR unless the deadline is next year, so no one will argue with you there.

Looks like I might have to play with Maxwell 2 a little.

[chuckie7413]

yeah i am quite impressed with the renders and the times i am seeing from Dave. I pretty much a mental ray user myself but am open to use what ever software/hardware necessary it takes to get the best results and the job done.

There will always be advantages and disadvantages to renderers, it is just a matter of finding the right balance and what works best for you or your studio.

I will definately have a play when the maxwell v2 demo is released.

Cheers

Rich

[dmeyer]

Is maxwell usable for animations now?

We've been using 1.7 for animations for a couple of years now.

[cgbeige]

so it never had problems with flicker?

[dmeyer]

so it never had problems with flicker?

Not in my experience.

[bazuka]

can we see some animations?

[cgbeige]

well I don't really do animation since I only do print work but I can post a quickie of that sushi Realflow sim or throw together a giant instanced tree walkaround or something.

[Algoris]

As a long time maxwell user I liked this software a lot. First of all it was very easy and userfriendly compared to MR and later Vray (which I used for archviz). But the main thing why i chose maxwell was because it was physically correct. In 1.7 there was somehing suspicious about lighting quality but I didn't care. Now in M2 i must admit that Maxwell is not so unbiased and physically correct as it was in the beggining. I compared it with Luxrender and Fryrender. Used simple test scenes with lambert materials so the comparison could be as fair as possible. I found out that Maxwell's primary lighting bounces are biased to somewhere near 3-5 and the secondary bounces are biased to 2-3. i don't know what are other corner cuts but I'm sure M2 is not a true MLT renderer any more.

For comparison Luxrender was the main tool because it is opensource and if you look into source-code you can see for yourself that it is fully unbiased and don't use corner cutings for speedup. Fryrender is also much more correct now than M2 but I'm alsmost sure that Fryrender also use a biased number of primary and secondary light bounces but I'm not sure of their numbers. However they are higher than Maxwell.

Luxrender is also some 10x (or more) slower than maxwell but if maxwell would been fully unbiased it's speed would be close to Lux.

In archviz industry ordinary clients are satisfied with Vray-like renders but i have many clients who expect the most physically correct renders and I can not use Maxwell for that any more. I will rather spend more money on faster hardware and use a physically more correct renderer.

PS:. Sorry for my bad English.

[cgbeige]

interesting - thanks for the input. I tried luxrender once but it was too bare bones for me and ya, it's very slow. Can you point to some discussions or comparisons?

Anyway, I can't vouch for Maxwell's accuracy beyond saying that it is far better than VRay and Mental Ray.

[threedeworks]

As a long time maxwell user I liked this software a lot. First of all it was very easy and userfriendly compared to MR and later Vray (which I used for archviz). But the main thing why i chose maxwell was because it was physically correct. In 1.7 there was somehing suspicious about lighting quality but I didn't care. Now in M2 i must admit that Maxwell is not so unbiased and physically correct as it was in the beggining. I compared it with Luxrender and Fryrender. Used simple test scenes with lambert materials so the comparison could be as fair as possible. I found out that Maxwell's primary lighting bounces are biased to somewhere near 3-5 and the secondary bounces are biased to 2-3. i don't know what are other corner cuts but I'm sure M2 is not a true MLT renderer any more.

For comparison Luxrender was the main tool because it is opensource and if you look into source-code you can see for yourself that it is fully unbiased and don't use corner cutings for speedup. Fryrender is also much more correct now than M2 but I'm alsmost sure that Fryrender also use a biased number of primary and secondary light bounces but I'm not sure of their numbers. However they are higher than Maxwell.

Luxrender is also some 10x (or more) slower than maxwell but if maxwell would been fully unbiased it's speed would be close to Lux.

In archviz industry ordinary clients are satisfied with Vray-like renders but i have many clients who expect the most physically correct renders and I can not use Maxwell for that any more. I will rather spend more money on faster hardware and use a physically more correct renderer.

PS:. Sorry for my bad English.

i cannot confirm this at all, maxwell's render precision was improved, the shading model is now even nearer to the physical reality than in version 1, the rendered results should confirm this for everyone. can you please show us all the renderings on which you are basing your observations? can you post the scenes as well? what do you mean exactly by saying 'biased to 2-3' or 'biased to somewhere near 3-5'?

markus

[mverta]

I found out that Maxwell's primary lighting bounces are biased to somewhere near 3-5 and the secondary bounces are biased to 2-3. i don't know what are other corner cuts but I'm sure M2 is not a true MLT renderer any more.

This is in not in any way true, and I look forward to seeing your "proof." Good luck.


_Mike

[alexentremont]

Yes, meet cgbeige, who consistantly loves to troll...
You seem to know quite a few things about trolling too.

[inguatu]

You seem to know quite a few things about trolling too.

Thanks, regardless of the fact that conversation had moved away from that awhile ago, I am curious to hear the post about Maxwell being a non MLT renderer. I'd also be curious to see examples of Vray vs maxwell for speed, noise and aesthetics in a still as well as an animation. Maybe cgbeige could figure that out since I think he now has both.

Have an adequate day, Alexa!

[Algoris]

I don't want to start a flamewar in any mean. Simply because i used maxwell till present. I'm just trying to tell my observations which are of course my own opinion and nodody is forced to accept it. I'm 99% sure that the primary and secondary lighting bounces of maxwell are limited. I don't know the exact numbers simply because i can not look in maxwell's source code but the fact is that they are limited and this means that it is biased in some way. I think that they tried to compensate these bouncing limitations with more and more robust shading algorithms. Also it looks that Maxwell now uses some form of radiance catching to reduce the number of shooted rays because in some places the tones of shades on an object are way too smooth and on the sides of shade tones where the transition of light changes, these transitions are very sharp.

