Render engines for Maya

#1

I’m doing some research for a TV show, we are in the process of figuring out what to render it in. It is a high profile show, full CG with dense meshes, displacements, fx etc.

We’ve used Mental Ray before but want to look elsewhere. Renderman or Vray are the two big candidates otherwise. Can anyone shed some light onto how these three compare in a big production?

My view on things right now:

Mental Ray: Free, slow and fairly buggy still? Haven’t looked at it in Maya 2012 yet but I hope they fixed more bugs than they introduced in MR. Big talent pool.

RenderMan: Fast, great for heavy meshes. Not as big talent pool, especially in TV? Pricey, but needs to be compared with render times etc.

Vray: Faster than Mental Ray at least, this doesn’t seem to be used much outside feature work or architectural work. Why is that? Smallest talent pool? Gets results quickly imho.

I’m not sure about learning curve on RenderMan, Vray was great - pretty much plug and play.

Any input would be great, thanks.

#2

Hopefully this thread will stay free from fights :surprised :cool:
Obviously there are a few more renderman compliant renderers other than just RmS but I’m currently testing MR/Vray/RmS for my own needs but essentially looking for hassle free work-flow.
They’ll all have strengths and weaknesses, RmS is probably your best bet for a large production but you’ll also need a few good TD’s and leads to ensure it runs the way you want it.
(well, ok that’s the same for any pipeline)

For all of it’s issues MR does, like you say have a fairly large user base and there are lots of work around’s for issues, plus ‘Mental Core’ could be a saviour for some.

I think you’ve probably pin-pointed the main things issues though.

This’ll be an interesting thread :slight_smile:

(p.s if you need any lighting guys in the uk already … :wink: )

#3

It’s job search? I’m ready for to fly on England. :slight_smile: ahaahhaa
Anyway with mental core mental ray inside maya is great. I hope AD incorporate this passe system inside maya. vRay is push the botton et voilà or something like this. It depend on how is great the experience.

ciao

#4

Oddly enough, Nickelodeon uses mental ray for TV animation. It was also used on Gnomeo and Juliet and Tim Burton’s 9. Zoic used mental ray for years (Battlestar Galatica, V, etc.) Recently switched to Vray for cost, but performance is the same. Render times here rarely go above 3 hours and average about 1-2 hours a frame with mental ray. That’s not particularly slow. You just need someone that understands what they are doing. Yes the talent pool is larger, but not necessarily better when it comes to “don’t do that”.

3delight would get you Renderman functionality at lower cost. Smaller talent pool. A Renderman dev could help with it since the animal is nearly the same.

Renderman would have a larger talent pool than 3delight, but honestly, to get the most out of Renderman or mental ray, a developer or three would be worth their weight in gold. It’s a framework for you to make your own. That’s the best option really but requires more overhead in the beginning.

Vray is very inexpensive and actually does have a decent sized talent pool, but it’s not very flexible and occasionally noisy. Similar issues with that talent pool when it comes to “don’t turn that on” or relying on presets. I believe 2.0 also has an improved SSS shader. Training is easier to come by for Vray so it’s a bit more democratic than Renderman for learning.

#5

We have run many jobs through Vray, both features and commercials.

Tron Legacy, Real Steel, Xmen, and also, many car commercials we have done here at DD.

Vray gets 80-90% what you need quickly. There are no noise problems when used correctly.

Now, Vray does have some drawbacks in that it does not have the same amount of capability as Mental ray in particular cases like fluid shading, particle rendering, postFX etc. But, its motion blur kills anything in comparison with mental ray as well as its DoF.

Its been a staple here at DD going on 3 years in maya and much longer in max.

I can’t recommend it enough.

#6

Im also using vray in production, so far didnt have any probl, all my work was rendered with vray, for hair/fur stuff u can use 3delight and thats the pipeline i use…

like PerfectLine said, they have best experience in highend production with vray, im also close to them coz im with vray from beta stage :wink: too, but for the motion blur and DOF u can do that in post production…

cheers

#7

Mental Ray: Free, slow and fairly buggy still?

You got the last part right. Buggy in Maya, still in many areas. But slow? It’s the fastest at what it does, due mostly to the vast ways you can optimize and speed up your renderings. Vray’s good too though - both rock from a timeframe perspective. I’d hardly call mental ray free though. Where are you getting it for free? Sure it comes with Maya, etc., but if you’re going into a production environment you’ll likely want Standalone and some kind of render-server setup. Nothing free about all that!

