prman 11 or mentel ray 1.5?

[stickyblue]

well MR 1.5 is free:) hehe

[Ultimatebadass]

I like the new MR 1.5 because it's much easier to work with and free ;).

[playmesumch00ns]

Well MR is free, do you really have $15,000 for RAT? But PRMan is a much better renderer...I love renderman

[somlor]

I thought Rat 5.5 with Prman11 was only $8k?

In any case, if you can afford it, matred, it's probably the best renderer on the market.

Ambient occlusion :love:
http://zj.deathfall.com/occsurfii.htm

(s)

[matred]

thank u so much for the reply..i got both..prman is bit difficult when comparing to MR...thanks somlor for the link..

[playmesumch00ns]

Only $8K? AND Mental Ray as well, is that the free plugin or the standalone version?

[enforcer2k]

PR MAN 11 by all mean looks like a better offering because of what you see in movies, but remember this, that's mostly because those movie have tons of artist working at the same time to achieve those effects, renderfarms that can commit up to days per cpu to render 1 frame of animation (we're talking a lot of CPUs) and a ton of compositing layers and effects. Secondly, it's PR (public relations), Mental Ray is also widely used in movie production, maybe not as widely as Renderman, but to a respectable degree, you can check http://www.softimage.com and see some of the movies mental ray has been used in. We're talking about a lot of movies from Panic Room to Jurrassic park 3 and Fight Club.

Comparing renderman and MR 1.5 side by side, MR 1.5 offers better integration and 98% support for maya's shading network (Some dynamic links don't work, minimal) which renderman DOES NOT support (it uses a separate shader builder module call SLIM for material AND lights), the NEW raytraced effects are SLOW, and thought the radiosity and and caustics are nice, MR competes with more mature methods there too, though I need to see some further tests. Mental Ray has very fast raytracing with high quality results, it makes use of the new Ocean and RAMP shaders that are part of Maya 4.5, anf has very very fast raytraced and area shadows. The depth mapped shadows in MR for some odd reason are very slow, but you can ditch them in favour of the raytraced counterparts in a heartbeat.

PR MAN 11/RAT (Renderman Artisit Tools) 5.5 has some good points, one being that it's very stable, supports distributed rendering (MR supports this too with on per frame basis, so even your test renders can be accelerated) and can yeild quick renders if you plan to fake raytracing, radiosity, and not use raytraced shadows. If you plan to use any of these, it is will fall behind MentalRay very very quickly in terms of speed and quality. On the down side, you need to really learn what the MR variables do because thay are not immediately intuitive, so you need to do you READING and experimenting cautiosly to get the best results, but it's not hard if you take things in step.

Personally I chose MR, but I hope this helps you make your own decision.

[matred]

thanks enforcer..i have already come to that decision(choosing MR)..thanks for the brief answer..

[MaDSheeP]

but.. MR doesn't support SubD's =(

ah well.. soon enough it will :)

[enforcer2k]

True true. Alias still needs to put some more work into subds and eliminate the performance issues. Most people work with smoothed polys I believe. However, it will be nice to have that it supported soon, along with glows and paint effects! ALIAS ARE YOU LISTENING????

lol :wip:

[Ckerr812]

Originally posted by enforcer2k
ALIAS ARE YOU LISTENING????

lol :wip:

http://www.aliaswavefront.com/en/companyinfo/contactus/productfeaturerequest.shtml

I am one of many people that go to the program testing gatherings Alias holds every now and then to make suggestions and try out beta stuff. It's very rare they will see anything on a message board. Use that link if you want to message them.

[matred]

thanks cherr for the link..it would be very useful for the people to post there expectations..

that's a nice question "are u listening alias?"

it would be great that MR or anyother 3rd party render software support maya's post rendering effects (ofcource there are doin a bit now..but still)

[~~ZIP~~]

mentel ray 1.5 ONLY maya4.5:hmm::annoyed:

[thesaint]

Just a couple of points here that may be valid.

Firstly, what is the choice? If you are considering $5,000 for PRMan 11 ($8,000 including RAT) you are really looking a very different working pipeline than MR. You are comparing apples to oranges.

If you have time to learn PRMan and the RI (Renderman Interface) together with SLIM (MTOR does in fact have minimal adoption of Maya Shaders) then you will find a level of sophistication that MR is simply not able to touch. Or, since you are spending so much, Animal Logic sell MAYAMAN that will translate the whole of your scene, including shaders, lights and all into a RIB to serve out to Alfred and on to PRMan.

Secondly, Ambient Occlusion (like Global Illumination) is still too slow in ANY package to be of serious production value, PRMan 11 simply steps into the arena with this release. Ray tracing is rarley if ever used because it is soooo expensive on cpu time. To the best of my knowledge in Bug's Life only about 3 or 4 scenes used it (served via BMRT's ray server). Mapping did the rest. Same in Both Toy Story's, Buzz's helmet was a reflection map with the center knocked out to avoid obscuring his face. You can waste a lot of time raytracing when a simple image map would have done the same job in a tenth the time.

MR is a great package, but PRMan it ain't. If you have the cash, the time, the energy and the resources get PRMan. If you want to get Renderman savvy, get AIR for $400 or 3Delight for free. Most serious studios rely on some sort of renderman compliant renderer so it is good to get in the swing of things.

Just my 2cents (mmm, maybe with this length it is more like 10cents!)

D

[enforcer2k]

Well like I said, not that Pixar's renderman is not capable, but the good work it produces does require teams and teams of artists. And most users on this board simply don't have those resources. And when it comes to reflections and raytraced shadows, maya's default renderer does a way better job, even generates shadow maps faster. So in all practicality I can only seldom recommend it.

[thesaint]

What??? Maya's own render engine???? I thought we were looking at MR vs. PRMan, two production capable renderers. Maya's own render engine is not production ready for anything but thier own promotional work.

Now that is Apples to Oranges.

Most studios use renderman in some flovor or another, so it is worth knowing how to manage RIB files and so on.

I agree that in this context MR is much better for shear usability! I did not mean to infer that MR was in any way 'stupid', just that asking us to compare the two was like trying to compare a cow to a horse.

Now, to address the claim that Shadow maps are faster in Maya. That is true only in the most basic of scenes. Once things get a little more complex the tables turn in a big way. PRMan has several functions and calls that make it substantially faster than any other render engine i have used. Irma is one of them, LOD is another and i could go on. Like you, i seldom recommend PRMan because it is a learning curve, but once over the initial stages it is a fast, powerful, beautiful engine that can handle scrutiny up to IMAX resolution.

Seriously, do you think ILM, DD, Image Works, DreamWorks, Disney and others would spend all that money on a render engine if it took hundreds of people to operate and hours longer to render for results that were no better than the renderer that came with Maya???? no, surely not. I use it everyday at work (though that is to be expected) and its stability, speed, predictability, extensability and quality are awesome and second to none

BUT, Mental Ray is the right choice for most posters. I do however believe that there is no such thing as 'posters here'. We are all trying to be the best we can, to stretch to new heights and extend our abilities, no matter where we are in our careers. Learning 'renderman' is a valuable excersice and can be done for free (with 3Delight and so on) so what have you to loose???

[enforcer2k]

lol. If you think so, great. Not every has a 1,000 object scene to benchmark your assumptions against. If anyone is really into raytracing, Renderman is SLOW. Now Renderman is highly programmable and an has a flexible pipeline, and interfaces with maya in a very stable way, but it can overburden an artist or small studio with the complexities if you don't know how to approach the renderman shading language. U cannot do everything in the UI.

So understand this, I'm not saying Renderman artist tools won't get you professional results, anyone who has seen the sentinel texturing in the Matrix can't deny that, and the list goes one, *but the investment Renderman requires especially for a single user or small group at the cost of $5000+ a license is not worth it I believe. And can you get film quality results with Maya's basic renderer, YES. And with MR, yes. If you look at softimage's site, you will see MR has been used in a lot of film productions. Pixar has been around longer so it has more credits, but ILM recent made a commitment to using MR in production and acquired more licenses.

ILM and Mental Ray (http://www.mentalray.com/c225.html)

If you want links to film quality Maya renders I can dig up a few for you, and you can look on Highend3d.com yourself. Mental ray is nice in that is supports GI and has faster raytracing abilities, distributed and parallel rendering, and much more which makes it a capable production renderer like much like Renderman.

FIN.

[thesaint]

I think we are making a lot of the same points. Great discussion, i hope we get to pick it up again but for now, this poor guys post is getting lost!!! hehe

As for Maya, i would love to hear of any films that have used Maya's out-of-box render engine. It would be fascinating, and i am always ready to stand corrected...

Hope we get to chat soon.


D

[Johnlittle]

Nice discussion guys.:)
What do you guys think of Gi Joe for Maya 4.5 unlimited? How does it compare overall to Mentalray or to 3Delight.
So paintFX does not work with Mental Ray yet? Thanks for helping me out on these questions if possible. :beer:

[thesaint]

I haven't, but i have taken a look at the home page and script and it seems really sound to me.

I would like to know what the impact on render times that would have? Provided ray tracing is turned off i would imagine this would be an almost ideal way to achieve the effect, because it should , in theory, be cheap to render.

anyone used it? Any stats on it?? If not, i guess the ball lands in my court to do some tests....

[matred]

is there is any possibilities to render maya'a post rendering effects(paintfx) with prman 11 or mr1.5?

[gmask]

Originally posted by thesaint
As for Maya, i would love to hear of any films that have used Maya's out-of-box render engine. It would be fascinating, and i am always ready to stand corrected...


I have been using Maya's renderer for commercials, broadcast design and even film since version 1.

I realise that it cannot do everything and it is a memory hog but considering the more or less unlimited rendering licenese it's a bargain despite it's flaws.

Last I heard it was a grand per CPU to load up with MR standalones and I don't really feel like spending $10 grand or whatever to beable to get my work done and if I were to spend that kin dof money I wold want my sub-d's to work and for their to be shared memory between MR and Maya.

For small shops and freelancers in the rest of the industry besides film cannot afford PRman or MR and does not really need it IMO.

anyway check out www.pixho.com for some examples of getting the most out of the standard renderer. This guy has a fake for everything.

[gmask]

Originally posted by thesaint
I haven't, but i have taken a look at the home page and script and it seems really sound to me.

I would like to know what the impact on render times that would have? Provided ray tracing is turned off i would imagine this would be an almost ideal way to achieve the effect, because it should , in theory, be cheap to render.

anyone used it? Any stats on it?? If not, i guess the ball lands in my court to do some tests....


According Emmanuel GI Joe is a little slower than MR in acheiving the same effect. I will be doing some tests soon myself to see if this is true. So far I would say that working with MR in Mayay is slow.. much slower than through XSI but this is mainly due to the memory space not being shared and having to wait for the mi file to be generated when doing tests.

Is is chaper to render this way but is probably less time consuming than PRman in setup I would imagine. MR is deep as well but does create some nice result with little setup.

[uday]

my .000002 cents
paint fx is a post effect.
like a photpshop blurr plugin.
try geeting a reflection of any paint effect on to a reflective plane
thats mayas inbuilt renderer we r talkin about
so u have to forget prman or MR as they r third party solutions

[gmask]

Originally posted by uday
my .000002 cents
paint fx is a post effect.
like a photpshop blurr plugin.
try geeting a reflection of any paint effect on to a reflective plane
thats mayas inbuilt renderer we r talkin about
so u have to forget prman or MR as they r third party solutions

Wel there are ways to do this by rendering the Pfx from a second cameras POV and using it as reflection map but that would require some effort :eek:

[thesaint]

In reply to Matred and uday:

No, at least not with PRMan, it is a post render effect, so it would not show up in any RIB. Not sure about MR, but if it doesn't handle Subd's i can't imgaine it would handle post-render processes. But it might.

In fact, as i think about it how can paint fx work with things like transparency. If the render engine, even maya's own, is processing the paint effects post to geometric calculations it must appear as a kind of 'on / off' switch. So if i created a piece of netting through which i looked at a grass field created with paint fx and mapped a transparency map to it, in theory (to my limited knowledge of the renderer) the paint fx would ignore it and appear right over the top of it. Is that right???

Anyway, i would love to see some credits for a film rendered in Maya's own render engine. It always looks so muddy and the AA is really quite full of artifacts, if i took a maya image and threw it up on a screen 20 feet wide would it really hold up??? I just can't see it. Sorry for those who take this to heart, really i am, but Maya's own render engine is not something i can imagine an art director looking at closley and being alright with it. But i am talking about film. I have used it on small shorts myself, and been happy with it too. Intergrating with live action would seem like a big problem too. Anyone got any experience with that? I have to say that i do like MR (not as much as PRMan, but what do you expect?!) if it wasn't a ray tracer only i would use it.

[gmask]

Originally posted by thesaint
in theory (to my limited knowledge of the renderer) the paint fx would ignore it and appear right over the top of it. Is that right???


That's correcty you;d have to render the window for example as a seperate layer and composite.


I have requested that Pfx have the option of being converted into geomtry but I doubt this will become a reality :-(

[thesaint]

that's too bad. Hope they listen to your request and impliment that, it would be really nice (and maybe not so hard) to create a basic geometric surface from that data. I mean, all that data about location in space must already exsist.

Is there a third party plug-in that they don't want to tread on?

Anyone know maya's API????

[gmask]

Originally posted by thesaint
that's too bad. Hope they listen to your request and impliment that, it would be really nice (and maybe not so hard) to create a basic geometric surface from that data. I mean, all that data about location in space must already exsist.

Is there a third party plug-in that they don't want to tread on?

Anyone know maya's API????

that's what I allways thought.. why can't you just run a mel script that extrudes on it's paths??? Yeah maybe through the API.. The only thing siumilar to it that I can think of is xFrog www.xfrogdownloads.com

this plugin is very cool..