When i have more time, i will upload some comparison images just for those people who are only curious and don't take this as an offense to the tool they use. Also anyone can make a comparison if he/she compares maxwell with some opensource unbiased renderer. I chose Luxrender but i think there are other opensource unbiased renderers too. I think it is important to compare it with opensource renderers because this way you can look in it and assure yourself (if you have some experience in programing) that it uses fully unbiased algorithms.

And one more time, please people don't take this as an offense or flameing, bashing of any kind! It is just my personal opinion and everyone is free to have one. Also nobody is forced to accept someone else's opinion if he/she disagrees with it.

ps:. sorry for bad English

[Bubbaloo]

Used simple test scenes with lambert materials so the comparison could be as fair as possible.

Can you tell me where in nature you can find a 100% lambertian surface? It doesn't exist. To make any test for physically accurate lighting, shouldn't you start with a surface roughness that can be recreated in a photo comparison?

[cgbeige]

these transitions are very sharp

I remember Zack, the mental ray guy chiming in on a thread about how Lambertian falloff is actually a lot sharper than most shaders render it so the more physically accurate implementations look wrong because we're so accustomed to seeing it the other way. Could this be what you're seeing?

As for comparing VRay and Maxwell, I do have both but I don't have time to set up a scene in both renderers and try and get them to match. I'm only posting here between renders for a deadline. And anyway, I am new to VRay so it will take me a while to get accustomed to it – but I didn't buy it for its photo realism, I bought it because I wanted a good multipass/linear workflow renderer for more stylized renders to replace the Mental Ray. I've never been convinced that VRay could produce what Maxwell does as far as physical accuracy goes from what I've seen.

[Algoris]

I used lambert simply because it is easy to transfer between renderers. The shading algorithms are a totally different animal. But the main part of renderer IMHO is it's handling of light rays. A made comparison about light behavior, not abou shading algorithms.

[mverta]

You are not offering an opinion - you are stating what you believe to be a fact about the rendering technology. If you cannot demonstrate your theory with proof, then I'd suggest you not spread misinformation and misrepresentations about the product.

As it happens, you are absolutely incorrect about your assertion, so I am aware no such proof will be forthcoming. But perhaps if you post what you think is evidence of such, we can address what is leading you to draw false conclusions.


_Mike

[InfernalDarkness]

So would "unbiased" mean that a renderer has no limit (save time) on how many bounces the light mechanism traverses? Or would that mean that in the case of mental ray, there were no limit to the Final Gather bounces? I'm still not clear on this topic...

[Jozvex]

Unbiased is essentially saying that the full extent of the mathematics and such, no matter how long and arduous, will be performed to do all the different calculations. Whereas mental ray or whatever will get so far, then with the interest of speed in mind go "hmm that's pretty close" and round up a number (or whatever, depending on what it's calculating). With final gather or photon tracing, you figure out roughly how the light would go, then blend it all together to look as nice as possible in a reasonable amount of time, but that introduces splotches and flickering or aliasing, etc.

Maxwell is saying, given enough time I can figure out 100% exactly how the scene would look, given the algorithms it has etc.

I can only reinforce what Mike and others are saying, Maxwell is definitely not doing a small number of bounces, that would just be pointless! I mean you can easily render scenes that have many many levels of refractive/reflective caustics, so each of those is a light bounce.

Looking at a render with 3 bounces of GI versus one with many many bounces is glaringly different and noone could claim their 3 bounce GI is unlimited bounces.

[InfernalDarkness]

Fair enough! And thank you kindly.

Been waiting for Maxwell 2 for some time now, looks like it's ready to rock. I don't see it replacing MR on my deadlines just yet, but the quick material setup alone is a devastating feature; 50% of my working time in Maya is just building shaders these days.

[juancg]

Hi Algoris,

You might want to take a look at this post:

http://www.maxwellrender.com/forum/viewtopic.php?f=1&t=32629 (http://www.maxwellrender.com/forum/viewtopic.php?f=1&t=32629)

So we will save time discussing if Maxwell v2 is more physically correct or not.

I found out that Maxwell's primary lighting bounces are biased to somewhere near 3-5 and the secondary bounces are biased to 2-3.
You would need to review the basics of the theory behind the things that you are saying, if you want to give the people without basic knowledge in raytracing the impression you know what you are talking about.

i don't know what are other corner cuts but I'm sure M2 is not a true MLT renderer any more.
It is not a true MLT indeed, since it is not an MLT render engine. For your information Maxwell 1 was not using MLT either, so that "anymore" word does not fit fine in the sentence.


Luxrender is also some 10x (or more) slower than maxwell but if maxwell would been fully unbiased it's speed would be close to Lux.
Precisely one of the aims of Maxwell is to make unbiased technologies usable in real world productions. Maxwell has never sacrifized quality to gain speed. If in the future we consider developing a new engine based on biased paradigms to get lower render times be sure the speed boost won't be 3-5x but maybe orders of magnitude higher, but so far we are not very interested on that, since there are good biased renderers out there.

I am sorry this is my first post here; I'd like to start in a very different way, but when ignorance and injustice come together it's difficult to stay calm.

Thanks,

Juan

[cgbeige]

well one of the main nice things about Maxwell is the huge library of free arch shaders from MXMgallery.maxwellrender.com - there is a browser for these built right into Maxwell 2 and Maya/other plug-in apps now.

That project you were working on with the shower head would have been done a lot faster if you could have just grabbed this:

http://resources.maxwellrender.com/search.php#page=2&mode=1&category=&id=4799&search=oil&v1=0&v2=0&tipo=

and tweaked the shader (they are all editable).

But I didn't want to troll your thread

[tomNL]

Dear Alexander,

Obviously, you've started to abuse this forum just before reaching your 10th post. So, I guess you have a good reason for being here making fun with companies and their commercial products and I'm surprised this behavior is tolerated. Frankly speaking, your posts are violating the following forum rule, because from a technical point of view we find them utterly ludicrous and far beyond, it seems you're having big fun hoping those posts will deceive people and give damage to the product.