Vray’s a great way to go if you need simplicity and ease of use. You’ll get better renders quicker, but without as many options as MR. Renderman… I wouldn’t even look into it, unless you don’t care about raytracing or realism. It’s also awesome at what it does, but lacks many of the features the raytracers have.

#8

i say go for renderman for maya. its very very fast and easy to use without a support team.

its true mental ray is faster for hardcore raytracing effects. i recently did a candy commercial and was struggling with highly glossy refractive and reflective SSS displaced multi layered shader at 4 in the morning and decided to switch to MR and i got it rendering fast by 8 am…but i coudnt render with moblur or DOF which i like to do in 3d

however that may have been operator error cause im still learnign RfM and theres not that much learning materials unlike all the mental ray er…tutorials that exist.

for general ray trace stuff renderman for maya is fast enough on glossy reflections and kills for anything with particles and displacemt and sss and hair. i find every single time on many different systems when im doing particles in mental ray, i turn on motion blur and maya crashes…ive seen the waterfall nparticle examples and i cant do it cause of the crashing… but id probably choose MR immediately if i had to do another hard candy commercial for a picky director. u just got to plan motionblur and DOF in post

obviously a lot of people have overcome these problems here, but renderman for maya is a f r e a k i n joy to use. its orders of magnitudes faster for many types of projects than other solutions

#9

This is the stat that boggles my mind - "

In fact, Pixar’s RenderMan has been used on every Visual Effects Academy Award Winner of the past 15 years, and 47 out of the last 50 nominees for Visual Effects have chosen Pixar’s RenderMan to produce the highest quality special effects possible"

I know its marketing, but that is a crazy stat.

Anyhow, you will probably be successful no matter which of those three you choose. Its an abundance of riches these days.

Edit: I own Prman for Maya and there is plenty of learning material in the manual, forums, and user webistes. There is also FXPhd and Escape Studios tutorials.

#10

At the end of the day, you need a team that knows the renderer very well in order to get the best out of it, whether that’s speeding things up, work-around’s or making pretty pictures.
Even Maya’s software renderer can be capable in the right hands :cool: (not me)

One thing though, please promise us that you’ll let us know what you ended up going with, and what the critical reasons for that decision were, it’ll be interesting to find out :thumbsup:

#11

Agreed about the marketing stuff, a quick search shows how much fluff that is. And Avatar was mostly done with PanaRay, from what I’ve read, with only the beauty passes done in PRman. But it’s still awesome - I’d love to be in a studio using a different technology. One that actually works how it’s supposed to, allegedly! I only use MR because the boss won’t spring for Vray just yet, myself! Someday when I’m all grown up though… someday I’ll get to use or even see Renderman!

#12

ive seen most of those except the escape studios ones but i didnt find they went too deep particularly with renderman for maya although there seems to be more material for RMS.

finding information for example on cgtalk or youtube or vimeo on how to incorporate RIBboxs in RfM or if its even possible without the pro system is alot harder to come by than finding info on how to use MR proxies. i just did a search on vimeo for renderman tutorials…3 videos came up and one of them was mine lol…pages and pages for MR

the pixar forums are pretty good tho.

lol on the other hand maybe that difference represents the fact that people dont have so many problems in renderman…shit works…you dont see people desperately asking why it crashes when trying to IPR SSS

#13

I have used RfM and mental ray a lot and I can tell you one thing, they both suck. I haven’t used vray much, but I am sure it sucks too…

What ever you get you will have issues. Having the knowledge to get around the issues is #1. Better the devil you know.

These are my views

RfM:
Great for Motion Blur, Particles, Displacements

mental ray:
Mia shader is awesome, cheap if you have floating licenses of maya, iray(if your adventurous)

For you I would lean towards RfM if your budget allows. You can get great images out of any of the three though.

#14

I have used RfM and mental ray a lot and I can tell you one thing, they both suck. I haven’t used vray much, but I am sure it sucks too…

haha, totally agree. I used RMS studio 1 & 2, both where horibbly buggy, I can see why Animal logic/Framestore/Weta/Double Neg all use there own renderman front ends.

I worked on a animated tv show, we where rendering 30mins of animation a week. We went with Prman and MayaMan, and a lot of custom pipelining. Worked pretty well, had full 3d motion blur and depth of field. If you do go down the Renderman route, get some really experaiced TDs, makes a hell of a lot of differnce. Almost none of the lighters had renderman experaince, but most picked it up after a few days.