But you still have this problem with fur unfortuantely :-( But again.. still why not convert it to gemoetry.. Shave and A HAircut can do that.. guess shave and a haircut is a better option.

[thesaint]

that's what I allways thought.. why can't you just run a mel script that extrudes on it's paths??? Yeah maybe through the API.. The only thing siumilar to it that I can think of is xFrog www.xfrogdownloads.com

That is a very cool plug! What is with the shadows / reflections though? They appear to be about twice the width of the casting geometry?

You think something like that could be done in MEL?? I'd have a crack at it if someone had some pointers about where the Paint FX spacial data was located. If it did require an API, well, i'd leave that to an expert. I should ask one of the TD's here.
I understand they know a couple of things about Maya :p

Well, back to the thread. PRman rules. And i am paid to say that...
MR looks nice, but it really needs to be much deeper (in terms of ability to get abstract results) And it relies entirely on expensive rendering techniques which often get in the way. Imagine if Buzz's helmet accuratley reflected the entire scene in every shot?? It would have obscured his eyes and face, and would have required way too much ancillary animation to sell the scene and been so time consuming at render time it would have been a joke. The fact is that using pre-rendered maps that could have transparencies attached to them (to reveal his face) saved time, helped us get the results and no-one has ever asked me why the reflections were inaccurate. Well, we could have used Maya's own render engine and saved a buck or two now i know it is film quality :-) (just kidding with ya)

[gmask]

>>You think something like that could be done in MEL??

Oh.. definately not.. you'd have to do it through the API if you can at all. I would be curious to know if it can be done.

GI and all that is the latest buzz word.. You never really need all these features turn on at once..I often feel that those who wish for that are hoping that the one button solution to rendering will make their work better and it really won't.. if anything it will make it look like anybody else's work who clicked the same button :-(

I got into PRman years ago but it's alot of work and for commercials and the like it takes too much time to tweak but I well oiled team is going to well function like a well oiled machine and beable to use complex tools like Renderman despite short deadlines.

[thesaint]

I asked the guys here and they said (paraphrasing):
'sure, why not. At the API level you have access to most everything the Alias programmers do, so you should be able to write routines that would attach geometry. However, they did also say that it would be like the cart pushing the horse. Laying down what is essentially just a texture in coordinate space and using that to create actual polys would be quite the trick. Which is why it is post process and not in any of the pre-calculate routines.'

So it is possible, but almost not. Which is probably why Alias haven't touched it. It would be very cool though. Alias, are you listening????


excellent point gmask, i agree. PRMan without SLIM, especially, is pretty intense. And i have to say thath i agree about that 'one button' comment. It gets a little old to hear about GI, caustics and other forms of ray tracing and ray casting. In the new version of the released PRMan they included a lot of ray tracing routines, people say 'about time too' but honestly, even in Finding Nemo, very few shots use it. One nice thing about scriptable renderer's is that you can assign almost any attribute (or write code for it) to a .sl (SLIM -- the shading app in MTOR) node. In SLIM the function call that supports raytracing is called SwissArmy. So if you want certain items ray traced, you attach that node to the object. You can then tell it what you want it to do (IBI, GI, Caustics etc) and then attach that node to any other and layer it, and leave other objects to be lit traditionaly using shadow maps. This helps in the outdoor shots because you can SwissArmy enough objects the sell the outdoor look, but leave everything else to 'normal' routines so the render time does not go through the roof. This kind of layering is possible only in apps that allow that kind of control. MR does not i believe, since it is a 'pure', i.e. only capable of ray tracing. You could do a similar thing in Maya by simply exluding certain objects from the GI Joe (or other fake) routines, but that gets difficult to manage in larger scenes. This wouldn't work for MR because it can only ray trace, so you'll have the expense of that math even if you exclude certain lights. So Maya has one up on MR in this respect and in my opinion.

Hope this post wasn't too long. gmask set me off, blame him ;-)

[gmask]

>>>So it is possible, but almost not. Which is probably why Alias haven't touched it. It would be very cool though. Alias, are you listening????

hmmm.. let's find out ;-)


>>>In the new version of the released PRMan they included a lot of ray tracing routines, people say 'about time too' but honestly, even in Finding Nemo, very few shots use it.

Wasn't it there all along but for some reason hobbled? I have read about Franken renders and the lik and they sound like totlaly nightmares and why not with a name like that!

However without raytracing how would you handle refraction in PRMan?


>>>This wouldn't work for MR because it can only ray trace, so you'll have the expense of that math even if you exclude certain lights. So Maya has one up on MR in this respect and in my opinion.

I guess so.. I'm only just starting to play with MR but it will be awhile before I ventur einto on a paying job unless I am being paid to do specifically that.

>>>Hope this post wasn't too long. gmask set me off, blame him ;-)

Don't blame me.. it's all relevant.. and we are still on topic..yeah! :applause:

[thesaint]

LOL

I wish i had the time to dig up that API and mess around, this kind of stuff intrigues me, alas, i am a lover not a fighter when it comes to C++!!

To the best of my knowledge (which is weak concerning public release) it was not there. I think what you are refering to is the RI (renderman Interface) which included the specs for Ray Tracing a long while ago. The deep Shadow maps used in Monsters were developed long ago. 4, maybe more, years (at least it seems that way to me)

For refraction and reflection (when really needed) we used BMRT until just after Bugs Life, mm, maybe quite a while (but not now, obviously!).

Let me give you and example from Geri's Game.
During the short Geri's glasses are used to enlarge his eyes to give him that angelic appeal, however, if we had used real refraction routine it would have meant that we had to view Geri from only a small set of angles, and that is no good. An algorythm should never decide an artistic function like camera placement. So instead the eyes where mapped onto the glasses and then moved until they where clearly visable from the camera, but in reality this left them in totally the wrong place. Even when we pointed this out to scholars or physics they were unable to tell it had been faked. I ahve attached an image for your reference. Isn't it amazing how 'wrong' they are, and yet it is not even appart when looking at it. We all have such a dismall ability to understand refraction and reflection and so on. In Toy Story we used only a handful of reflection maps (stills) for Buzz's helmet, yet did you notice that it wasn't the room reflecting in there? Neither did anyone else.

Oh my gosh -- how i do keep taking up this thread! Sorry folks!

[thesaint]

missed the images:

how do upload more than 1 image into a post??? Anyone??

[thesaint]

And the other...

[somlor]

Yeah, it really does look just perfect. There is a really interesting chapter about this and other sorts of physically "wrong" effects you can get away with in the first chapter of Advanced Renderman, as I'm sure you're aware of.

Great discussion.

(s)

[Lyr]

Great discussion going on here. A bit off topic, but I thought I jump in real quick. There is an opensource renderman compliant renderer that I just came across called aqsis, it's even got a little tool to interface with maya. I haven't used it yet, but I thought some of you guys might be interested in checking it out. http://www.aqsis.com/

[thesaint]

i really need to get that book. Do they cover Geri's Game or do they (Larry and Tony) use other examples? Why is every book for our industry around $50? oops, that's for another thread...

Yeah yeah, captive small market. Blah blah. It stinks all the same.

I totally thought the guys and girls here would really be into thier ray tracing and ray casting techinques, but this doesn't sound like it is so, i think that is really neat. I confess that i had to 'get over them' myself. As soon as i realised what was possible with simple maps i was sold. Even for my own personal shorts i don't use anything more sophisticated than deep shadow maps.

So back to the original question which was MR or PRMan.

What lets PRMan down is the complexity.
What lets MR down is its inflexability.

How do we unravel this to advise the original poster as to which software to buy. I always presumed that if you had the cash to plonk down for Artist Tools you should, just because you can grow into it and it is always good to stretch yourself a little, but the barrage i got for suggesting this makes me think that easier is best, even if it ultimately limits you. What is the solution?

I agree that for sheer learning curve and usability out-of-the-box MR is a good choice, but is it not always merely a companion to the Maya renderer, since there is stuff you can only do in Maya's, like Paint FX and Subd's??
You can't do Paint FX in PRMan either -- but the subd implimentation in Artist Tools is first rate at least.

How do we advise this artist?

[thesaint]

i have played with aqsis at home. I like it and i like 3Delight too.

www.3delight.com

The thing that makes them hard to use, especially for newbies to the rendering paradigm is the missing MTOR or MayaMan tool set.

getting to a RIB can be overwhelming without them, although both have great support from tutorials and forums. In fact Aqsis is supported by MayaMan is it not??

Did you guys know that Pixar runs a user group for renderman??

https://renderman.pixar.com/

In the support section, bottom right.
check it out.

[stunndman]

that's really one interesting dicussion going on here - i just read through the last few posts and therefore i might ask something that has already been answered

@thesaint: why do you think MR is inflexible - i'm just wondering because to me it seems to be very powerful - where does PRman distinct itself from MR (just considering the feature set and not performance issues)

>>>This wouldn't work for MR because it can only ray trace, so you'll have the expense of that math even if you exclude certain lights. So Maya has one up on MR in this respect and in my opinion.

either i am not able to follow your argumentation or you might be wrong with your assumption - here are three quotes from the MR documentation concerning primary and secondary rays


The software [MR 1.5 for Maya] uses advanced rendering acceleration techniques such as a scanline algorithm for primary visible surface determination and the BSP ( binary space partitioning) algorithm for secondary rays. These algorithms can be fine-tuned by optional user-settable parameters to achieve even higher performance than normally achieved by the built-in automatic scene cost analysis.



trace on|off
Normally, mental ray will use a combination of a scanline algorithm and ray tracing to calculate samples of the scene. If trace off is specified, ray tracing is disabled, and mental ray will rely exclusively on the scanline algorithm. Since the scanline algorithm can only compute straight rays from the pinhole camera, reflection rays cannot be cast and refraction rays are computed like transparent rays, which do not allow control over the ray direction based on the index of refraction of the material. Lens shaders cannot alter the ray origin and direction. However, reflections onto environment maps do work. Shadows are also affected if ray tracing is turned off. Ray tracing is turned on by default.




scanline on|off|opengl
This statement allows turning off the scanline algorithm. By default, mental ray tries to use a scanline algorithm for straight rays from the pinhole camera, such as primary rays. In most cases this gives better performance than pure ray tracing. Turning scanline off forces mental ray to rely entirely on ray tracing. This will generally slow down rendering but in some cases, for example when the task size is very small, the overhead of initializing the scanline algorithm may outweigh its benefit and turning it off can result in an improvement in speed. The opengl option will cause mental ray to use OpenGL hardware if present to further accelerate rendering. If possible, the master host will use OpenGL to generate acceleration data that the scanline algorithm then uses for intersection testing. Also see task size below.

[thesaint]

Scanline algorythm reply:

PRMan uses REYES not scanline. In fact scanline is one of the reasons AE has issues rendering for film, but that is more a bad implimentation than a bad foundation. REYES (Render Everything You Ever Saw) uses micro polys to render an image and is widely accepted as the best method, although Scanline is no slouch, as MR results have shown us. Also PRMan uses what is easily the best AA algorythm out there, and it is the subject of numerous patents.


Trace on | off reply:

What i did a bad point of trying to say was that in MTOR you can actually select the object in the scene you want to have affected. Like a shader node in Maya, you can assign an object to be ray traced (and subsequently how you want it ray traced). You can script this shader as well as layer it with other shaders too. There really is nothing you couldn't do. Even if PRMan doesn't have a function for it you could write your own into the RIB or .sl and get it to do whatever you need.


scanline on| off opengl reply:

I cannot make out what it is saying here. Ray tracing is a method of calculating the path of a ray of light from the camera back to the light source, while scanline (to my understanding) is a method of breaking the scene into managable chunks of data that are then rendered one pixel line at a time. I have no idea how or why you would turn this off, or what bearing it would have on how the scene shadows, refraction and reflection is calculated, except that all effects the time to render.
Any help on this one is appreciated.

There are the difference that are immediately apparent.

[stunndman]

thanks for taking a look at all three quotes - actually i just wanted to clarify your statement


...
This kind of layering is possible only in apps that allow that kind of control. MR does not i believe, since it is a 'pure', i.e. only capable of ray tracing.
...


i would conclude by the documentation that MR is not a 'ray tracing' only renderer

here's some information on scanline from the MR docs


scanline (default miTRUE) enables the first-generation scanline algorithm if applicable to improve speed. miFALSE disables scanline rendering, and 'o' replaces the standard scanline renderer with an OpenGL renderer that used the OpenGL hardware of the system for fast rendering (SGI only).


and second MR does ...


...allow detailed control of how a particular object contributes to certain rendering stages. These Flags include: Visible, Trace, Shadow, Caustic, and Globillum.


i think that's what the SwissArmy thing does in PRman?

and don't get me wrong - i'm not trying to defend MR here or something - i'm just trying the get some more insight into the occult subject of render engines

[somlor]

thesaint: Yes the Advanced Renderman book goes into the exact cheats used on Geri's Game and Toy Story (among others) that you explained which is why I assumed you read it. :) You should definitely pick it up, and you can grab it used off amazon for under $40 usually.

As for Aqsis lacking MTOR, check out Liquid http://liquidmaya.sourceforge.net/ which has been successfully compiled against the Aqsis library and is available in the downloads section of http://www.aqsis.com/

And this is totally off-topic, but IMHO I think that as Brazil matures, goes standalone, accepts RIBS and it's Orchid scripting language (to provide full support for Renderman compliant shaders) becomes useable it will be a more juicy option than MentalRay for people without the funds/resources for Prman, but who want to work with a Renderman compliant renderer with less hassle. Of course, Entropy filled this niche quite nicely but that's a whole 'nother thread. :annoyed:

(s)

[thesaint]

uh oh. I wondered when someone was going to mention Entropy. Gulp.