*** Totally ludicrous subjects ***
While the actual definition of this one may be debatable, we prefer to keep totally ridiculous subject matter that is totally unrelated to the industry, away from the forums as much as possible. While we all do enjoy some manner of light hearted or amusing conversation, we ask that members rather use the CGTalk IRC channel for this. A certain level of such matter is tolerated, but if a thread is really of very little substantial merit, the author of that thread will have to accept that it might be closed or removed completely.

It's a shame doing this... You could instead kindly present your proofs -if you have any- and ask for confirmation. But, coming up here making posts with unreal assumptions / conclusions about a product is not something innocent, nor could be taken granted as they if they were only critics. You here obviously blaming Next Limit and you say Next Limit is telling a lie about Maxwell Render being 100% unbiased. So, now prove it or make an excuse.

Best regards,

Tom

[cgbeige]

For anyone looking to see a big list of the changes in Maxwell 2, check this out:

http://www.maxwellrender.com/forum/viewtopic.php?f=1&t=32629

Judging from the content of that page, accuracy was still paramount in version 2, not performance.

edit: After reading that page, I feel like the $400 upgrade price was definitely justified. That's a shit-ton of engineering work.

[cgbeige]

haha (from the bottom of the page):


____HIDDEN FEATURES__________________________________________________________________

5.1 Vow to Stay a Hundred Percent Unbiased!
Sorry, do not ask for the opposite anymore.

[Cheesestraws]

One thing Maxwell 2 proves is that Maxwell might be 100% unbiased, and might have always been, but that certainly does not mean it is 100% physically accurate. :)

[Tora_2097]

Ha!
Half of the Maxwell force has arrived in this thread.

How exactly did you count the primary bounces btw? And when looking at your testimages, how did you arrive at the conclusion that MXW has now used all of its 3-5 primary bounces and was now spawning secondary rays?

I think you might need to revise your technical BG asap.

Ben

[Algoris]

Dear Alexander,

Obviously, you've started to abuse this forum just before reaching your 10th post. So, I guess you have a good reason for being here making fun with companies and their commercial products and I'm surprised this behavior is tolerated. Frankly speaking, your posts are violating the following forum rule, because from a technical point of view we find them utterly ludicrous and far beyond, it seems you're having big fun hoping those posts will deceive people and give damage to the product.

*** Totally ludicrous subjects ***
While the actual definition of this one may be debatable, we prefer to keep totally ridiculous subject matter that is totally unrelated to the industry, away from the forums as much as possible. While we all do enjoy some manner of light hearted or amusing conversation, we ask that members rather use the CGTalk IRC channel for this. A certain level of such matter is tolerated, but if a thread is really of very little substantial merit, the author of that thread will have to accept that it might be closed or removed completely.

It's a shame doing this... You could instead kindly present your proofs -if you have any- and ask for confirmation. But, coming up here making posts with unreal assumptions / conclusions about a product is not something innocent, nor could be taken granted as they if they were only critics. You here obviously blaming Next Limit and you say Next Limit is telling a lie about Maxwell Render being 100% unbiased. So, now prove it or make an excuse.

Best regards,

Tom

I don't have any intention about deceiving people or giving damage to a product. What i wrote down is my personal opinion and i didn't force anyone to accept it. However you are telling me to kneel and ask for forgiveness which is ludicrous indeed. I didn't do anything wrong to deserve something like that. And i certanly wont do such a thing. You want to tell me i don't have the right to think anything that contradicts with your product? As far as i know cgtalk is not a private marketing place for Maxwell and reacting so agressively and childishly to individuals opinion -even if it is criticism- shows how imature some members of this company are. I stay on my opinion that Maxwell's renders become more unreal with every new release and because I'm not a Maxwell developer I can only refer to what i see in final render outcome. However this is my personal opinion and I'm not encourageing anyone to follow it.

If I hurt any person however, then i deeple apologize. It was not my intention.

[inguatu]

I stay on my opinion that Maxwell's renders become more unreal with every new release and because I'm not a Maxwell developer I can only refer to what i see in final render outcome. However this is my personal opinion and I'm not encourageing anyone to follow it.

.

I think a large part of that is due the the inexperienced users broadcasting their Maxwell images to the community, leading people to believe that's the level of quality and realism that can be obtained using that software.

[cgbeige]

I really hope you're insulting me because I don't see any portfolio under your name, just "AntiMac." You judged Maxwell 2 as still being noisy by looking at a minute-long render and I'm supposed to take you seriously?

Algoris - having an opinion while throwing around fake statistics is misleading and disingenuous.

[Algoris]

I really hope you're insulting me because I don't see any portfolio under your name, just "AntiMac."

Algoris - having an opinion while throwing around fake statistics is misleading and disingenuous.

I didn't state any statistics, just speculated.

[ytsejam1976]

I don't have any intention about deceiving people or giving damage to a product. What i wrote down is my personal opinion and i didn't force anyone to accept it. However you are telling me to kneel and ask for forgiveness which is ludicrous indeed. I didn't do anything wrong to deserve something like that. And i certanly wont do such a thing. You want to tell me i don't have the right to think anything that contradicts with your product? As far as i know cgtalk is not a private marketing place for Maxwell and reacting so agressively and childishly to individuals opinion -even if it is criticism- shows how imature some members of this company are. I stay on my opinion that Maxwell's renders become more unreal with every new release and because I'm not a Maxwell developer I can only refer to what i see in final render outcome. However this is my personal opinion and I'm not encourageing anyone to follow it.

If I hurt any person however, then i deeple apologize. It was not my intention.

