Renderman vs MentalRay - a quick rendertest


#1

Hi boys and girls,

I just wanted to let you know, that I did a quick and dirty comparison of MentalRay for Maya and Renderman for Maya (the new plugin).

If anyon is interested, you are invited to have a look at my site: -> http://babylondreams.de/wordpress/index.php/2006/06/16/renderman-mentalray-comparison/


#2

For the motion blur, you didn’t seem to use the rasterizer (previously known as “rapid scanline motion blur”)?

For the hair, you didn’t seem to use the detail shadow maps, which is similar to the “deep” shadow maps of PRMan.

/Z


#3

So far I only had problem with Rapid Motion Blur MasterZap, but you are right, I didn’t use it in this test. Still I guess that it wouldn’t stand a chance against the Renderman Motion Blur. But maybe I am biased here. Renderman just won me over from the very first render in this area.

As for the fur, you are right. Didn’t use them, I completely forgot. :sad:

I will update my post to reflect these shortcomings. Thanks for the hints. :thumbsup:


#4

If you do not, you are not even comparing apples and oranges…

PRman’s motion blur undersamples the shading, whereas mental rays “full” motion blur oversamples temporally to make real motion blur, including:
[ul]
[li]motion blur in shadows[/li][li]motion blur in reflections and refractions[/li][li]multi-segmented motion blur (curved motion blur paths)[/li][/ul]So to even begin comparing and to make the comparision anywhere near “fair”, you should use the rasterizer, which also undersamples shading.

/Z


#5

Like Zap mentions, comparing render times with the settings you have is not comparing like with like. Unfortunately the version of mental ray in Maya 6.5 is not nearly as good as the version in Maya 7, and is significantly older than the Renderman for Maya product. If you have access to Maya 7, I would love to see the tests repeated, this time with the same basic settings for both renderers. The defaults do not give you equivalent setups, so some adjustment of settings is necessary. You need to look carefully at the number of samples per pixel, geometry tesselation settings, and so on.

Additionally, the test scenes which shipped with Renderman for Maya are clearly chosen to make that product look good. A neutral set of test scenes would be much better for a comparison, although it will also be more work.


#6

maya 8?? doh! :eek:


#7

Sorry, Maya 7 :slight_smile:


#8

Well, as I said in my post as well as in the article. It is a quick and dirty test. And I didn’t claim to be a master of either of the renderers. But that’s what makes this test interesting for the casual user (who is probably not a master of either of the renderers either). Which renderer is more user friendly in terms of speed and features. Sure I can tweak each and every detail in both packages, but for most projects you don’t have the time to do that. So I was looking for out of the box values.

I beg to differ. How more neutral can you get then a teapot in motion with one directional light? If I switched from a teapot to a bunch of spheres, do you think that would influence the results, just because Maya created the spheres and not Pixar?


#9

I think the point Zap was making about the motion blur test is that MR’s is more accurate, hence the slower speed. So it isn’t really ‘out of the box’ 'cept that MR defaults to a higher quality/technique so to speak. Switch to rasterizer to make a fairer speed comparison. As far as ‘casual user’ goes, I don’t get that, how can you casually use any renderer if you want to get owt worthwhile out of it :wink: .

A comparison of SSS would be good, RfM is a dream to setup in comparison to MR here, but then it’s not as flexible or controllable as it is in MR. The other thing I like about RfM is the ease of setting up multiple output passes, still in MR we have Francesca and Maxs Ctrl_buffers for that.


#10

Ok, maybe the word casual was a bad choice. What I really meant, was what I said in the sentence after that.

I was speaking about project in a commercial environment of a small studio, where you do not have days to tweak your settings.

And the out of the box value for MentalRay is the accurate slow MotionBlur in this case. I guess this discussion is only going around the Motion Blur as far as I can see anyway. If you wish I can redo this special test with Rapid Motion Blur.


#11

just for the sake of completeness, there’s another comparison here:

http://www.maxschoenherr.de/animation/Maya/RendermanForMaya/rendermanMain.html

which comes to roughly the same conclusions


#12

which comes to roughly the same conclusions

oh my word, that’s terrible, he has no idea where to find the MR features to run a comparison. This all goes to show how lacking MR documentation is when you unpack your Maya box.

RfM is a damn good renderer in terms of speed/quality and ease of use, no doubt. It’s not very flexible compared to MR, or compared to the full Renderman package however. MR provides much more, a proper GI solution for one. RfM’s GI, er, isn’t is it ? I don’t think comparing MR to RfM is sensible really, t’would be better to compare it to Turtle or Gelato I reckon. They are two other products trying to provide the same thing as RfM, MR is something else really. Not only that, as Turtle and Gelato are renderers you also have to pay for a comparison would be more useful I think, considering the differences in price.


#13

I could be a complete jackass, however… Your tests don’t include much of mental ray, almost more of a Maya Software comparison, some have mental ray, but… Why not all? Don’t understand how the conclusions can be roughly near the same in anyway if not all the tests were inclusive of the same renderers …

Am I wrong, or is there something I am missing here?

az


#14

see, I was just linking to another test not conducted by myself, which came to roughly the same conclusions and confirmed the conventional wisdom as far as

  • Motion Blur
  • Displacement
  • GI

are concerned, these being areas of particular interest to many users

cheers


#15

test? this is a test? my god! :wink:


#16

for gawd sake’s if all the time the weisenheimers spend niggling would have gone into testing, then we might have assembled the definitive super-comprehensive, ultra-fair and hyper-exact PRmanMentalRay comparison database by now :rolleyes:


#17

Agreed RabbitRun.


#18

there is a “vraylike” test thread, with billions of tests, try to test some scene with renderman instead of try to render some sphere or a teapot! :rolleyes:


#19

yes, I am aware of this thread and it’s very informative; all I’m saying is, it might be more productive, if we could get something similar started with respect to Renderman vs. Mental Ray

cheers


#20

I recently used Prman in production for the Transformers trailer. This was my very first time working with Render Man, and I must say I was impressed with the output. I only had a couple things to tweak in render settings, and once I got familiarized with the palette system, setup time was comparable to MR…

I was rendering 8k displacement maps in full cinema frames in about 6 minutes on my workstation… And it was pretty gorgeous. In fact, my supervisor asked me not to use any bumpmaps, everything was being displaced, even little scratches on ‘the robot’s’ body.

Maybe I suck at Mental ray, but I never got this kind of performance from it.