I understand Liquid is still in need of a windows compiler. Do you know what the status of that is? That would fill one deep niche for wintel users and would open up RenderMan compliant renderers in a big way.

As for Entropy, i could never, nor would i want to, assume what was going on there. I have kept up with Larry and his exploits since he jumped ship and have always wished him best. I never used entropy but those who did say it was competant and fluid to intergrate. He is a formidable talent in the field of rendering, something i am not. I am a simple animator with big opinions about technical stuff! (i know you didn't solict that, but i am glad to get it over with!)

I know nothing about Brazil. Who makes it? What is the platform and why should this original poster consider it right now (trying and pull it back on topic!) if all the cool stuff is just around the corner??

[somlor]

Heh, yeah, like I said, the Entropy thing is a whole 'nother thread. :rolleyes: Didn't mean to touch on that hotbed.

As for Brazil, well I'm not exactly sure how soon the things I mentioned are going to materialize, as I read them from a Brazil design overview written early in 2001: http://www.splutterfish.com/sf/b_technology.php3

And so really I am just babbling and should keep my mouth either shut or on topic <grin> but since I know very little about MR, and only a little about Renderman I figured I'd throw in my two cents. :shrug:

;)

(s)

[matred]

hello guys,
this post is really getting a very interesting and informative post. thanks "The saint", "gmask", "enforcer2k". i just want to post this link regaring paintfx in renderman..check it out

http://www.highend3d.com/maya/tutorials/paintfxqa/b.3d

[Mauritius]

Hi everyone. I completely missed this thread since it was in the MAya forum that I don't visit two often, not 'Lighting and Rendering'-

Anyway, here's me two c:

From my pov, MRay and PRMan 11 are pretty much on par, feature-wise. Pixar even adopted much of the MRay ray-tracing api for SL, so this is even more true from a programmer's pov. I know that the next 3delight release will also have all of these new PRMan 11 SL functions plus subdiviosn surfaces including all deRose extensions, like holes, interpolated boundaries and arbitrary sharp creases. I took special care -- by providing feedback -- that
it integrates seamlessly with RAT.
Speaking of other RMan renderer's, here's the list:

3Delight (still free)
AIR ($300-$375 aka a "no-brainer")
Aqsis (free)
BMRT (free, still available from other sites than Exluna.com and legal to use)
RenderDotC ($595-$2.995)
PRMan 11 ($5k)

Of course, MRay is not a micropolygon renderer though it is capable of handling a scene that is consisting of geometry diced to micropolygon size.
A Maya user gets one MRay license for free which is a strong point but seldomly is one license enough.
So here we are at the strongest point against this renderer from the pecuniary pov: with MRay you are forced to buy further MRAy licenses. With RMan renderers you have the choice. Let me just tell you that twice this year I worked at small boutiques who owned one MTOR license. I was hired as a RMan TD because they where too afraid to use it w/o one. This btw. brings up another topic: Superstition when it comes to the complexity of (P)RMan -- in particular with MTOR this simply is not true. I teach MTOR courses twice a year at least and Maya artists constantly tell me afterwards that they find (P)RMan much easier to control and understand than e.g. the Maya renderer.
Anyway, to get back to my examples, in both cases one PRMan license wasn't enough. No problem! I installed 20 3Delight licenses on the renderfarm and everything went smooth deadline-wise. Do you have this choice with MRay?
Of course you have to be aware where the featree set of your replacement renderer doesn't match with PRMan (or whatever other RMan renderer you stareted the project with) but that's pretty much it.

A few comments on Shaders.
Shaders are created using SLIM if you own MTOR. So saying RAT is not well integrated simply isn't true. It just circumvents Maya's Hypershade entirely to provide more powerful and hence better access to some functions in RMan SL that rely on a diffrent philosophy.

But even if you use Liquid (and a pre-compiled Liquid is available since two months for Windows now, only not from the sourceforge site), you can use ShaderMan or DarkTree if you're an artist. ShaderMan is free and DarkTree is the most artist friedly shader tree system I ever saw. Plus it comes with an extensive -- to say the least -- lib of preset shaders. A new DarkSim DSO enables DarkTrees to be rendered from any (!) RMan compliant renderer that supports them (I think only Aqsis is currently not in this list).

Ok, last but not least some comment on ambient occlusion. This is called 'final gathering' in MRay and its available there since years. It is also possible since years using BMRT in combo with any of the aforementioned renderers. We used this to render dirt maps in 1997 using BMRT with PRMan but we didn't call it 'ambient occlusion' then.

It is a myth that it is too slow for production in most cases. This is only true if you have characters that deform which requires the solution to be re-calculated for each frame. If you have ridgid geometry like buildings etc, you can use texture projection or use a pre-pass which is even more effeicient if e.g. the pov changes. This is supported by Entropy, AIR and soon 3Delight. I guess RDC will follow suit soon too.

.mm

[gmask]

Originally posted by Mauritius
So here we are at the strongest point against this renderer from the pecuniary pov: with MRay you are forced to buy further MRAy licenses. With RMan renderers you have the choice. Let me just tell you that twice this year I worked at small boutiques who owned one MTOR license. I was hired as a RMan TD because they where too afraid to use it w/o one. This btw. brings up another topic: Superstition when it comes to the complexity of (P)RMan -- in particular with MTOR this simply is not true. I teach MTOR courses twice a year at least and Maya artists constantly tell me afterwards that they find (P)RMan much easier to control and understand than e.g. the Maya renderer.

This is all really good info.. So how easy is it to replace PRman with one of this other renders in connected it to Mtor? I say this without knowing very much about Mtor setup but with a little about setting PRman on a network.

In regards to PRman being easy to understand than the Maya renderer.. why do you think that is? Is it because of the numerous limitations of the maya renderer that force you to use compositing tricks or that it does not have a sophisticated shading language? I imagine that the type of students taking your classes are more inclined to accept learning the technical side of things. Whereas I encounter many users who want to master 3D but have no interest in getting under the hood.

Thanks

Adrian

[Mauritius]

Replacing PRMan is as easy as hacking in one command line into a dialog and choosing 'custom' as the target rendere. Addistionally, you can also use this new renderer to render shadr preview ball in SLIM or just add a secondary shader compiler as target. The greatest problem so far was the path separation via ':' in the RIB search paths used by only PRMan. However, I wrote to Paul Gregory and the 3Delight guys, so Aqsis and the aforementioned renderer support this too now.

Speaking of complexity I simply refer to the number of parameters affecting quality of the resulting image and the time to compute it. Those seetrings that control these issues in a RMan compliant renderer a much more straightforward to understand than in Maya. Only think of the tesselation controls!!! These become obsolete when switching to a RMan compliant renderer.

SLIM certainly takes time to get used to. Its much more like the old PowerAnimator or 3ds Max' shader editor, where you did/do build shader trees, but wouldn't see ther tree structure.
You can see this structure in SLIM, but you can't edit in tree-view. :( Component-wise, after installing all those stuff available via highen3d.com and deathfall.com, there isn't much left you can do with Hypershade -- out of the box --, which you can't do with SLIM. On the other hand there are quite a few things which are much easier accomplished with SLIM and near impossible to do for a newbie with Hypershade.

.mm

[thesaint]

in fact, i would go step further. Thre are abstractions you can create looks and abstractions using code that you simply could not do in Hypershade.

Slim, i find, is actualy as easy to use as Hypershade if you want to get rich (meaning complex and subtle) results. The GUI has really improved over the last few generations of release and the ability to create 'trees' and render swatches at any level to see detail through the shader network means you can follow, pretty intuitively, the network from the most basic color elements to the geometric driven procedural displacements. Speaking of which, i dont believe there is any way to get a geometric driven displacement in MR or Maya, that is, a displacement map that shrinks or distorts its procedural data to comply with changes in physical geometric model, ala Mike's elbows (inside and out) have subtle creases that appear as the arm bends and flexes. Such detail is not modeled, for obvious reasons, but is mapped, linked and driven by Slim. Try that in Hypershade (better get a fresh bottle of asprin first)

If i owned a renderfarm at home, which is getting pretty practical these days, i would renderman purely for the reason put forth by Mauritius. Flexability, Scalability and Functionality (you know you'll never grow out of it, only every into it) and Price, you could do it all using freeware. That is worth the price of admission, i.e. learning how to use it.

[gga]

Sorry. I had to comment, since a lot of misconceptions and just plain wrong information was being discussed about mental ray.
I work at ILM where we daily use both prman and mray and I personally have about 5 years of experience with each package, as both an user, and developer writing shaders... and can tell you stories of pulling my hair with both ;)
This is sort of a list of good and bad points of each renderer.

1) mental ray is a hybrid renderer. As a matter of fact, it has been such since v2.0. A hybrid renderer is new to prman, but it is NOT new to graphics community. Besides mray... lightwave, maya, etc. are all hybrid renderers and have been so for quite some time.
This means that indeed you can use the package only on scanline mode if no tracing is required. Most good hybrid renderers like mental ray, prman and lightwave, can mix rendering scanline with raytracing within samples, even. Bad ones can't or have horrible limitations (maya comes to mind).

2) mental ray3 IS a micropolygon renderer, which is not based on a reyes architecture. The talk of not being able to do driven displacements thru the hypershade is just bogus.
mray is a micropolygon renderer that it is based on an object caching architecture (this is both good and bad). Instead of splitting objects per buckets, objects are split into subobjects (think of doing dicing in screen space vs. world space, I guess). The quality of displacement mapping has little or nothing to envy prman anymore. The object caching is also less subject to patch cracks as prman is. mray, unlike prman, tries to always maximize the memory consumption and keep as many triangles in memory as possible when raytracing (ie. the caching is less agressive). This is not bad, it makes the renderer faster during traces, but yes, when you fire a mray render without specifying memory limits, you will immediately see it try to take over as much memory. Acceleration of raytracing is done by hierarchical grids or bsp, albeit bsp is still the preferred approach.
mray3.1 was the first implementation of this and it did suffer from some bsp creation problems which would spike memory like crazy albeit was usuable already for most projects. mray3.2 beta (not yet released) has addressed most of the issues and solidified the architecture more. Currently, I can already testify that the object-cache architecture is already proving to be as successful as prman's reyes ones in many scenes, more so when raytracing is involved. Still, raytracing of micropolygons IS expensive compared to just scanline rendering or to raytracing objects with good but not micropolygon tesselation. mray still provides the best (read fastest) implementation of this I have ever tried. Over the Christmas break, I raytraced scenes of over 10,000,000 polygons with little issue and under 2.5hs. Still, as we throw more and more complex cases, we do find bugs and issues and we usually report to both vendors as problems they need to address. For example, I recently did find an obscure bug with the caching being flushed too much leading mray to technically render 150,000,000 ( yes, that's 120 million triangles) due to re-tesselation. What surprised me was that as ugly bug this was, the architecture handled it and albeit it took long (16 hrs), it worked just fine. And 16hrs for 120million triangles sounds REALLY cheap to me, actually --- specially considering the code doing most of the re-tesselation was mine and not optimized, and not as good as mray's tesselation code ;)

3) subdivision surfaces. mental ray provides a very comprehensive subdivision surfaces package which is sold as an add-on shader library. It is a silly marketing decision. As such, indeed most translators do not support it. Also, regarding its being sold in the USA, not sure if it could expose mray to legal action by pixar, since pixar holds some rather obvious patents on subdivision surfaces.
The subdivision kit is more complete in terms of subdivision types and in terms of hierarchical subdivision than prman is, but it does not resolve subd's to bsplines as in prman implementation.

4) mental ray's architecture has proven time and again to be MORE flexible than prman's, not less. We usually try to do everything with prman when possible since our pipeline is built with it more in mind and due to having been used at ilm for so long, we also have many more artists familiar with it. But albeit SL is nice and friendly, prman's lack of openness bites us whenever you want to do anything BEYOND SL. Technically, we do have ways (technically and legally) to hack into prman code. However, we have very few people who can do so (technically and legally) and many times we don't have the time. And prman's DSO architecture is simply not mature enough. mental ray suffers from none of that (basically, anyone that can code C can be writing mray shaders in no time and the API is way more ope). As such, when prman fails or is not good at a certain thing, mray usually ends up being used as a complement ( raytracing, blobbies, atmospherics, etc).

5) mental ray's scanline rendering is actually FASTER than prman. Yes, you won't hear this very often, but it is true. Now, this does not mean the renderer is faster, thou ;) Let me explain:

mental ray's scanline renderer is on average about 3 times faster than prman10 and ~1.5 than prman11, when equivalent shaders are used for your average ilm creature (say, our superduper plastic shaders with million of knobs). This is so for preview renders without motion blur.
What DOES make prman amazingly fast during scanline renders is this great fact:

When motion blur is on, performance hit for any prman render is (usually) less than 50% of what it was without.