Quote your personal and free opinion in a Free forum

I don't know what' s the problem in this personal opinion.
If one man change the sales of maxwell, this is a problem. :)
If an opinion it creates problems, I think that autodesk would have had to ask the closing for all the threads of the Maya forum rendering for the problems on mental ray, the slowness of fluid effects and the other

[tomNL]

Alexander,
It's not as innocent as saying "they were my opinions". We build a hundred percent unbiased engine and you say you are 99% sure you figured it's not. Let's remind what you say:
I'm 99% sure that the primary and secondary lighting bounces of maxwell are limited. I don't know the exact numbers simply because i can not look in maxwell's source code but the fact is that they are limited and this means that it is biased in some way. I think that they tried to compensate these bouncing limitations with more and more robust shading algorithms. Also it looks that Maxwell now uses some form of radiance catching to reduce the number of shooted rays because in some places the tones of shades on an object are way too smooth and on the sides of shade tones where the transition of light changes, these transitions are very sharp.
Now your answer conflicts with your experiments. Let's compare:
BEFORE:I'm 99% sure that the primary and secondary lighting bounces of maxwell are limited. I don't know the exact numbers simply because i can not look in maxwell's source code but the fact is that they are limited and this means that it is biased in some way.
NOW:I stay on my opinion that Maxwell's renders become more unreal with every new release and because I'm not a Maxwell developer I can only refer to what i see in final render outcome.
One post ago, you were so sure it's biased for many reasons and suddenly now you say it's only your opinion by looking at the render output.
You want to tell me i don't have the right to think anything that contradicts with your product?Of course, you can. You can say you don't find it realistic, you don't feel satisfied with it, you don't think it's a good engine or is not successful comparing to others and many more... You can say anything you like. But instead, you are trying to invalidate something we say >> "Maxwell is 100% unbiased." and this stands for counter-advertising and moreover an accusation. It's far beyond being your innocent opinion about the product.

Best regards,
Tom

[cgbeige]

well saying "meshed nFluids are not as good for realism" is not the same as "Now in M2 i must admit that Maxwell is not so unbiased and physically correct as it was in the beggining." To make the same harsh claim about Maya, you'd have to say "Maya is no longer an animator – it used to do nonlinear/nodal and now everything it does is linear and flat."

[Algoris]

well saying "meshed nFluids are not as good for realism" is not the same as "Now in M2 i must admit that Maxwell is not so unbiased and physically correct as it was in the beggining." To make the same harsh claim about Maya, you'd have to say "Maya is no longer an animator – it used to do nonlinear/nodal and now everything it does is linear and flat."

How can you proove that it uses fully unbiased algorithms? Did you saw it's source code?

[cgbeige]

no, but I don't make claims I can't substantiate, even if I claim it as my opinion.

Anyway, I don't care to drag this on any longer. If anyone has any questions about my experience with Maxwell 2, feel free to ask. It doesn't seem like the demo is up yet. In the meantime, I'll be trudging through this Mental Ray project and counting my photons like they're prize jewels...

[InfernalDarkness]

well one of the main nice things about Maxwell is the huge library of free arch shaders from MXMgallery.maxwellrender.com - there is a browser for these built right into Maxwell 2 and Maya/other plug-in apps now.

That project you were working on with the shower head would have been done a lot faster if you could have just grabbed this:

http://resources.maxwellrender.com/...v1=0&v2=0&tipo= (http://resources.maxwellrender.com/search.php#page=2&mode=1&category=&id=4799&search=oil&v1=0&v2=0&tipo=)

and tweaked the shader (they are all editable).

But I didn't want to troll your thread

I very much agree with you my friend, and wish I'd had Maxwell to use for that entire project. My rendertimes got pretty absurd, even on this workstation, and I still never approached realism. What's done is done though; the client has signed, I already started laying the tile, and life goes on. But yeah, my shower material still looks horrible.

And you're never a troll, CGBeige. You're a welcome voice around these parts, much like Aunt Flow or perhaps The Tooth Fairy. We need your brand of naysaying and frank talk on occasion. Thanks for being a pal.

[cgbeige]

I had a little trouble getting it working at first but the web material integration is amazing. It just opens a searchable window, you find the mat and hit import and it's on your object and all reference textures are linked up from your sourceimages directory. This is the biggest benefit to Maxwell's shader format - it's platform agnostic and editable, unlike MR shaders that are compiled and non-editable.

[mverta]

As Tom mentioned, Algoris has stated no opinions, he has made false accusations, period. An opinion is, "I don't like the look of Maxwell." An accusation is, "[despite NextLimit's stating that Maxwell is unbiased] the primary and secondary lighting bounces of maxwell are limited."

Your assertions are false and misleading, and despite having principal programmers and developers attempt to clarify the truth, you refuse to acknowledge your error. So on top of everything, you're calling them liars.

Ignorant is one thing - malicious is another. Algoris, your behavior is both.



_Mike

[Algoris]

As Tom mentioned, Algoris has stated no opinions, he has made false accusations, period. An opinion is, "I don't like the look of Maxwell." An accusation is, "[despite NextLimit's stating that Maxwell is unbiased] the primary and secondary lighting bounces of maxwell are limited."

Your assertions are false and misleading, and despite having principal programmers and developers attempt to clarify the truth, you refuse to acknowledge your error. So on top of everything, you're calling them liars.

Ignorant is one thing - malicious is another. Algoris, your behavior is both.



_Mike


I'm sorry Mike if you feel that way. I'm not calling anyone a liar and the numbers of bounces were my speculation which i wrote down already. I never stated that my opinion is definately 100% correct. I simply don't understand why you find it so offensive.

[Bubbaloo]

I stay on my opinion that Maxwell's renders become more unreal with every new release and because I'm not a Maxwell developer I can only refer to what i see in final render outcome.

Done with Maxwell V2:
http://forums.cgsociety.org/showthread.php?f=121&t=794664

You were saying?:)

[cgbeige]

ya, I saw that image on the Next Limit site. Unbelievable. VRay is good but it's nowhere near that good.

[InfernalDarkness]

I never stated that my opinion is definately 100% correct. I simply don't understand why you find it so offensive.

Like the others said, it's not that we find it offensive. I'm sure nobody is actually "offended", or even upset. It's simply that your opinion is wrong. When someone around you goes about stating an error, it's human nature to try to correct this error. Since you've given no proof your theory is factual, and offered none, we've all just assumed that the error has been corrected already. You know as well as we do that Maxwell is unbiased; your continued assertion that it's otherwise is just stubbornness, an admirable trait in many cases.