#15

Honestly, for your arch viz work I think PRMan is always going to be the distant choice, especially if it’s Renderman for Maya. Forget mia_material_x, you’ll have your hands full stacking Blinns on top of each other. I really hope some of the physically plausible shaders trickle down into RfM because it needs modern shaders badly (as well as the enhancements from RPS 16, including progressive raytracing).

That said, despite the hassle of baking I’ve gotten way too accustomed to flicker-free color bleeding and the super-cheap motion blur. It’s addictive. The downside for the original poster is that like others have said, you’ll need RMS and some TDs to really make it sing.

#16

depends on how big the project is. renderman is very stable, but as mentioned, youd need a couple of td’s to get the most out of it, plus as mentioned, youd be using maya’s native shaders unless you scripted your own in rsl, though, there are loads of rsl shaders on the net, but most of them are ancient and look very dated.

as for vray, its very proficent, but im not sure how well tested it is in a production environment within maya.

mental ray is way better than it used to be, it can render pretty much out of the box without too much pain. ide say stick with mental ray.

#17

I’m not an animator so I can’t give the same level of advice as others here but what I can say is this: V-Ray support is awesome. When you have a problem, it’s not unheard of to see it fixed in a nightly build within a week or two, depending on it’s complexity. Good luck seeing a mental ray bug fixed within a couple years and, while mental core and others are encouraging, having to involve a third party in a tool with a lot of shit broken at the bottom isn’t appealing. What happens if those developers suddenly find it’s not worth their while to support the tools? Open sourcing it isn’t going to help you much without someone to crack with a whip when stuff breaks.

The talent pool for V-Ray is large and, since the workflows are basically the same between Max and Maya, someone with more experience lighting in V-Ray for Max wouldn’t take too much time getting up and running in Maya if they had even a basic knowledge of it’s workings.

I learned V-Ray for Maya pretty much entirely from V-Ray: The Complete Guide, which is for Max.

#18

The talent pool for V-Ray is large and, since the workflows are basically the same between Max and Maya, someone with more experience lighting in V-Ray for Max wouldn’t take too much time getting up and running in Maya if they had even a basic knowledge of it’s workings.

One could say the same of mental ray. But I totally agree with you about it being a broken, steaming pile - at least its implementation in Maya. When you spend more time bug-hunting than you do rendering, from a purportedly “finished” product like Maya, something’s wrong!

Looking forward to Vray this year, still!

#19

It all depends on your needs.

RendermanStudio is great, but its expensive and if you need raytracing, its much work. You can get a good performance out of renderman, but its not an easy task to build a pipeline based on renderman and you have to if you are building structures for a tv show. But its great for flickerfree indirect lighting, motionblur etc.

Mentalray is reliable and of course you have to know it very well to get most out of it, but it can handle huge amounts of geometry if you do it right as well as vray, much more than renderman.

From the programmers point of view you get a good support for renderman. The support is terribly for mentalray because you have to ask autodesk if you have a mentalray problem and because autodesk has no idea what mentalray is, they give it to mentalimages, that can take months before you get a useless answer.

But even if the normal support for vray is great, the support concerning developers is horrible, even worse than autodesk. The sdk is badly documented, the examples have almost no comments at all.

Normally you have to do coding work for a large production to get the best performance. And it can be cruical if you encounter a showstopper and get no support at all.

Developing shaders for renderman and mentalray is quite easy now. Renderman shaders have a disadvantage because they are monolitic and you cannot simply plug other nodes into a compiled renderman shader.

#20

@Haggi OMG, best description yet. :bowdown:

True! Keep in mind that some of these things are changing. It’s probably a year out or so. But Autodesk is changing how renderers are exposed in Maya. This means changes for plug-ins as well as how mental ray is handled.

This is 1000% true and why I mention flexibility in a renderer. I expect Vray 3.0 will begin to offer more in the form of a production renderer than is currently available. Having to shoehorn things into the render is somewhat painful unless you have a huge renderfarm or an army of compositors.

If I had unlimited funds, I would go with Renderman or mental ray and a small brigade of developers to make it mine.

The problems experienced in here (CGTalk) are more related to artist issues and usability in most cases. So most would benefit less from having a more robust API.