NO renderer on the market is even close yet to offer this... I know, we usually try them all.
With mental ray (and most other renderers), motion blur implies about a rather constant 500% increase in render times at least. This usually offsets ANY benefits of the mray scanline renderer when motion blur is on. Many places only turn on motion blur for final renders. Not so at ilm. EVERY one of our renders is motion blurred. Usually our supervisors require it. Kind of silly, perhaps. But certainly doable for prman and it is the one feature that makes prman worth its price tag, in my opinion.
And yes, you can do tricks like 2D motion blur to compensate... they just don't look as good, that's the problem.
Now, that's for scanline rendering.
For straight raytracing... prman is 5 to 10 slower than mray. No suprise there. After all, this is Pixar's first implementation of raytracing. It is likely to get better with newer versions.
With raytracing and motion blur on ... well, no miracles here. We usually try not to since it is usually too expensive with any renderer. However, mental ray is still better. As slow as it is, prman is (and I'll be nicely optimistic here ;) 5 to 10 times slower, without even mentioning the other issues about it.
For global illumination ( final gather / caustics ), mray is slightly ahead in most aspects while prman is ahead on at least one. However, speed and quality are relatively comparable and the differences will hardly make any difference to your average user: mray supports caustics and volume irradiance while prman has a more flexible api to work with irradiance caches in shaders.

prman is king in terms of primitives ( mray lacks CSG, built-in blobbies, sphere primitives, no point primitive, etc. and mray's hair is very new and not production proven ). mray's only strength here is support for high order curves and surfaces (as those used in CAD and engineering).

mental ray's basic shadow mapping support is now much cheaper than prman since they can be created on the fly at any resolution with little performance hit. On prman, a separate pass is always required. Prman now supports deep shadow maps which are great for volume effects and more so for hair. mental ray shadow maps are not yet deep shadow maps, meaning hair shadows are still better raytraced right now.

Prman's atmospheric shaders are slow at best and not too great. mental ray shines in this allowing you to create anything from blobbies shaders to very fast raymarchers. The latter package also has some internal handling of in/out volumes to allow intercrossing and travelling thru volumes.

Prman allows stitching of subds to handle displacement without cracks. Note however, that this is only for subd's not for generic patches. mental ray has stitching for patches, but they do not work for displacement.

RIB format is slightly more compact than mray's .mi2 format, but MUCH harder to debug and read, since it is ALL binary, unlike mi2 which is a hybrid format. RIBs are also more compact if you spit them in a hierarchical fashion, making them even harder to read and debug.
RIBs were originally a standard that did not catch on mainly due to Pixar still keeping hold on the copyright and licensing of it. However, in the past few years, other renderers have indeed started using the RIB as an input mechanism. This is both good and bad. It means the standard has already deviated but it also means that if you'd like NOT to use prman, you can switch to other competing products much more easily than before.

mental ray's format is as much a standard as prman's rib is, albeit no other renderer uses it yet as input. It supports shader networks, user data, and geometry instancing which ribs do not. More important, it supports progressive scene changes, for IPR functionality. It also does NOT work hierarchically, which is a big relief.
RIBS parameter passing and attachment of variables to geometry is VERY elegant, and although the same can be done in mray, it is not as nice. Only phenomena is elegant in mray.

PRman works with a SIMD architecture while mray does not. This is the other BIG plus for prman. This allows prman to calculate derivatives for antialiasing shaders trivially, while in mray, the shaderwriter has to do more work for that.
Now, however, that raytracing is introduced in prman11, the benefit of prman SIMD becomes less, since afaik, neither renderer does derivatives calculations beyond the first ray hit (could be wrong about prman, need to recheck the latest one). The only package on the market that was superior in this aspect was the now defunct Entropy, which did allow them.

In terms of documentation, mray's documentation is and has always been poor at best, lacking examples. Prman documentation while far from being great is full of examples and the new application notes for raytracing are simply outstanding.
For both renderers, however, it is still recommended to get the publication books sold.

mental ray integrates seamlessly with a number of products on the market with more or less successful interfaces -- usually more tighly built around the package workflow. Currently 3DMax, Houdini, Maya, Softimage and XSI support it.
Prman's main translator is the one that comes with RAT (irma/mtor/slim combo). Houdini support is native. For max, you need something like MaxMan and for Soft, softman. There's also some very ugly lightwave support. For xsi, there's no option to spit out ribs.

Prman's stability is simply legendary and is no overstatement saying than some of the best programmers in the world have contributed to it. Simply put: what other software product can you say has lasted for over 20 years with little core modifications? When Pixar releases a new feature, it has usually gone thru about a year of testing internally, and it shows. Hard to find bugs in prman.
mental ray's reputation is shaky at best in this area. Many users tried pushing mray to do things it was never meant to do in its old times and to this day, mray still lacks that internal testing that it is so important. As such, whenever mray introduces a new feature, my guideline is to stay away from it until the following revision of the software.

In summary, if you can afford both, do so. Then you should learn what they are good for and not try to use a hammer when a screwdriver is better.
If you can afford one, look at the list of good/bad features above and choose. Yes... something, somewhere will not be possible or be as good or easy to do. Just know what that something is, and avoid doing projects that require that (or find creative cheats).

Finally, if not convinced with comments, try them yourself. Both Pixar and mental images offer demo versions of their products.

[playmesumch00ns]

Liquid has been successfully compiled with MSVC++ 6 and .NET. The project files etc are on the sourceforge page, or at least linked to from the forums. Still couldn't get it to compile myself though.

[stunndman]

thanks gga for sharing insights from your experience and clearing up some prejudices regarding MR that were stated here as facts

i think that MR is a very interesting and capable renderer - and a great addition to maya of course

[thesaint]

I am in screenings most if not all of toady, but gga has certainley opened up some interesting questions that i would like to respond to, and i for one now have some more questions and challenges for him and the other knowledgable people here.

Later.


Great discussion.

"I'm out there Jerry, and lovin' every minute of it"
--Kramer on the subject of going 'commando' in Seinfeld.

[beaker]

>>Also, regarding its being sold in the USA, not sure if it could expose mray to legal action by pixar, since pixar holds some rather obvious patents on subdivision surfaces.

The only patent pixar holds on subdivs are Semi-Sharp creases.

[gga]

Originally posted by beaker

The only patent pixar holds on subdivs are Semi-Sharp creases. [/B]

Pixar to my knowledge used to hold 3 patents about subdivision:
- One regarding scalar fields (as a way to allow more friendly st coordinates assignment and other potential uses)
- One regarding hybrid subdivision schemes where subdivision is done hierarchically with different methods at each step (say, merging Catmull-Clark and Doo Sabin).
- And there was one indeed mentioning semi-sharp creases, but cannot find it anymore listed in the patent web site. probably my fault.

[gga]

Originally posted by thesaint

Trace on | off reply:

What i did a bad point of trying to say was that in MTOR you can actually select the object in the scene you want to have affected.
[/B]

Yes, this is also the case on all commercial raytracers, including mray. You have visible/trace/shadow flags on a per object basis. In the case of mray, these attributes can be propagated thru a hierarchy easily.

Originally posted by thesaint
scanline on| off opengl reply:

I cannot make out what it is saying here.
Ray tracing is a method of calculating the path of a ray of light from the camera back to the light source, while scanline (to my understanding) is a method of breaking the scene into managable chunks of data that are then rendered one pixel line at a time. I have no idea how or why you would turn this off, or what bearing it would have on how the scene shadows, refraction and reflection is calculated, except that all effects the time to render.
Any help on this one is appreciated.
[/B]

Didn't I say mray docs are hard to read compared to Pixar's? They sometimes seem to be written by Germans :D

Basically, this allows turning on or off the scanline algorithm and also allows things such as shadow maps to be calculated with your card's opengl (using opengl for shadow maps is a great idea in theory, but not yet in practice. Albeit faster to calculate, you loose, since you spend too much time sending complex geometry thru the card's as of yet limited bandwidth. There's also some precision lost).
Now... Why would you EVER want to turn off a scanline algorithm?
There's mainly one reason: you have a lens shader (something missing from the RIspec).
The idea is that a shader can be attached to the "lens" of the camera to simulate all types of effects, from fishlenses to actually position the camera randomly on each pixel or sample. Lens shaders are called before render begins and displacements are run. Since the camera can stop being a pinhole perspective camera then, all your usual concepts from shading rate to texture filter area become unknown to the renderer. As such, the scanline algorithm should not be used at all (nor any view dependant tesselation).
This is useful to us to do texture baking, by isolating each object and moving the camera around it following its st parametrization. This allows baking textures all around the object perfectly, and not just from a single view and is particularly useful to bake ambient occlusion for expensive objects that will not move.
Nowadays, mray offers another form of texture baking with a new type of shader, but it has one or two issues regarding consistent image size and control across multiple patches. As such, we still like our brute force method better.

[visionmaster]

very interesting thread :)
is there a connection btw maya 4.5 and mentalray 3.1 ??
i want to render gi fur and hair in mental ray but it dont work.
do you think it is possible with shave and haircut?

[beaker]

>>very interesting thread
is there a connection btw maya 4.5 and mentalray 3.1 ??

The MR 1.5 connection for maya uses MR 3.1.

>>i want to render gi fur and hair in mental ray but it dont work.
do you think it is possible with shave and haircut?

You never want to use GI on fur. The rendertimes would be beyond slow(see you next year). Shave and Haircut uses their own renderer and then composites the hair into the scene(like how maya fur works).

[beaker]

Originally posted by gga
Pixar to my knowledge used to hold 3 patents about subdivision:
- One regarding scalar fields (as a way to allow more friendly st coordinates assignment and other potential uses)
- One regarding hybrid subdivision schemes where subdivision is done hierarchically with different methods at each step (say, merging Catmull-Clark and Doo Sabin).
- And there was one indeed mentioning semi-sharp creases, but cannot find it anymore listed in the patent web site. probably my fault.

Looks like pixar has been busy these last few years with patents. Larry Gritz had always mentioned on the newsgroups that pixar only had patents on semisharp creases. So I never checked on any others. One of those patents you mentioned was just granted a month ago.

[gmask]

>>>BMRT (free, still available from other sites than Exluna.com and

I can't find it anymore??? Is it still downloadable from somewhere???

[tropistic]

BMRT was pulled as part of the Pixar vs Exluna lawsuit settlement.

Jay

[gmask]

Originally posted by tropistic
BMRT was pulled as part of the Pixar vs Exluna lawsuit settlement.

Jay

But it was still avaible for awhile I guess it got yanked???

[tropistic]

Nope, it was pulled simultaneously with the press releases of the lawsuit settlement if I recall. Therefore, any copies still floating around are *not* legit.

Besides, BMRT hadn't been updated for almost a couple of years, and it'll never be updated/supported now. It's probably best at this point to explore some of the other free (or relatively inexpensive) renderman compliant renderers available, if you're so inclined :)

Jay

[thesaint]

BMRT v 2.6.18 NT just happens to be sitting on my hard drive here... hint.

ALSO
Copies that are floating around ARE legit, just not formal distribution. The app is still legal and very much in use out there. It is not open source, so development has stopped, and i might also take this time to recommend www.3Delight.com
They just posted a new updated version that is very good.

[beaker]

BMRT is still available on a few sites:

http://www.cirkus.fr.st/CG/cg.html
http://herakles.zcu.cz/education/mpa/index.php

[Gonzo The Great]

Why don't you just use Maya's own renderer? It's built in! :scream:

[gga]

Originally posted by beaker

You never want to use GI on fur. The rendertimes would be beyond slow(see you next year).
[/B]

Well, sort of.

Depends on what you are looking for when you say global illumination with hair. I see 4 possibilities of what you want to see:

a) You expect each strand to contribute ambient bounce light to other strands. Yes, you are likely insane.

b) You expect strands receive the contribution of the skin surface (head). Relatively easy. Just run a pass creating and saving an irradiance (or final gather, depending on software terminology) cache to disk without the hair and then re-use that same cache in a beauty render with the hair in it and with its shader querying that cache. Increase in render times should be negligible (well, at least it should be as bad as any standard irradiance pass).

c) Having hairs occlude light from the skin (or add bounce light to it) is also somewhat doable, but will depend more on your pipeline and in being somewhat smart.
The idea is that you should split the render also in two, using a raytrace pass and a beauty pass. For the raytrace pass, your hair is invisible to primary (camera) rays but visible to traces.
By making the hair invisible you will soon realize you can cheat quite a lot, making hairs thicker and less denser than those hairs used for the beauty pass and that will improve your render times quite a lot. This can be easy if you have a hair generator that uses guide splines to create new hairs by interpolation (if you have many guide splines even just rendering those may be enough). Lacking such a pipeline, you can achieve a workaround if you have time to model a rough mesh representing the hairs to give you a decent approximation.
For short hair in particular, you may also cheat by only generating hairs around the edge of the head, and not in areas of high volume and concentration (where shadowing will likely occlude any ambient contribution anyway).
In either case, you also should pay attention to your raytracing acceleration scheme and settings for it if your render allows tweaking them, since "good" settings for tracing the scene without hair are likely not optimal for tracing with hair.
We currently have 3 movies in production with characters being rendered doing something akin to this.
For long hair, we experience about an additional 1/2 hour of additional render times for closeups for an equivalent occlusion pass without hair, and it does force a normal scene that would not need cache flushes to require them.
For shorter hair, it is much more doable, requiring just an additional 5-10mins and usually no flushing.

d) You want b) AND c). Unfortunately, irradiance caching will usually only give you one or the other. If you need both together, you need to do more than one bounce, leading to issues and a big slow down in any package.
This is still one of the big limitations when it comes to irradiance caches (and straight montecarlo tracing, too).