[inguatu]

ya, I saw that image on the Next Limit site. Unbelievable. VRay is good but it's nowhere near that good.

depends who is using it.

[Sincrol]

the question is not "who" but "what".
a good software always make a good result... independt of whos using it.

[dmeyer]

the question is not "who" but "what".
a good software always make a good result... independt of whos using it.

You've got that backwards. Tools don't make art, artists do. Tools just make it easier.

[TRick]

ya, I saw that image on the Next Limit site. Unbelievable. VRay is good but it's nowhere near that good.

I read a lot of your (good) comments in this thread, but I can certainly not agree on this one. Since Maxwell is an unbiased engine and VRay is not they can not be compared. Just like a photographer you can make perfectly real images with Maxwell and only the talent of the artist (BOTH in software knowledge and artistic qualities) defines whether you make a good image. The same artist with the same artistic qualities can make the same image in any software if he knows the software good enough....even if he has to use photoshop to fill in the holes.

Personally I think that there are not that many artistic images made with Maxwell, which may probably be due to the fact that renders just take too long, which makes experimenting a lot harder then in other engines. You first have to be very technically savy to make good images, but this does not mean they are artistic....

[cgbeige]

The whole idea that all renderers are created equal is horse shit. Maxwell is not as well suited to animation as Mental Ray but Mental Ray is not as capable of realism. Sure, a good TD is able to fake their way to a nice image with a bunch of fake added lights but that doesn't mean it's as good a renderer for photo realism. I'd like to see someone get this with one source emitter in Mental Ray:

http://www.3eige.com/cgtalk/fiorito_mediumres.jpg

Sure, a good artist can make up for the tools to a certain extent but you're proving my point: that's compensating for the poor renderer and you might as well be painting in fake bounced light in Photoshop and saying that Mental Ray is just as good at that point. Until Mental Ray stops having independent light intensity and photon intensity, it will always be inherently unrealistic.

And I bought VRay for Maya for much the same reason you mentioned TRick – there are times when I don't want a photo, I want stylized environments or harder NPR lighting over realism and DoF. Unfortunately, with Mental Ray's passes, buffers and linear workflow in the state that they are, that meant throwing it out for that too.

[inguatu]

The whole idea that all renderers are created equal is horse shit. Maxwell is not as well suited to animation as Mental Ray but Mental Ray is not as capable of realism. Sure, a good TD is able to fake their way to a nice image with a bunch of fake added lights but that doesn't mean it's as good a renderer for photo realism. I'd like to see someone get this with one source emitter in Mental Ray:

http://www.3eige.com/cgtalk/fiorito_mediumres.jpg
.


The whole idea that Maxwell is the end all be all for rendering is horse shit which is what you keep telling everyone who will listen. I know you'll never get up on waving your fanboy flag around so it's pointless to reason with you.

There have been a myriad of postings from talented individuals using various render engines from mental ray, to brazil to renderman and so on. Granted, each of them have their own issues, strengths and weaknesses. Yet, people are still producing fantastic results with them. You refuse to acknowledge the fact that these renderers are worth their weight in gold in the right hands. By doing that, you're slapping the face of every user who doesn't use Maxwell by calling their tools "horseshit". Being an elitist doesn't prove your talented.

As for that sorry ass image you keep craming down people's throats, it's not that great as someone posted before in the many threads you've plastered it in. I've definitely seen better, even with the default Maya renderer. Quit using that as a basis for showing people how great Maxwell is, as it does not prove that point. Who really cares if you hit the "make stuff render automagically" in Maxwell with a single emitter? That's not talent. Tha'ts relying on the renderer to solve the scene without understanding how light, shadows, colors and elements all work together. I wouldn't want anyone to spend time in any renderer trying to recreate that shot as it doesn't have enough dynamic elements in it to showcase what a renderer or a talented person can do. Just give up on it already dude and show something new and fresh.

[Algoris]

The whole idea that all renderers are created equal is horse shit. Maxwell is not as well suited to animation as Mental Ray but Mental Ray is not as capable of realism. Sure, a good TD is able to fake their way to a nice image with a bunch of fake added lights but that doesn't mean it's as good a renderer for photo realism. I'd like to see someone get this with one source emitter in Mental Ray:

http://www.3eige.com/cgtalk/fiorito_mediumres.jpg

Sure, a good artist can make up for the tools to a certain extent but you're proving my point: that's compensating for the poor renderer and you might as well be painting in fake bounced light in Photoshop and saying that Mental Ray is just as good at that point. Until Mental Ray stops having independent light intensity and photon intensity, it will always be inherently unrealistic.

And I bought VRay for Maya for much the same reason you mentioned TRick – there are times when I don't want a photo, I want stylized environments or harder NPR lighting over realism and DoF. Unfortunately, with Mental Ray's passes, buffers and linear workflow in the state that they are, that meant throwing it out for that too.

When maxwell will be able to light the scene with a level of realism close to this, then we can talk about a physichally correct renderer: http://randomcontrol.com/images/randomcontrol/gallery/fryrender/01.%20exteriors/020.jpg

http://randomcontrol.com/images/randomcontrol/gallery/fryrender/01.%20exteriors/018.jpg

But with every new release Maxwell's renders are more and more fake and simular to vray-style renders. Maxwell's fake lighting can be mostly seen in exterior renders. It just can't produce a photoreal exterior.

[cgbeige]

wow - the idiot squad is out in full force now. ****HEAD - WE'RE NOT TALKING ABOUT TALENT, WE'RE TALKING ABOUT QUALITY RENDERERS. If you want to have a talent-based competition, start dancing.

why is it that every insistent shithead tossing the most garbage around here have no portfolio? I think you guys better go back to watching TV because you suck at armchair tech journalism.

[Bubbaloo]

Dave, please don't respond to these kinds of posts. Let's just say that Maxwell V2.0 is an amazing piece of software, and I for one can't wait to see more masterpieces made! Also, we can answer questions about the new version, but anything else is just not worth the time or effort, right?

[Sybexmed]

I'm looking for a way out of Mental Ray as well, i can't deal with all of these paremeters for every single scene i create. Its very time consuming and time equal money when you are a freelancer. Even though i have been using mental ray since i can remember, it is still a pain to work with.

[Bubbaloo]

I'm looking for a way out of Mental Ray as well, i can't deal with all of these paremeters for every single scene i create. Its very time consuming and time equal money when you are a freelancer. Even though i have been using mental ray since i can remember, it is still a pain to work with.

It's a trade-off. With mental ray, you have your time consuming settings and with Maxwell, you have possibly longer render times (depending on the scene). But Maxwell is getting faster, as demonstrated by Version 2. Now's a good time to give it a try!

[InfernalDarkness]

I gotta agree with you on this one too, CGBeige.

http://img38.imageshack.us/img38/4033/nogoh.jpg

I've never seen a mental ray render of the Server not found scene that equals your Maxwell render of the Server not found scene. It's easily the most realistic render of what you actually see on your screen at times that I've ever seen.

Probably gonna have to switch to Maxwell, so I can render perfectly realistic Server not found scenes too! Those are the bread and butter of any CG worker, these days.


If you're not laughing at me, you're not as funny as you think I am.

[cgbeige]

hehe - ya, I'm in the middle of migrating my host and that's the DNS transfer screwing up. And ya, the last time I checked, talented wasn't synonymous with masochistic. I gave up using QuarkXPress years ago too - Indesign didn't cause my eyes to bleed or induce fits of rage so it made sense to use. I guess the talented people are still using Quark though. Apparently QuarkXPress 8 just got... wait for it – drag and drop! WELCOME TO THE FUTURE.

I'm thinking of dropping Nuke and just using Powerpoint. That's where all the talented guys are doing the real work.

[Algoris]

hehe - ya, I'm in the middle of migrating my host and that's the DNS transfer screwing up. And ya, the last time I checked, talented wasn't synonymous with masochistic. I gave up using QuarkXPress years ago too - Indesign didn't cause my eyes to bleed or induce fits of rage so it made sense to use. I guess the talented people are still using Quark though. Apparently QuarkXPress 8 just got... wait for it – drag and drop! WELCOME TO THE FUTURE.

I'm thinking of dropping Nuke and just using Powerpoint. That's where all the talented guys are doing the real work.

It seems that i must repeat myself:

When maxwell will be able to light the scene with a level of realism close to this, then we can talk about a physichally correct renderer: http://randomcontrol.com/images/ran...teriors/020.jpg (http://randomcontrol.com/images/randomcontrol/gallery/fryrender/01.%20exteriors/020.jpg)

http://randomcontrol.com/images/ran...teriors/018.jpg (http://randomcontrol.com/images/randomcontrol/gallery/fryrender/01.%20exteriors/018.jpg)

These renders are from >>>>>Fryrender<<<<< which i find much more realistic >>>>>unbiased renderer<<<<< then maxwell. And it is much easier to handle. But still Luxrender is the most physically accurate and realistic engine.

With every new release Maxwell's renders are more and more fake and simular to vray-style renders. Maxwell's fake lighting can be mostly seen in exterior renders. It just can't produce a photoreal exterior.

[cgbeige]

haha. I didn't respond to your original posting because it was ridiculous to claim that any version of Maxwell couldn't have produced that image. please stop - you're just embarrassing yourself. First you pick luxrender and now you go to Fryrender's gallery and grab a random image – pick an argument and stick with it.

[mverta]

Figures.

I can't believe I wasted my time responding to this. I feel like I need a shower.


_Mike

[eikonoklastes]

I love software wars so much.

That new features list is mightily impressive though. What about those thin surfaces huh? Tasty stuff!

Still, mental ray is the best and kicks Maxwell's butt!


<runs...>

[Jozvex]

Oh my goodness, there hasn't been a thread this ridiculous in a long time! Haha! :thumbsup:

The only person who has to like the renderer you're using is yourself. I use mental ray, Maxwell and 3Delight for different things and I like all of them for different reasons. I don't rank them in order of best to worst and make little medals to hang on their packaging, because NOBODY CARES. It's not a representation of who you are as a person.

One thing is for sure, this thread is getting more and more biased with each iteration and yet it's still very noisy!

[redbranch1]

Josvex, that was a simple, rational, and well thought out post.

It clearly has no place in here.

Use whatever gets the job done, and pays your bills. If you are comfortable with the renderer's workflow, and the client likes the output, the rest is pointless.

I like to hear why people like, or don't like a renderer. I like to see what people can and can't do with a rendering engine. Because these help me learn, and decide what I want to do, and where I want to expend my efforts. I liked Mental Ray perfectly fine, until I heard enough about other renderers like VRay and Maxwell to take a look for myself. I appreciate people like cgbeige when they post about what works for them in the software they like.

But there's really no point at all to all the ranty, rendererX suxxors stuff. If you don't like it, don't use it. Rather than complain about a renderer, put together a render that shows how the software you like does it.

Not that this is going to make a difference, people flame war, because people like to flame. And unfortunately, this puts people like the maxwell developers in the position of having to respond, to refute factually incorrect posts

[inguatu]

. I appreciate people like cgbeige when they post about what works for them in the software they like.



Only if it's in an unbiased manner, and doesn't come off like a complete moron or eliteist, especially when there are people out there with more talent singing the praises of their individual software. I respect people who provide an unbiased reason for why they're using the tools they use. I respect people more who realize they need more than one tool on occasion and state the reasons why.

Lastly, I enjoy seeing great peices, REGARDLESS of what tools were used. Again, unbiased. If someone uses monkey crap on a page and it looks better than the best Maxwell render out there, then, hot damn.. that's awesome!