[Jones_Stork]

I have a question (esspecially to gga) as there is a lack of good documentation for mRay ( tutorials!). Help docs tells you that you CAN do animations not HOW to setup mRay for that.
And that is a main problem. Our company is rather small, we do not have render farms, only a couple of PC`s. And some game intros to create. We would say: "the hell with GI and FinalGathering, it takes too long and there is a LOT of artifacts" but we have done similar tests (just few boxes and sky dome or indoor scene with one light) with finalRender and Brazil for 3DsMAX and tadaaa! we have smooth, clean animation in realiable time (4-8min per frame with 320x240 resolution wich is enough for our purposes). But...we have to use Maya because we just jump to it from MAX and cannot come back.
So the questions is... Is it possible to have such render times in ANY renderer in MAYA (having all GI/FR on)? Do you know how to set up this free mRay to Maya 1.5 for animations, wich as you said is the newest version of it? Sorry for such a disturbance in your disscusion (you can reply on my mail if you like) but you seems to be a right person to ask and I already started to learn Maya after a few years of working with 3dsMax. And it really makes me mad when program with such abilites, used in top high-quality movie production studios is lack of most important thing like simple, easy to use, realistic renderer wich is available for not very popular, game-purpose package, like 3DsMax..:annoyed:

thanks in advance

[stunndman]

So the questions is... Is it possible to have such render times in ANY renderer in MAYA (having all GI/FR on)? Do you know how to set up this free mRay to Maya 1.5 for animations, wich as you said is the newest version of it?


are you rebuilding FG for every frame ?


when program with such abilites, used in top high-quality movie production studios is lack of most important thing like simple, easy to use, realistic renderer


well, mental-ray must be something like the 2nd most frequently used renderer in feature films

asking for something simple and easy to use for high-quality output is a contradiction - and in the end simple and easy depends on your insight into the product

[mayadini]

Great thread, too bad that you high-end guru guys in the industry seldom occupy any user discussion forums nowdays:shrug:

I also have a question (which by the way is not at all intended to drive away gga and other valuable voices from this thread/forum at all, if too busy no need to answer:):
I'm sure you guys at ILM use your own proprietary translator from maya to mray (or maybe Alias's one is what you use since they seemed to have worked hard to inegrate it as much as possible), but I don't seem to be able to find any way in telling mray to use micropolygonal rendering of geometry (nurbs, since we don't have the subd plug and mostly use patch in pipeline). I just started experimenting with mray and I can't wait to test your statement that its "by far the fastest micropolygonal raytracer you've seen"! Personally lack of micropolys in any other-than renderman compliant proggie was the biggest drawback to any renderer I've tried...
Another more technical/newbie question:
How do you guys set up patch-modelled texturing for prman since there are no shading switches in RAT like in Maya (I've been told by your collegue that you use patches also mostly not as much subd but he didn't know exact method for textures since he's a modeler). All this stuff is no rush for an answer. Any will be greatly appreciated, though , thanx in advance!

[gmask]

>>>Do you know how to set up this free mRay to Maya 1.5 for animations, wich as you said is the newest version of it?

There is nothign to setup.. you can render an aniamtion with it using batch render and you can issue a render from the commandline. However it is not designed to allow you to use it freely like the Maya standard renderer. You have to have a license of maya installed on each machine and issue the commands manually on each machine.

[mayadini]

also, forgot this, when you said that prman renders subds as splines did you mean as NURBS? If so does it mean that it creates its own "patches" on the fly? Otherwise how does it go around the standard nurb drawback of the old "only 2 holes possible per patch" problem? Thats very interesting stuff, sorry about "newbieness" i'ma just a humble conventional "artist" forced to 3d but love tech stuff :)

[Jones_Stork]

>There is nothign to setup.. you can render an aniamtion with it using batch render and you can issue a render from the commandline.

Again, my question is how to set up GI and FinalGathering to achieve animation without artifacts, not how to render animations. I`m newbie in Maya, but not THAT much ;-)

>are you rebuilding FG for every frame ?

I do not know... I think "yes" (lock samples - off?). And I think it is a main problem. As far as I know one have to rebuild the photon calc. in Final render to achieve changes of lighting regardles to the animated objects movement. And usually, in simple scenes, it works fine. But I can`t understand the workflow of mRay because its documentation tells nothing about that issue. So if I AM rebulding it -is that good or wrong? Do I have to lock samples/photons somehow? (how?)
Is building photon calc. is for animations with only camera moving or can I use that for i.e. character animation? Do I have to increase amount of photons? Accuracy? What is a "golden centre" to achieve good quality and realiable rendertime?

Please, anyone, answer... I know everybody hate newbies, but who other than you should teach? Sometimes a few words can stop a newbie from hunging himself.:argh:

"Help me Obi-Wan Kenobi! You are my only hope..."

[gga]

Originally posted by Jones_Stork

Again, my question is how to set up GI and FinalGathering to achieve animation without artifacts, not how to render animations. I`m newbie in Maya, but not THAT much ;-)

>are you rebuilding FG for every frame ?

I do not know... I think "yes" (lock samples - off?). [/B]

Lock samples has little to do with Final Gather. It is a control mainly used to tell mray how the sampling in general should behave (and it is mainly noticeable in area lights' shadows). Having few samples is not good, but sometimes you do have to make the painful sacrifice of trading quality for speed. When you do, areas without enough samples may exhibit a noise or pattern.
Depending on the situation, the lack of quality samples may go less detected if the sampling pattern is locked or altered from frame to frame (thus, lock samples).
Lock samples is sometimes good if the camera moves little or nothing. The viewer will see the pattern (of an area shadow for example) as just another element of the texturing of the object. However, if your camera moves, the static nature of the pattern will be obvious --it will swim-- and as such lock samples off is preferred (which will give you "noise" instead).

Anyway, back to your question...
To rebuild Final Gather you should turn on, not surprisingly, the "Final Gather Rebuild" switch (in the final gather folder, just above the folder you found lock samples). If you toggle it off, the final gather file will be saved and kept in one of the maya subdirectories (you can specify the name of the file too). mray will reuse it for future calculations and renders (only creating a new one if it cannot be found). If you alter your materials or lights, you should always recalculate the final gather or else you will be misled to be looking at some old final gather which will likely be incorrect.
Final Gather Rebuild is neither good or bad.
For say architectural fly thrus where nothing but the camera moves, you will likely not want to rebuild final gather on every frame. This will improve the calculation and as the camera moves, new samples will keep being added to that same file, making subsequent frames even faster to render.
However, as soon as something other than the camera animates from frame to frame, you do need the switch on, since mray cannot reuse the potentially now "wrong" calculation of the previous frame.
To control the quality of final gather, you need to play with the min and max controls mainly, while experimenting also with the number of final gather rays. These min/max values are a minimum and maximum distance in world units (it can also be changed to be in screen pixels instead, but not from the maya2mr gui as far as I could tell).
What do the values do? Well, If a ray hits geometry near a sample that is less than min distance from a previously calculated final gather point, it will automatically use an interpolation of what it had calculated so far to approximate the result. If the ray hits between the min and max distance allowed, mray will use its own heuristic to decide whether to interpolate or not. If the sample hits farther away than the max distance, mray will need to do a new sample for sure.
By default, the values of 0 tells mray to use its own heuristic to try to choose good values for the min and max based on your scene's bounding box, which will usually give okay results for scenes with objects roughly the same size and not too far apart (read: hardly ever ;).
You may think that changing the maximum value would be more useful, right? Well, not in my experience. Changing the minimum value seems to usually be the more important control where you will see the quality/speed difference trade-off right away.
Every time mray cannot interpolate a sample, it will be forced to shoot a number of rays around the hemisphere above the intersection point. The average of the colors of the objects hit will determine the color of the sample. If there are many objects around the point with hugely varying colors, and you did not shoot enough rays, your average may not end up being too accurate and could potentially change from frame to frame, leading to noise in the animation.
Increasing the number of rays will thus help decrease noise (the more detailed geometry you may have in your scene, the more likely the # of samples will need to be increased from the 1000 default). Using a smaller min distance will decrease the "splotchyness" that can sometimes be produced due to the interpolation, and which is mostly noticeable in animation.
Most of this sampling happens in a pre-pass process (during which you will see nothing being shown and just a small progress bar), but some new samples may also be taken while the render progresses.
Needless to say, the more rays and smaller min size you use, the longer you will be kept waiting.
You will likely need to experiment and render 4 or 5 frames with different values until you get the hang and the "eye" for it to see what's the good tradeoff in speed/quality for your scene.
mray has a switch to turn on a debugging of the final gather points (so that they are shown on top of the image being rendered as green/red points, but alas, it does not seem to be exposed in maya2mr yet either).
To control the intensity of the final gather contribution maya materials have an irradiance and an irradiance color control in them which is independent of the diffuse intensity of the object. You can think as one of the controls being how much "bounce light" it can emit, while the other controls how much it can potentially receive. The good thing is that the irradiance color (the receiving) does not need recalculating the final gather solution (you can turn off rebuild). However, changing the irradiance one always needs recalculating.

Finally, when rendering with mental ray in general, it is recommended you turn on the verbosity of the render to show Progress. This will print a lot of stats and progress to the shell window or DOS console you fired maya from. If using windows, you may need to fire maya with a batch file instead of the usual icon to get such a console. This can help you when you are a newbie since sometimes you may set some setting wrong and send mray to calculate forever. The stats while not explaining all, may give you an idea of what part of the render mray may have trouble with. Paying attention to the stats is also the difference between a very fast and a terribly slow render. Particularly the bsp size, and the number of rays and samples taken can tell you a lot about how good your settings were. The triangle count, alas, in mray3.1 has gone missing, unless you turn verbosity really high. Overall, the two main issues that can make mray be slow is excessive number of triangles in your objects usually as a result of over tesselation and wrong settings in the bsp (and btw, maya2mr's defaults on these two are not that great... they err on the side of caution more than in the side of speed).

Whether mray's final gather will match the speed you want, it is impossible to say without knowing more about your scene and setup.
Overall, the times you quoted from other renderers seemed to me at first glance somewhat excessive for your resolution, but it all depends on the number of textures, shaders, cpu and many other factors, I guess.
My point of comparison is film res frames taking between 20 to 40 mins (on our single proc 1.5/2Ghz machines) for a complex creature with no hair and a relatively simple environment. My speed difference largely being on how close the character was in frame (and how well settings were adjusted, of course ).

If you have not yet seen it, you can see a sample movie file of a user interacting with final gather in real time in a simple scene and download the same sample tutorial file from the A|W website for studying:

http://www.aliaswavefront.com/en/products/maya/mentalray.shtml


BTW.... note that final gather has little or nothing to do with photons. photons are used for caustic effects and for global illumination where more than one bounce may be needed.

[gga]

Originally posted by mayadini
Great thread, too bad that you high-end guru guys in the industry seldom occupy any user discussion forums nowdays:shrug:


Well, noone is either high-end or low-end anymore other than how big your budget is.
And most certain noone is a guru in our industry, unless perhaps your last name is Catmull or Blinn.

Originally posted by mayadini

I'm sure you guys at ILM use your own proprietary translator from maya to mray (or maybe Alias's one is what you use since they seemed to have worked hard to inegrate it as much as possible), but I don't seem to be able to find any way in telling mray to use micropolygonal rendering of geometry (nurbs, since we don't have the subd plug and mostly use patch in pipeline). I just started experimenting with mray and I can't wait to test your statement that its "by far the fastest micropolygonal raytracer you've seen"! Personally lack of micropolys in any other-than renderman compliant proggie was the biggest drawback to any renderer I've tried...


Well, the biggest drawback to me has always been motion blur, that's why we all love prman.

Yes, we have our own translators but from propietary software to mray and prman. And we use all sort of propietary shaders.

To use micropolygons in mental ray the object (or scene) needs to have an approximation statement in it. This can be a displacement approximation or a normal approximation. displacement approximation has the advantage it only kicks in if a displacement shader is used, but it can be slightly slower... specially on a shader tree with a noise texture.
The micro polygon approximation that is somewhat equivalent to prman's shading rate of 1 in mray would be "fine view length 1".
In maya2mr, which I don't have in front of me, you can find it if in some weird location. I believe it was near the pull-down for render globals, where you will see a submenu for mray listing approximations and custom text.
Approximations is the one you want. From that, how to get fine view length 1 should be easy. mental ray also uses a min/max to impose a per-object limit on the tesselator, too.

Originally posted by mayadini

Another more technical/newbie question:
How do you guys set up patch-modelled texturing for prman since there are no shading switches in RAT like in Maya (I've been told by your collegue that you use patches also mostly not as much subd but he didn't know exact method for textures since he's a modeler). All this stuff is no rush for an answer. Any will be greatly appreciated, though , thanx in advance!

Yes, we use patches and not subdivision. Mostly now due to software accumulated since T2 or so, more than for any other real technical reason.
We currently do not use RAT as we use prman with custom lighting software, albeit we are looking at it again these days.

[stunndman]

as usual very useful insights gga - so far i haven't played with the min value much - i'll have to explore this one further

and by chance anyone knows what the Precompute Photon Lookup setting in the Final Gather section does ? - i can't seem to find any information on this

If using windows, you may need to fire maya with a batch file instead of the usual icon to get such a console.

mental-ray for Maya will output progress messages to the Output Window which is opened by default as a separate window at startup

[stunndman]

some more information on MR - the Softimage XSI documentation on mental-ray's GI and FG features - http://www.iro.umontreal.ca/~roys/softimage/html/shade/GIcaustics.html

[gga]

Originally posted by mayadini
also, forgot this, when you said that prman renders subds as splines did you mean as NURBS? If so does it mean that it creates its own "patches" on the fly? Otherwise how does it go around the standard nurb drawback of the old "only 2 holes possible per patch" problem? Thats very interesting stuff, sorry about "newbieness" i'ma just a humble conventional "artist" forced to 3d but love tech stuff :)

No, I said b-splines. B-splines are just NURBs wanna-bes :) They look somewhat similar but they lack weight controls on the vertices and, more problematic, cannot have arbitrary knots or parametrization (meaning it can give you headaches in texturing). The control cage will also look somewhat different to a NURBs to obtain the same resulting shape. Their biggest benefit is that they are likely a little cheaper to evaluate. They can be found and used in other software like Softimage.
And yes, prman does create its patches on the fly as it renders the subds. That's why the Pixar guys will get somewhat offended when anyone with what's basically a polygonal rounding tool calls themselves as having subdivision surfaces. You can find the description of their method if you search for Tony DeRose or if you attend some of his presentations like those in the Subdivision Courses of Siggraph.
Pixar's method was the first and required of reyes. Today it is no longer the only way to get patches, nor it is likely the best approach to render them anymore.
The way maya renders the wire frame and the opengl of subdivision is likely the best approach to render subds today since the evaluation is direct, but it is also a patented one (by A|W) and has not been extended to Pixar's weights due to another patent (by Pixar). Maya subds sometimes may not seem fast to users, but they technically are. Maya just went a little crazy with features for also allowing subds to be hierarchical (where you do pay a big prize for it).
In any case, if nurbs is what you want... you can get them and avoid all patent issues if you use a method by Peters (which is the one maya4.5 uses when you click subd->nurbs), which, btw, is the most efficient method known so far to indeed turn a subd into a group of nurbs with proper continuity.
If you don't care about continuity at all, then there's Lightwave's MetaNurbs (which is unpublished, but relatively trivial to understand how it works). It obtains smoothness only during the polygonization of the objects by merging edges and recalculating their normals. Its surface does not entirely respect the shape of Catmull-Clark subdvision and since it cares not at all for continuity, it is even faster than all of the above for display (but the polygonization with the fix of edges process makes them slower for a renderer).

[CJcuervo]

This serves for some great reading material here fellas.

once again....off topic..and for a silly question as well...


I'm new to maya (and MR) and was just wondering what the differences are between mr1.5 for maya and mr3.1 stand alone?




CJ

[gmask]

Originally posted by CJcuervo
This serves for some great reading material here fellas.

once again....off topic..and for a silly question as well...


I'm new to maya (and MR) and was just wondering what the differences are between mr1.5 for maya and mr3.1 stand alone?
CJ

The standalone can handle renders from any program that can generate the proper file for MR. The MR pLugin for Maya cannot parrallel render. IE when you ahve the standalone renderer installed on your network machines you can use the whole network to render and image from inside maya which has it bovious advatnages wehn it comes to tweaking redners.

[Phearielord]

hey

this has been an amazing discussion, by far the most informative read I've had in a while.

Anyway, somebody mentioned earlier something about a demo version of PRMan.
Is that true??
For I haven't heard anything about it.

Somewhere I've read that you can control the level of subD's of your character in Renderman, that they can have different levels of subdivision thoughout your character.
How does work, if true?

gga, recently ILM signed an agreement with Mental Images that they will use MR in their movies.
What does the agreement entail?
That you will render solely with MR (unlikely), or simply that you guys will help develop it through what you guys use it for, or what?

keep this discussion going, very interesting
:thumbsup:

[beaker]

>>Anyway, somebody mentioned earlier something about a demo version of PRMan.

You have always been able to get a 2 week demo license of Prman. Just call them up to request it.

>>Somewhere I've read that you can control the level of subD's of your character in Renderman, that they can have different levels of subdivision thoughout your character.

Sounds like your talking about hierchial edits in subdivs like maya's subdivs have, which Prman does not have or support.

[a23]

hi gga,

just a quick question.

since you prefer PRMan for speed when using motion blur, how do deal with shadows?
especial selfshadowing!!!

i always get artifacts, because shadowmaps don't motion blur. in PRMan 11 one could use the pseudo ray trace area lights, but that seems expensive and deepshadowmaps do extra computation one wouldn't really need.

i know that blurring shadowmaps to death and use insane samples give you a motion blurred shadow, but as i said before it gives me headaches with selfshadowing.

any tips and tricks?

AndY

[mayadini]

Great info gga, I managed to get the micropolygons (or microtriangles as mentalImages refferes to them in the manpages). One thing's for sure, there would be no way for me to find it on my own, thanx lots!
Also your other explanations opened my eyes on things yet again so thanx for that too.
"you can get them and avoid all patent issues if you use a method by Peters"
Peters who?
Btw, is that how maya renders their own subds? By converting them to nurbs and than tessalating them? How it goes about texturing than (unless planar which i understand is geometry type independant in its nature). I know Maya renders everything as triangulated polygons, but with this new technique you've explained to me with Mental ray and aproximations in order to get the mycropoligonal render going it does make sense to have subds batch converted to nurbs at every frame (with baked animations perhaps or whatever technique) and render that instead of paying for that subd plugin :))
Thats some cool stuff. So, is this technique by "Peters" used by any comercial (not in-house proprietrary) renderer?
thanx

[gga]

Originally posted by stunndman

and by chance anyone knows what the Precompute Photon Lookup setting in the Final Gather section does ? - i can't seem to find any information on this



It gives you slightly better quality in shadow areas by merging final gather and photon mapping.
If you are not using photon mapping, you won't care.

Here's a much better introductory set of tutorials with simple scenes.

http://www.3dluvr.com/intercepto/maya/tutorials/mray_thegathering/hdri.htm

These tutorials show you how to mix both techniques (however, in my opinion, photon mapping is still very expensive to calculate on a per frame basis. final gather alone is much more reasonable).

[a23]

hi, my guess is that it's this Peters...


jörg peters:

http://www.cise.ufl.edu/research/SurfLab/papers/

AndY

[gga]

Originally posted by mayadini
Great info gga, I managed to get the micropolygons (or microtriangles as mentalImages refferes to them in the manpages). One thing's for sure, there would be no way for me to find it on my own, thanx lots!
Also your other explanations opened my eyes on things yet again so thanx for that too.
"you can get them and avoid all patent issues if you use a method by Peters"
Peters who?
Btw, is that how maya renders their own subds? By converting them to nurbs and than tessalating them? How it goes about texturing than (unless planar which i understand is geometry type independant in its nature). I know Maya renders everything as triangulated polygons, but with this new technique you've explained to me with Mental ray and aproximations in order to get the mycropoligonal render going it does make sense to have subds batch converted to nurbs at every frame (with baked animations perhaps or whatever technique) and render that instead of paying for that subd plugin :))
Thats some cool stuff. So, is this technique by "Peters" used by any comercial (not in-house proprietrary) renderer?
thanx

Sorry.
Jorg Peters, from SurfLab. The paper I refer to is his Siggraph 2000 paper for Patching Catmull-Clark surfaces. He also has previous papers for patching other subdivision schemes (DooSabin, for example), which I definitively recommed reading first since those are presented more clearly. The Siggraph paper tried to fit it all in two pages, and while not really a complex method, the algorithm and structure he uses is a little cumbersome. The explanation is very subtle, in particular to the nomenclature and structure. One of his students also has an implementation of the algorithm, too... as sample C++ code, that uses a lot of advanced C++ libraries like CGAL.

http://www.cise.ufl.edu/research/SurfLab/papers/

As I said, if you have maya4.5, you already have the feature. I'm not too familiar with maya's renderer, but I assume by now it may be using the same technique.

Texturing of subds is roughly similar to meshes. You assign coordinates to the vertices of the control cage and the subdivision algorithm interpolates them. The assignment of the coordinates is usually the headache. You can use projections as a start and then some programs like maya allow manual manipulation of the coordinates. It is tedious work but usually less than trying to get continuity with patches.
The interpolation of uvs during the subdivision process can be linear or it can also follow the subdivision rules. The two methods are useful for different purposes. Depending on the method used, you may also be able to texture the cage at lower subdivision level and generate the corresponding coordinates for the original base level (that's kind of one of the pixar patents).
There's still no magic automatic solution to assign texture coordinates. Many software allow some form of automatic assignment, but these usually will break the model in small sections that are not friendly to users or 2d packages and will usually still require manipulation.
The best more automic approach I am aware to assign coordinates is a paper presented (was it from Secret Lab?) a couple of years ago, where a model was cut into pieces using guides set by the user and then each one was textured with a relaxation algorithm using maya dynamics. Don't have the reference anymore and a little fuzzy now on the details.

Overall, converting to patches would not be required anymore if you could use maya's direct evaluation method and create a tesselator that way. But since the method is patented, no renderer can without paying A|W for it. Converting to patches and then tesselating is still usually faster than using some form of localized view dependant mesh subdivision (which is what you'd have to do otherwise... the micropolygons do have to come from somewhere) so most high end renderers will take that route.

BTW... instead of using the approximation editor, the maya2mr render globals in the advanced mray section has overrides for tesselation that effect all objects. That's usually easier when dealing with multiple objects (and more akin to prman's shading rate 1 for all objects).

[gga]

Originally posted by a23
hi gga,

just a quick question.

since you prefer PRMan for speed when using motion blur, how do deal with shadows?
especial selfshadowing!!!

i always get artifacts, because shadowmaps don't motion blur. in PRMan 11 one could use the pseudo ray trace area lights, but that seems expensive and deepshadowmaps do extra computation one wouldn't really need.

i know that blurring shadowmaps to death and use insane samples give you a motion blurred shadow, but as i said before it gives me headaches with selfshadowing.

any tips and tricks?

AndY

Well, not much you can do really if moblur is your problem. But somehow I doubt that's your issue.

If you are getting shadow map artifacts in prman10 or 11 with gaussian blurs, this may be due to a different reason: a bug slipped by where the filter area was being calculated incorrectly. I reported the issue and Pixar already has an upcoming fix. For the time being, using a box filter, which is inferior and not usually what you want, is the current workaround and should give you better results. Previous versions of prman did not suffer from this.

The other typical headaches with shadow map artifacts is due to using a low resolution or a bad shadow bias, both of which can also create self-intersecting shadows. Check the shadow bias or switch to prman's better approach of using midpoint shadow maps. Stick to 512 to 1K maps for video res, and 1K or higher for film when possible and use some blur with them. Shadow maps in prman also need to be a power of 2 and square ratio. Your scene converter may not enforce this, perhaps?

If all fails and indeed motion blur is the issue, then yes. you will need to use either raytracing or deep shadow maps. Deep shadow maps should be faster, but not as good for motion blur (have not tried them for this yet). Raytracing one light won't kill you, but don't go crazy with making all lights the same way.
Also, for raytracing with motion blur, you will need to turn on an option in the rib. prman11 by default will not perform motion blurred raytracing.

[CIM]

Anyone else notice how poor (buggy) the Mental Ray for Maya connection is? It seems like Maya crashes alot when it's translating shading networks for rendering. There are tons of weird quirks I've found, as well.

Is there some kind of workaround anyone has figured out concerning the shading networks thing? Maybe we'll just have to wait until Alias/wavefront decides to release a working version of the connection. :rolleyes:

[gmask]

Originally posted by CIM
It seems like Maya crashes alot when it's translating shading networks for rendering. There are tons of weird quirks I've found, as well.


They say that most Maya shader nodes are supported but maybe you are using the ones that are not??

Anyway check the Known Limitations file provided with the install.

file:///C:/Program%20Files/AliasWavefront/mental%20ray%20for%20Maya%201.5/doc/html/UserGuide/known_lim3.html

I see that you are a LW fan.. is somebody twisting your arm to use Maya and MR? :rolleyes:

[Phearielord]

Most maya shaders are supported, but for best results, try and use the MR shaders as much as possible.

[beaker]

Originally posted by CIM
Anyone else notice how poor (buggy) the Mental Ray for Maya connection is? It seems like Maya crashes alot when it's translating shading networks for rendering. There are tons of weird quirks I've found, as well.

Is there some kind of workaround anyone has figured out concerning the shading networks thing? Maybe we'll just have to wait until Alias/wavefront decides to release a working version of the connection. :rolleyes:

A/w took over some of the development from MI about a year and a half ago. Oringinally a/w had nothing to do with it. MI started on the MRtoMaya back when maya 1.0 came out. By the time maya 3 came out they had finally added support for many of maya's shaders, but it still didnt have any support for GI/FG. It took them till the 1.0 release fbefore GI even showed up(4 years into development). You really shouldn't be blaming a/w for this since they inherited 4 years of MI snail pace work. The speed has increased by leaps and bounds since a/w took over. In the beta around maya 2.0 with MR2.1 it was 2-3x slower than maya's renderer with any moderate amount of geometry.

[gmask]

Originally posted by beaker
The speed has increased by leaps and bounds since a/w took over. In the beta around maya 2.0 with MR2.1 it was 2-3x slower than maya's renderer with any moderate amount of geometry.

I spent some quality tim etoday with Mr for Maya and am pretty satisified so far with it's spped.. I didn't turn everything on all at once yet but individually everything feels pretty nice. I did get a couple crashes here there that seemed to be script relted but only needed to restart Maya to clear it.

[gga]

Originally posted by Phearielord


gga, recently ILM signed an agreement with Mental Images that they will use MR in their movies.
What does the agreement entail?
That you will render solely with MR (unlikely), or simply that you guys will help develop it through what you guys use it for, or what?



I have not read the agreement so I know as much as you do.

However, make no mistake about it.
Prman is still a key component of our pipeline and we currently see no reason for that changing.

Albeit the use of mental ray at ILM is now more widely known, the truth is that we have been using it for quite some time now, ever since the first Men in Black. What started as an experiment in raytracing a couple of shots back then has now grown to become just another component and toolset of our pipeline, as tcsh scripts or python is.

At this point in time, the plan is to stick with both renderers moving forward, since we see their strengths and weaknesses and find each other complementary.

We do have a good relationship with both mental images and Pixar and we indeed request new features and provide feedback in terms of how things work (or don't work) in a real production environment.

Both vendors have mutual respect for each other and are now very aggressively developing their products not only looking at what they can do for next year but also to what they can do long term to push rendering to new levels.

[Jones_Stork]

Thanks for your answers, it helps a lot. But that is not all I want to know... :p

I have such a message when I render with MR to Maya:


"RC 0.2 warn 082016: lens shader runs in scanline mode, it must not modify the origin or direction of the ray"

What the hell is that?