[InfernalDarkness]

To paraphrase Octavia Butler, it's the combination of our two key elements which keep such arguments alive...

From an outside, alien perspective, humans have vast intelligence crippled by their hierarchial nature. So, we're always trying to one-up each other, to the detriment of all. But at the same time, that's how we better ourselves and evolve, and progress.

I personally subscribe to a New Level of computer elitism: the older, the slower, the cheaper, the better. Call it a negative bias.

Gonna try Maxwell on my Commodore 64 ASAP! My guess is it'll blow mental ray AWAY. And after that, some benchmarks on my TRS-80, just for a nice clean comparison...

Off-topic, but Commodore finally released their dual-core C64, which should run Maya twice as fast:

http://img401.imageshack.us/img401/2296/c54x2.jpg

[Sybexmed]

Only if it's in an unbiased manner, and doesn't come off like a complete moron or eliteist, especially when there are people out there with more talent singing the praises of their individual software. I respect people who provide an unbiased reason for why they're using the tools they use. I respect people more who realize they need more than one tool on occasion and state the reasons why.

Lastly, I enjoy seeing great peices, REGARDLESS of what tools were used. Again, unbiased. If someone uses monkey crap on a page and it looks better than the best Maxwell render out there, then, hot damn.. that's awesome!

Anti-Mac, i love that haha.

[cgbeige]

ya, he's a multi-faceted troll

[Xharthok]

Is it now possible to render instances AND motion blur together in the new maxwell version?

[cgbeige]

doesn't seem to - no. Mike - can you confirm?

[inguatu]

ya, he's a multi-faceted troll


luckily the ignorant assume it means something its not. It referred to an election although this is supposed about Maxwell 2.

Speaking of which, I wonder if Maxwell will eventually go the route of multi-GPU processing versus sticking with software.

[cgbeige]

I really doubt it will go GPU - I'm yet to see a good GPU renderer and most are focused on the bare-bones "great for presentations!" type of thing. Anything's possible I guess but considering that VRay and Mental Ray and every other company under the sun is working on a side-project GPU renderer and Next Limit's always been about accurate simulation (Realflow, XFlow, Maxwell), it doesn't really seem to make sense to make a 2/3 quality solution for simulation. It's like PhysX - great for games, shit for film production since it's completely inaccurate.

[Hamburger]

VrayRT on a GPU doesn't cut corners as far as I know, it's to do with how the renderer is processed, not the GPU cutting corners. It is not the same as what Gelato was, it is not the same way game engines cheat either.

[Kel Solaar]

It's like PhysX - great for games, shit for film production since it's completely inaccurate.

CUDA was used on Transformer II and Harry Potter to do Dynamics Simulation.

KS

[cgbeige]

CUDA-based dynamics software or PhysX? I don't doubt that you can use the GPU for accurate dynamics stuff, it's just that PhysX isn't very good.

[Kel Solaar]

I understood you were meaning that GPU is not accurate when it comes to calculation, hence the fact I reported CUDA being used by ILM.

KS

[inguatu]

I could see large studios like that investing R&D into tech we don't have enough info on or can quite use ourselves just yet. I can see GPU processing eventually reaching the masses once it continues to mature. Some applications will take longer to adopt whether it be from budget constraints or the lack of talented developers within the company.

[SheepFactory]

I have been playing with maxwell 2 for a day and I have to say I am very impressed. The maya integration is amazing. It actually feels better integrated than mental ray to be honest and the render speeds were extremely fast.

Cant wait to try it some more.

[cgbeige]

just make sure you have the latest Maya plug-in. The first release had a few bugs like incorrect tone mapping for 32-bit images sent back to Maya and if your render failed because it was missing a texture reference, it would crash Maya. The latest one is 2.0.30

and don't forget to test multilight - that's where the magic happens. live light tweaking - as it renders.

[cgbeige]

and Maxwell is definitely not the slowest renderer - that title goes to VRay thanks to it's absolutely shit use of multiple cores. I've been waiting 15 minutes for a tiny square of this 480x360 image since it's decided to just use one core for this last bit:

http://www.3eige.com/cgtalk/vray_choking.jpg

http://www.3eige.com/cgtalk/vraycpu.jpg

Its CPU usage is all over the place. It's really bad and pissing me off to no end. I'm being told that this is just how it is with VRay by the experts

[SheepFactory]

Yea I am pretty much sold on maxwell. I went to the website to buy it but apparently you cant buy it yet if you are a new customer.

how much does a single license cost? i cant find any pricing info there.

[Hamburger]

and Maxwell is definitely not the slowest renderer - that title goes to VRay thanks to it's absolutely shit use of multiple cores. I've been waiting 15 minutes for a tiny square of this 480x360 image since it's decided to just use one core for this last bit:

http://www.3eige.com/cgtalk/vray_choking.jpg

http://www.3eige.com/cgtalk/vraycpu.jpg

Its CPU usage is all over the place. It's really bad and pissing me off to no end. I'm being told that this is just how it is with VRay by the experts

Never had a problem with Vray and it most definitely not the slowest render, scenes are always fast. Post a scene in the Vray thread and I'll check it out if you want.

No way that scene is worth 15 minutes on your kind of system. 1 minute maximum.

[SreckoM]

You could set smaller size render buckets.

[wurp]

gotta love all these uninformed and naive rants :)

I'm not gonna get into an argument about which rendered is better, but I do think that the brute force approach that for example maxwell uses will be the future, all these other tricks like irradiance caching were all invented because brute force gi was way too slow back in the day, today hardware is finally catching up and I for one hate all the settings you have to deal with when using vray, mr etc, with a brute force renderer you only have a few parameters to tweak and thats it.

and cgbeige, if one bucket takes 15 minutes in that picture you have not set it up correctly. Also, when using a bucket renderer it will only use one core per bucket, so if only one bucket is left rendering then it will only use one core, nothing wrong with that. If you are rendering very small images then simply decrease the bucket size.