I found that the quality of my animation depends on the angle of the light. But even if I use settings that suppouse to be high, I have artifacts. I will post one frame, so meaby you will tell me what i wrong with that scene.

[Jones_Stork]

here is the image. I pointed where artifacts are ussually shows.

P.S. I use 50000 photons in one spot light

[semuta]

This is an awesome thread, I can't beleive how much useful information is getting thrown around here, so, heres a question.
how do you use the mental ray custom environment in maya camera atributes? ie we're trying to use the mi envronment sphere as documented in the maya to mr docs. I'm new to mental ray and don't know if I'm missing a step or something.

thanks:scream:

[caulfield]

this free mental ray...does it leave a water mark? And wheres do I find it?

[stunndman]

Originally posted by Spike
this free mental ray...does it leave a water mark? And wheres do I find it?

no / aliaswavefront.com

[tripNfall]

This is a great thread.

OK, since there has been a ton of good information
uncovered, I thought someone could uncover my mental
ray question of the month.

I’m very impressed with mr and want to start using it as
my primary render. However I still need to learn more.
So, why is it that when I use a poly plane or a dome to
light a scene I get nice occlusion (darkening shadow areas)
but when I use a spot light (either with of without photons)
I don’t get the occlusion and sometimes it seems to brighten
the areas where I think it should darken? I have read this
question in a lot of post, but have yet to see an answer.

Also, are there negative lights in mr? I have tried entering
a negative intensity number but it does not seem to work?

Thanks

[ThE_JacO]

Originally posted by tripNfall

So, why is it that when I use a poly plane or a dome to
light a scene I get nice occlusion (darkening shadow areas)
but when I use a spot light (either with of without photons)
I don’t get the occlusion and sometimes it seems to brighten
the areas where I think it should darken? I have read this
question in a lot of post, but have yet to see an answer.


occlusion, or FG or RFOcasting or whatever buzzword you want to call it with basically scatters rays from objects to produce illumination.
the very moment you are using a light you are not lighting with FG anymore and therefore can't rely on it to "stop" the light and create a shadow.
with lights you have to rely on shadows of some kind, raytraced or mapped


Also, are there negative lights in mr? I have tried entering
a negative intensity number but it does not seem to work?

Thanks

try a negative color and it should work.
without going into how MRay deals with lighting data, let's just say a negative intensity is unconvenient, negative color channels intensity should work and allow color component controls in the light subtraction, way better then a single intensity value (that acts as a multiplier sort of thing).

[Phearielord]

I recently had the oppertunity to play with PRman 11 recently, and I must say, it was REALLY fast.
Though if you get too complicated with the ray tracing stuff, it could reallyslow down quick.
But it has some very nice tools, and from the quick glance I got at everything, I must say that IMO it is much better than MR.
Once you get to know it, the possibilities are endless.
Suddenly MR doesn't quite seem that cool anymore.

[Jones_Stork]

I must say that MR didn`t gave me what I really need. It is far too dificult to achieve smooth GI in animation especially in comparision to 3dMax Brazil, which is as simple as a hammer and really fast... Meaby it`s only that MR Connection to Maya 1.5 which is so piece of crap, not a standalone version, but i found it not interesting... Or meaby I`m too stupid too use it? But with lack of good documentation and nobody to give me the tutor.... what can I create with it? :hmm:

[Jones_Stork]

One revision to what i said in previous post. I recreated the same scene in Brazil and I must say that it renders slower then in MR. But at least I do not have artifacts and it took me 5 min to setup it properly.. But I will do some extra tests to MR....

[enforcer2k]

Jones_Stork: Totally agree with you man. I've use Max native GI, Vray and Brazil (Vray and Brazil are both max plugin renderers supposedly being ported over to Maya), and they are all way easier to use and achieve stunning artifact free results than Mental Ray. I really really tried to master MR, docs, faqs and all, and I still couldn't get consistent reliable results. I hardly needed to look at the manual for vray or brazil to get up and running.

Just so you know it's not you. the MR connection is too complicated to tweak for lots of reasons... What is awkward though is that Softimage renders tend to be smoother and more on target (softimage uses mr too, but a tighter integration). Vray and Brazil still come out tops though, but I give vray the extra nudge above brazil for the blazing speed. The Vray implementation of GI is so fast you are tempted to include it in everything you render :D.

Anyway since this is a Maya forum let me redirect... MR is a good step forward especially since it's raytracer is faster than Maya's and it can parallel render across pc's, so presuming you have 10pc's to toy around with, it can make for a lot of crazy fun. I think the render dialog for Mental Ray in Maya 'NEEDS' to be simplified it's too complicated as is. To illustrate what I mean, getting rid of gi artifacts in MR for maya means tweakinganywhere from 3 to 8 or more variables (fg min rad,max rad,samples,gi sample,photon emission,etc etc...). This is STUPID! Especially for artists who are primarily animators or modelers but who no less deserve to have a powerful rendering solution. In Vray, or brazil, you only need to focus on 1-4 variables, mostly no more than 2. My 2 cents.

[beaker]

>>Just so you know it's not you. the MR connection is too complicated to tweak for lots of reasons... What is awkward though is that Softimage renders tend to be smoother and more on target

Why would you want less options? Go use LW if you want less controlls. Maya to MR just needs more extensive docs and better presets so people can quickly setup a no brainer without messing with the settings.

[stunndman]

softimage's mental-ray is the same as maya's mental-ray - i have been using a setup were maya's MR 1.5 plugin used a networked softimage's standalone mental-ray node - and the buckets rendered locally were just the same as the ones rendered on softimage's mental-ray node (surprise)

i wouldn't call it stupid to have the freedom to tweak whatever parameter influences the rendered output - sure they could build in some logic which does this for you depending on scene scale etc. - and brazil or v-ray must have something like this although that's just a guess

concerning the documentation - i think this is definitely the weakest point - not that it is not complete or anything - but it is very technical which is great in the first place but they have omitted the chapter for us dummies - you really have to dig deep to get the big picture - if you are using PRman you definitely don't do that without a book or two - so get a book if you feel lost in the raw technical documentation

[enforcer2k]

stunndman: To clarify, I know that the Mental Ray Engine in Maya and Softimage are the same, what I meant is that the interfaces into the engine for softimage yield better results with less tweaking than Maya's (With more intuitive varibles and fewer of them at that). My opinion, but I've used both extensively.

I stand by it being STUPID to have so many exposed variables. Lightwave, Vray, Brazil or any GI capable renderer that comes to mind are just not as daunting (I would mention Renderman 11 but its GI and raytracing suck suck suck). It's like having a car with 4 steering wheels, one for each tire, stupid. Other renderers much more automated, user friendly, and more capable at getting you the results you want 'quickly' with solid consistency. I understand advanced uses like the extra control, this control should not come at the price of making a piece of software hard to get into, unintuitive, and nearly impossible to master. Sometimes increasing the sampling to remove artifacts from Mental Ray images quadruples the render time, and optimizing all those variables to get reasonable rendering speed can be daunting to say the least. It is hard enough to become a good CG artist, but must the hurdle of 'numerical mastery' become another obstacle along that path?

[stunndman]

Originally posted by enforcer2k
stunndman: To clarify, I know that the Mental Ray Engine in Maya and Softimage are the same, what I meant is that the interfaces into the engine for softimage yield better results with less tweaking than Maya's (With more intuitive varibles and fewer of them at that). My opinion, but I've used both extensively.


from the XSI 1.5 documentation (which included mental-ray 2.x) the render options for mental-ray (a screenshot of the render options dialog for mental-ray in XSI -> http://www.iro.umontreal.ca/~roys/softimage_xsi/files/docs/html/tutorial/images/illuma10.jpg) are identical in count and naming to maya's render options dialog (except for the photon volume accuracy and photon volume radius which might be due to the older mental-ray version) - enough of that

i totally agree that mental-ray is hard to handle and i have spent some weeks now to explore this ***** and i am quite frustrated - but i tend to attribute this to my lack of knowledge - it's like somebody gave me a ferrari and i just don't know how to drive it

i am really wondering how v-ray and brazil achieve this "no-tuning-necessary-to-make-my-radiosity-lit-scene-look-great" effect - does this mean that mental-ray's radiosity algorithms suck ? - maybe someone can tell us the whole story

[enforcer2k]

Originally posted by stunndman
i am really wondering how v-ray and brazil achieve this "no-tuning-necessary-to-make-my-radiosity-lit-scene-look-great" effect - does this mean that mental-ray's radiosity algorithms suck ? - maybe someone can tell us the whole story

It's not no-tuning, but very-little to moderate. Frankly I think mental rays age is showing in comparison to other renderers. VRAY's object displacement is in my opinion way ahead of mental ray when it comes to speed and memory use and setup ease is even better. It think it comes down to better written software made more accessible. I mean come on the 40% of the mental ray docs are fine, but you have to be a Physics or Math major to read the rest of that crap! It's hilarious! Who do they think was going to be reading the docs? Artists or Science majors or programmers???

[gmask]

Originally posted by enforcer2k
I mean come on the 40% of the mental ray docs are fine, but you have to be a Physics or Math major to read the rest of that crap! It's hilarious! Who do they think was going to be reading the docs? Artists or Science majors or programmers???

Personally I haven't found the docs to so oblique.. in regards to Maya though .. it is a little confusing at first where to find the neccessary controls. However I have learned how to use MR in Maya fairly quickly and even though I have had a try at XSI my familairity with Maya made it much easier for me to figure things out with MR.

[enforcer2k]

gmask: Welcome to the 5%. ;)

[beaker]

>>I stand by it being STUPID to have so many exposed variables.

In other words you just need a "make art" button? If your above statement is true then you shouldnt be using maya, because there are obviously way too many controlls (2-3x as many variables for everything over 3dsmax and LW). Maya is obviosly just way too complicated and should be dumbed down for the masses. :) :) :) (just a little humor there)

serious mode on :)
AW should make many presets for the the masses that want to use the software to create fast results, but in no way should they remove controlls/functionality. Thats just plain stupid. Just because Vray/brazil/max hides many of the controlls from you(and make many assumptions from the user), doesnt mean everyone else should follow suit.

>>(I would mention Renderman 11 but its GI and raytracing suck suck suck)

It doesnt suck, its just slow. This is the first incarnation, Im sure speed will come soon.

[enforcer2k]

>>This is my last post as I find topics concerning software comparisons usually turn into religious debates. Control is 'good', complexity is mostly 'bad'. They are not the same ideas. Because a process is very complicated does not mean it gives you the control you want. Maya is a great package which while being slighlty complicated benefits the user by steering a good amount of this complexity towards giving the user more control.

Mental Ray however, in my mind, represents the bad side of complexity. When an artist looking to get simple quality renders with a clear purpose in mind in forced to juggle 6 or more interlinked variables to achieve his goal. The simple conclusion is this, goals being the same, an artist using vray or brazil will get the results he or she desires in less time with less tweaking. These two packages like I said will probably be making their way to the maya platform soon and I think will make more accessible alternatives to using MR in Maya.

[Jones_Stork]

In my opinion MR to Maya need some presets. I created the same scene in brazil and I just click Production Level 3 preset button, made some tweaks in two parameters and TAADAAA!! But I found out that this scene push the limits of any renderer, and even brazil found it dificult (1h/frame 320x240) But, not even single artifact. I will make another test with Maya and meaby I really need to try some higer values to get rid of artif.
I must say that working in Max and Maya is like flying Cessna plane comparing to Boening 747, but I think Maya is lack of simplicity in many simple actions... Sometimes an artist need a simple click-and-go solution to preview his work before he will spend two months tweaking render nodes to achieve his goal.

[playmesumch00ns]

can you not just use MEL to save preset options?

[playmesumch00ns]

Originally posted by beaker
>>I stand by it being STUPID to have so many exposed variables.

In other words you just need a "make art" button? If your above statement is true then you shouldnt be using maya, because there are obviously way too many controlls (2-3x as many variables for everything over 3dsmax and LW). Maya is obviosly just way too complicated and should be dumbed down for the masses. :) :) :) (just a little humor there)

serious mode on :)
AW should make many presets for the the masses that want to use the software to create fast results, but in no way should they remove controlls/functionality. Thats just plain stupid. Just because Vray/brazil/max hides many of the controlls from you(and make many assumptions from the user), doesnt mean everyone else should follow suit.

>>(I would mention Renderman 11 but its GI and raytracing suck suck suck)

It doesnt suck, its just slow. This is the first incarnation, Im sure speed will come soon.

...ya they said when prman 11 was release that GI wasn't final, or even 1.0, just there as a placeholder for future developments more than anything else.

[playmesumch00ns]

Originally posted by stunndman
from the XSI 1.5 documentation (which included mental-ray 2.x) the render options for mental-ray (a screenshot of the render options dialog for mental-ray in XSI -> http://www.iro.umontreal.ca/~roys/softimage_xsi/files/docs/html/tutorial/images/illuma10.jpg) are identical in count and naming to maya's render options dialog (except for the photon volume accuracy and photon volume radius which might be due to the older mental-ray version) - enough of that

i totally agree that mental-ray is hard to handle and i have spent some weeks now to explore this ***** and i am quite frustrated - but i tend to attribute this to my lack of knowledge - it's like somebody gave me a ferrari and i just don't know how to drive it

i am really wondering how v-ray and brazil achieve this "no-tuning-necessary-to-make-my-radiosity-lit-scene-look-great" effect - does this mean that mental-ray's radiosity algorithms suck ? - maybe someone can tell us the whole story

Brazil and VRay are written for MAX users, hence they make a whole host of assumptions about the way you want your scene rendered. That's fine if you're looking for a "make cool" button, but what if you want to render for feature-film production and need to tweak the look down to the last detail? (not that you'd be rendering whole scenes with GI at film res, but that's another matter)

Perhaps more relevant is that MR was written a looooooooong time ago, back in the middle ages when lots of people doing computer graphics still did have maths or programming majors. Brazil and VRay, while no less competent, have simply been written with different users in mind. If you can afford all the different tools, why not hire someone to tell you what all the different options mean? You could even do a course in rendering theory!