[cgbeige]

got it - works now. Ya, that qualifies as a naive rant :p

hmm - I tried the bucket size with some other scenes and it's not a fix-all unfortunately. I have one that uses instances and it can't saturate half the cores for most of the render time. I do like VRay though - it's been great for the Maya integration and passes, linear workflow etc and I've got it working nicely with a Nuke relighting gizmo. The main problem with Maya integration and Maxwell is that Maxwell doesn't support Maya's procedural texture nodes - you have to convert them to file textures to use them with Maxwell.

[cgbeige]

Yea I am pretty much sold on maxwell. I went to the website to buy it but apparently you cant buy it yet if you are a new customer.

how much does a single license cost? i cant find any pricing info there.

It's $1000 I think. Go here:

https://www.nextlimit.com/newtpv/#productInfo

you contact sales to get your license - they make you a private license file that's not hardware locked and it activates the demo. It's the best licensing scheme around since you can use it on a laptop or move machines, etc and not have to worry about a dongle.

[cgbeige]

I gave up on trying to do this in VRay (it was a challenge I failed – caustics is trickier than I thought it would be in VRay) so I did it in Maxwell. A boring shot but I just want to sell the ice on turbosquid.

http://www.3eige.com/cgtalk/mxice.jpg

need to work on some spots, make the contact area wider and thinner and make it more interesting but not bad for an eight minute render.

[Bubbaloo]

Nice one Dave!

[cgbeige]

cross-posted from Maxwell's Maya forum

This might exist already but I was tired of creating these by hand so I made a script to automate linking of a distance locator with the selected camera's Maxwell focus distance. With a distance tool created and locator2 (the default second locator in a distance tool) to be moved to your camera location, select your camera and run this script.



if(!`attributeQuery -node $each -exists "mxUseFocusDistance"`)
addAttr -longName "mxUseFocusDistance" -attributeType "long" -defaultValue 0 $each;
setAttr ($each + ".mxUseFocusDistance") 1;

if(!`attributeQuery -node $each -exists "mxFocusDistance"`)
addAttr -longName "mxFocusDistance" -attributeType "double" -defaultValue 50.0 $each;

connectAttr distanceDimensionShape1.distance ($each + ".mxFocusDistance" );
string $mySelection[] = `ls -sl`;
string $shapes[] = `listRelatives -shapes -noIntermediate $mySelection`;
for($each in $shapes)
{
float $p[]= `xform -q -ws -t`; move -a $p[0] $p[1] $p[2] locator2;
select -af locator2;
Parent;
}


You will now have your Maxwell focus point always on the first locator.

edited: now includes the automated use Maxwell manual distance attribute. Thanks to Mihnea at Next Limit who does the Maya Maxwell support.

[cgbeige]

the Maxwell 2 demo is out. Make sure to get the plug-in for Maya as well

http://www.maxwellrender.com/mw2_demo_sec.php

Here's today's Maxwell render:
http://www.3eige.com/cgtalk/coconutscoconuts.jpg

[vlad]

A couple of anims showing what Vray in capable hands is able to achieve ;)
http://vimeo.com/5115304
http://vimeo.com/4765632
http://vimeo.com/4765159
http://vimeo.com/5407991

[cgbeige]

After seeing some animations with Maxwell, I do think that VRay is still better suited to animation, even if it's much slower than Maxwell 2 for final renders. The problem with Maxwell animation is that since it has no control over anti-aliasing or GI cache, it can look flickery.

Still, I think the words "capable hands" says it all: VRay takes a lot more work than Maxwell to get good final render quality results. Not nearly as much as Mental Ray but good caustics was a real problem for me in VRay. I'll get it eventually though.

Both companies have great support though – Maxwell 2.0.1 just came out and fixed a ton of small issues and my problem with VRay threads on OS X is apparently fixed now. Just a slight difference between how long it takes to get an issue sorted out with Mental Ray.

[vlad]

Take a look at CGarchitect's gallery, you'll find plenty examples that using Maxwell is no garantee of success. And quite frankly, once you master the ins and out of a particular render engine, setup time/ease of use becomes pretty much a moot point. And saying that Vray is much slower than Maxwell for stills just proves you have no idea of how to set ip up correctly. No offense. ;)

[cgbeige]

That was definitely the case with Maxwell 1.7 but take a look at 2 - it does ridiculous caustic scenes in no time.

[Mihai75]

After seeing some animations with Maxwell, I do think that VRay is still better suited to animation, even if it's much slower than Maxwell 2 for final renders. The problem with Maxwell animation is that since it has no control over anti-aliasing or GI cache, it can look flickery.

GI cache and flickering in animation??? Maxwell doesn't use a "GI cache", it does no interpolation of samples that introduce the usual flickering in animations. Could you post an example?

[cgbeige]

i know it doesn't use a GI cache - since it recomputes the GI for all frames, if there is any noise in the render, it flickers visibly across frames. I've seen it recently here:

http://www.maxwellrender.com/forum/viewtopic.php?p=312533#p312533

Am I wrong?

[Bubbaloo]

i know it doesn't use a GI cache - since it recomputes the GI for all frames, if there is any noise in the render, it flickers visibly across frames. I've seen it recently here:

http://www.maxwellrender.com/forum/viewtopic.php?p=312533#p312533

Am I wrong?

I don't see any flickering there, just low res aa issues. It's a completely different issue. Fix: render at higher resolution. ;)

[dmeyer]

I don't see any flickering there, just low res aa issues. It's a completely different issue. Fix: render at higher resolution. ;)

+1. I see some aliasing from WMV compression on the horizontals...could be fixed by higher res and/or tweaking compression settings.

We've been using Maxwell for animations for a couple years with no notable flickering issues.

[Mihai75]

Flickering in animations using GI are splotches that appear/disappear in different places at different frames because of the sample interpolation which Maxwell doesn't do. The patterns you see there are from the resizing probably, and also of the geometry itself, you see moiré patterns, same reason they don't recommend to wear a striped suit on TV :)

[CGTalk Moderation]

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.