On the SI/Maya comparisons, MR's been integrated into SI for donkeys years (conincedentally because the SI renderer was a pile of poo :rolleyes: ), This is really the first proper version of MR for Maya (discounting the MR connection plugin...), give them a chance!

[IRONIC3D]

Hi all, really sorry to bring this post up again (I just found it :) )

I have a question to gga but I'm afraid it's not maya related :beer: but it's a rendering related.

gga, you said that you do try other renderers. I wonder if you have any info regarding Mantra.

How does Mantra differ from Mental Ray and PRman in specific?

cause for some time now I thought that Mantra is close to the way PRman works :p

Also, how bad is Lightwave Renderer is Technically (just testing my luck here)

Thanks in advance

I3D

[CIM]

Also, how bad is Lightwave Renderer is Technically (just testing my luck here)

It's not that LW's renderer is bad (it's actually damn good), it's just that it is using older technology than these standalone renderers.

[pixelmonk]

Originally posted by CIM
It's not that LW's renderer is bad (it's actually damn good), it's just that it is using older technology than these standalone renderers.

Agreed. I think that was someone's attempt at trolling LWers into the Max "make everything for me" button versus the complexity of MR and PRMan debate.

[graphiouz]

Often GI renderers i see is way to exaggerated, with colorbleed from hell, and way to light on the dark areas,
and many of them often mixing daylight with indoor light,
often fake GI turnes out to be not fake picturewize:rolleyes:

nice thread:buttrock:

[Novakog]

Now I obviously don't know as much about renderers as a lot of people on this thread, but I will give my impression of Mental Ray and how things such as Final Gather are slower than in say... Brazil.

Someone else said that Mental Ray was programmed "back in the day" where most of the people doing CG were science or physics major, and herein lies what I think the problem is (this may be completely stupid and wrong, if so, sobeit).

I was thinking that maybe Mental Ray is so slow because it's "physically accurate." I've never heard any renderer other than Mental Ray actually say that, and I'm not sure whether it's true for any other renderers. Can someone who knows a little bit more reflect on this, that maybe most other fast raytracers use non-physically accurate equations that yield equally good results with less tweaking and faster render times.

Anyway, sorry again if this is completely wrong... render theory isn't exactly my area (well, I'm pretty bad at everything, but render theory is probably my WORST area).

[Phearielord]

That's a pretty interesting point there, Navakog :thumbsup:

I would really like to know the answer to that myself :)

[Mauritius]

I was thinking that maybe Mental Ray is so slow because it's "physically accurate." I've never heard any renderer other than Mental Ray actually say that, and I'm not sure whether it's true for any other renderers. Can someone who knows a little bit more reflect on this, that maybe most other fast raytracers use non-physically accurate equations that yield equally good results with less tweaking and faster render times.

Anyway, sorry again if this is completely wrong... render theory isn't exactly my area (well, I'm pretty bad at everything, but render theory is probably my WORST area).

It's completely wrong. The difference is rather that you have to read a few books to successufully use a renderer like MRay. That's why the big studios have rendering TDs.
Regarding physical accurate equations: since MRay has programmable shading, the equations are as inaccuarte or as accurate as the resp. shader TD thinks they must be.

The differences between MRay and st. like finalRender are huge, to say the least. It starts with stability, goes over features and ends with customization (e.g. programmable shaders). Comparing both is no point. But to answer your implict question: MRay can with no problem be as fast as finalRender while producing similar images.

A personal sidenote from myself (being a lighting & rendering TD):
I read Cebas' support forum from time to time and had to use finalRender under production circumstances in a company I recently worked for.
My very personal opinion on this issue was and -- after this job -- even more is that Cebas is full of hot hair and unkept promises to their customers around a very mediocre "renderer", encapsualted in a big marketing bubble. Some people (like that studio I worked at) payed for fR stage-1 over a year ago but fact is that it hasn't been shipped a of now stage-0 is full of bugs.

If you want somebody to elaborate more deeply on the why's and how's, repost you question on the highend3d rendering theory mailing list which is read/hosted by the authors of all those renderers. Thomas Driemeyer, lead MRay developer is there and also Edwin Braun, proud father of finalRender, also the splutterfish guys lurk on this list etc. etc.

Cheers,

Moritz

[Novakog]

Alright, so I was wrong (like I said, it's probably my worst area).

You said that you would have to read a few books to make good, fast renders with Mental Ray. Any recommendations (other than the obvious "Rendering with Mental Ray")?

[Jones_Stork]

Yes, meaby Mray is good when you have dozen people to manage it for single render.:thumbsdow
As with the whole Maya workflow. Why, when I`m pay for expensive program I have to pay more to dozen of people to use it? Why I didn`t received for such a price a good books and tutorials to understend that program? Why I have to pay again for "good books"?
Mray is good renderer? I`m an artist not a programmer. When I`m byuing a tool i want it to be ready to use, with all needed information how to use it.
Sorry guys, I`v tried to learn Maya, it is almost a half of year I use it everyday and I`m still think that MAX with Brazil is better (for single artist like me) and more I work with Maya I`m more sure of that.

[Mauritius]

Stop complaining. I'm in this business for eight years now -- I still learn everday.
Maya from -- my experience -- takes about nine months to become fluid with -- production wise -- for the average newbie 3D artist. If you have no one to ask and are purely self studying at home, it may take a lot longer as you don't face production related problems often enough.
Renderers are very complex pieces of software. You can wrap any high-end renderer with a few black box default shaders that have 50 knobs to dial and make them as "artist friendly" as e.g. Brazil. But fact is that the bar is put higher in VFX work every year. People are constantly asked to create new spectacular effects no one has seen before. This requires tools built as open as possibe so people can do things with them no one imagined when hose tools where actually designed.
Foe example, without programmable shading, PRMan would never have become the renderer of the feature film industry. Take any black box renderer from the late 80's and try to render a decent, state of the art, image with them -- it's impossible as the algorithms they use in their shaders simply don't cut it anymore. But I can still get pretty close to today's standard using my old 16bit PRMan for Windoze with some of today's cutting edge shaders compiled for it.
The more open a system is, the more knobs it will have for someone to dial. More knobs means higher complexity and a longer way to go for someone to learn how to use this system. This is true for apps like Maya and Max, but in comparison, Max is alot "flatter" than Maya, so your statement doesn't come unexpected. Start talking about Houdini ...

Cheers,


.mm

[kiaran]

I read this whole damb thread from begining to end and found it very interesting the whole way through. The guys at Pixar and ILM really helped to clarify a lot of questions I had regarding PRman and Mray.

Personally, I've used BMRT, Mray for Maya, and the new Max 5.0 light tracer/GI solution. I have to say that I liked them all very much. Its seems to me that all this anti-complexity towards Mray could very easliy be remedied with a much more intuitive collection of presets. I believe that there is a user friendly renderer in Mray just waiting to burst out. A|W, if you are listening, focus on making your software more user friendly and intuitive (ala Max). I don't believe that this necessarily has to be at the cost of extensibility either.

Just look at what they have done with Paint Effects for example... On the surface they appear to be a very high level 'make beutifull button' sort of tool. However, dig a little deeper and you'll find that they have quite a fine level of control.

I don't know if any of what I just said will make sense to anyone. I guess what I'm trying to say is that ease of use does not have to be attained at the cost of extensibility.
Now it's time to get back to work and stop reading insanly long forum threads. ;)

[Novakog]

I agree with you kiaran, Mray could be made more user-friendly without taking away all the deep features it has via presets. However, Mray really does have presets (PreviewFinalGather, Production, etc.) The only problem with these presets is that Mray settings are often dependent on the scene. For instance, most people I suspect use something a couple thousand FG rays for a scene. In my case, however, the measurements of my scene are in centimeters, so my scene is huge, and I use 20-30 thousand rays, but it works the same as using 2-3 thousand on a scene that's a tenth as large. A|W should write some presets that automatically set settings based on the scene.

The presets should also be feature-based. For example, under the Final Gather tab, should be a drop down menu that says Final Gather Quality: "Very high" "high" "medium" "low" and "very low" (like what the default Maya renderer has for anti-aliasing) but as I said before, the actual physical numbers that each FGQ setting assigns would be dependent on the scene itself.

You know, someone who knew their API and/or MEL could program something like this as a plugin or MEL script.

[dwalden74]

A|W should write some presets that automatically set settings based on the scene.


if all that you mean is setting some MR render global attributes based on the scene´s current units, that is quite easy to do with basic MEL knowledge.

-david

[Novakog]

Well, not current units, more like size (in the current units), but yes it should be quite easy to do, I just don't know MEL at all. Again though, the more important thing is the preset aspect (my 2nd paragraph) of it, and I'm not sure whether you can do that through MEL.

[orvski]

Hey guys. I am one of those people learning Maya on my own at home and using forums like this one to see whats happening out there and the issues and many things.

I just one to say thank you very much for the people who have contributed to this thread. Its been very informative and have inspired me to start downloading a renderman compliant renderer and get a renderman book so I can start experimenting.

Thanks again. I know its an old thread, but I felt like replying for all the valuable information ive gotten. :)

[enforcer2k]

I'll tell you why many people complain about Mental ray for Maya simply --- DOCUMENTATION!!!!!! It is very poor, bad, and often NON-EXISTENT for man Mental ray functions. So unless you shell out some more cash for several mental ray books and videos, you will have a hell of a time understanding it. But once you do, Mental Ray becomes a good friend. I must say I am dissapointed in Alias for the lack of documentation.

Fine. They revamped the online access to the maya manual in 5.0 and made it fast and searchable but the information you are looking for is simply not there. How do you tune caustics or final gather or GI??? hoe do you apply motion blur in MR and what do the different settings mean??? How do you setup DOF for mental ray in Maya??? How do you make custom shaders??? This BASIC INFO is not there!!! Ridiculous! You can buy the alias video on how to use Mental Ray, but I've seen it and you will only learn a little!!! So that's it.

Renderman Artist Tools for Maya on the other hand has very good documentation with lots of examples, but price wise and feature-wise it does not make sense for a HOME user or SMALL BUSINESS unit to think of buying it.

So Alias handed us a door with 4.5 and 5.0 ... But the kept the key to themselves :shrug:

[CIM]

Lets hope the poor docs. and the poor integration are fixed/improved greatly in Maya 5.5.

[NUKE-CG]

I agree, the documentation on Mental Ray is horrible.. simply horrible.

Is anyone else buying the mRay for Maya DVDs at *removed link, the author is a scamming asshole* ?
Seems to be a lot of good information on that DVD.

[xenoid]

Hi everyone,

First of all, thanx to all those who've contributed to this thread so far! It's really amazing! So much info packed into this thread alone... :thumbsup:

Got a few questions related to rendering and shaders programming on my mind now, as a beginner in rendering and shaders... Hope all those with experience and knowledge in this area to help enlighten me :D

1. Other than rendering using the integrated mental ray in Maya, are there any other ways to render with mental ray? As in, other than selecting all those stuff in Maya and render scenes, are there other methods of offline rendering like command line, etc? I read some stuff about using Visual C++ to do it, but got no idea how to. :shrug:

2. Am I wrong to say that the current renderers like mental ray and Maya renderer are software renderers? Is there any way to render them like hardware renderers to utilize the power of GPUs? I've read up stuff abt Nvidia's Cg shading language as the way for hardware rendering. What is the correct way to integrate Cg into mental ray/Maya rendering if possible?

Hope these qns dun sound so newbie till no one's willing to answer me. Really wish someone can enlighten me.

[NUKE-CG]

Gosh, quite an old thread you pulled up.

Can I just warn people about my previous message above that I posted over a year ago. That DVD is a scam, please do not buy it!

To make sure I'm going to kill the hyperlink.

[Dr. Ira Kane]

1. You can use mental Ray Standalone to render .mi files (you export your scene to .mi from maya).

2. Yes they are software renderers and I don't think you can use them as hardware renderers, you need pure hardware render to do this ( if I'm wrong please correct me ).

bye bye :)

[beaker]

MR can use Cg hardware shaders but last I knew they had serious aliasing/sampling issues.

[xenoid]

Dr. Ira Kane: Thanx for your reply. :) So the integrated package inside Maya does not have the command line function to render .mi scenes? :shrug:

beaker: I see. But how do you integrate the Cg hardware shaders with MR? Sorry tried to find info on it but can't get anything solid on this. Care to guide me to anything you know? :thumbsup:

[Dr. Ira Kane]

Dr. Ira Kane: Thanx for your reply. :) So the integrated package inside Maya does not have the command line function to render .mi scenes? :shrug:

You can do it only with standalone version, the integrated one in maya can only render maya's scenes from command line.

[xenoid]

Thanx Dr. Ira Kane! :)

[beaker]

beaker: I see. But how do you integrate the Cg hardware shaders with MR? Sorry tried to find info on it but can't get anything solid on this. Care to guide me to anything you know? :thumbsup:Just search for "Cg" or "Harware Shaders" in the maya online manual and it will pull up a bunch of info on the subject.

[xenoid]

hi beaker,

thanx for the tip, but these are not really what I'm looking for. I'm looking at how to integrate Cg shading language into the mental ray shading language so that the rendering process can be speeded up. Thanx anyway! :)

[beaker]

Yea, I know, it's in the maya manual. There are lots of docs in the maya manual about doing hardware shader calls in MR shaders and such.

[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.