CGTalk > Main > General Discussion
Login register
Thread Closed share thread « Previous Thread | Next Thread »
 
Thread Tools Search this Thread Display Modes
Old 01-03-2013, 03:26 PM   #1
wancow
A Noid
portfolio
Christian Holm
Livermore, USA
 
Join Date: Nov 2009
Posts: 208
So: What is the REAL Story on RenderMan?

On another forum (I should have asked here) I asked if anyone actually used RenderMan... In that I see a lot of great renders in Brazil, VRay and just about everything BUT RenderMan...

By RenderMan I meant the RenderMan produced by Pixar Animation Studios, which holds the RenderMan trademark.

So, my question for all y'all is this:

Is any "RenderMan Compliant" Render Engine actually RenderMan? Or is the source code for RenderMan actually considered the gold standard for render engines?
 
Old 01-03-2013, 03:29 PM   #2
oglu
Christoph Schädl
 
oglu's Avatar
portfolio
Christoph Schädl
Austria
 
Join Date: Mar 2003
Posts: 3,329
__________________
...
 
Old 01-03-2013, 04:20 PM   #3
wancow
A Noid
portfolio
Christian Holm
Livermore, USA
 
Join Date: Nov 2009
Posts: 208
That's a really nice article! Thanks!

It answers the question I asked, but what it doesn't tell me is why more folks who post here aren't using it... I would have thought it would be highly desirable to work with it...
 
Old 01-03-2013, 04:29 PM   #4
DePaint
Banned
 
DePaint's Avatar
portfolio
Emre M.
Istanbul, Turkey
 
Join Date: Jun 2008
Posts: 1,156
Quote:
Originally Posted by wancow
I would have thought it would be highly desirable to work with it...


There are 4 main reasons why RenderMan isn't used much outside the big CG/VFX studios

1) It is fairly expensive compared to render engines like Vray, or MentalRay (the latter of which you get free with Autodesk 3D products)

2) To get the most out of RenderMan, you need to hire trained coders who write custom shaders and other things for RM; if you don't have such coders, much of the technical power of RenderMan can't be used

3) RenderMan didn't raytrace very well until recently, so it isn't particularly well suited to things like Archviz or ProductViz GI renders.

4) The depth of technical features and customization RenderMan allows is overkill for a freelance 3D artist, or for a small studio with just a handful of artists


RenderMan is powerful, and very customizable, but it has always been designed to be a "Big Studio Renderer".

Hence you don't see many amateur, freelance or otherwise solo CG artists posting stuff rendered with it.

Also for some things - like Archviz interiors for example - smaller renderers like VRay give a good result in far less time.
 
Old 01-03-2013, 04:52 PM   #5
wancow
A Noid
portfolio
Christian Holm
Livermore, USA
 
Join Date: Nov 2009
Posts: 208
Okay, that makes total sense to me. On another forum we were talking about 3Delight, which is a "RenderMan Compliant" render engine, and one of the things that came up was that features that seem "out of reach" are those that require a lot of coding to access... what you seem to be telling me, O Helper of Noobs, is that RenderMan basically requires a lot of coding just to use at all
 
Old 01-03-2013, 05:00 PM   #6
DePaint
Banned
 
DePaint's Avatar
portfolio
Emre M.
Istanbul, Turkey
 
Join Date: Jun 2008
Posts: 1,156
Quote:
Originally Posted by wancow
O Helper of Noobs, is that RenderMan basically requires a lot of coding just to use at all


There is a simplified/artist-friendly GUI version of RenderMan, called RenderMan For Maya:

http://renderman.pixar.com/products/tools/rfm.html

That lets you use RenderMan for rendering in Maya without knowing a lot technicalities.


The big "Pro" version of Renderman only makes sense if you have programmers/TDs trained to use its inner workings, which are best accessed through writing scripts and custom code.

That way, for example, you can create very customized shaders & materials that you cannot easily create with GUI controls alone.


What may be confusing you also, is that RenderMan is both a renderer made by Pixar and a "rendering standard".

A renderman-compliant renderer like 3Delight or Air can use a standard 3D Software-to-Renderman interface to know what to render.


But under the hood, 3Delight and Air work differently from Pixar's RenderMan.


I hope that helps a bit...
 
Old 01-03-2013, 05:14 PM   #7
wancow
A Noid
portfolio
Christian Holm
Livermore, USA
 
Join Date: Nov 2009
Posts: 208
Regarding that RenderMan is a standard and a Render Engine has become clear to me over the past few hours.

I also am under the impression that what goes into the RenderMan render engine code becomes the RenderMan Standard, if not immediately, eventually.

Unfortunately, there's a whole lot of confusion out there regarding RenderMan compliant engines, so much so that people will actually tell you that 3DeLight IS renderman! I've had a couple tell me this, and I know for fact it's not true, that the code is just not the same and even if it's reading the same shader language it's not going to give the same results... gaw... so I finally had to break down and come to you guys, because I know I'll get straight answers here... have for years, and it's always refreshing!

Thank you so much for your help!
 
Old 01-03-2013, 06:21 PM   #8
shehbahn
Expert
 
shehbahn's Avatar
 
Join Date: May 2004
Posts: 376
Quote:
I also am under the impression that what goes into the RenderMan render engine code becomes the RenderMan Standard, if not immediately, eventually.


Not quite : the RenderMan specification is very carefully kept agnostic of implementation details, many of which are patented in their respective implementations as competitive advantages. For example, PRman's (Pixar's implementation) stochastic sampling algorithm has been an important competitive advantage for many years. Another more recent example would be some of the details of the Catmull-Clark subdivision algorithms, where Pixar released the patents last year because they interfered with a standardized geometric definition (aka. OpenSubdiv).

Historically, Pixar has been the main contributor to the Renderman spec, but that was because the entire industry at large was using PRman, with little in the way of significant commercial alternatives. Now that we have several viable competing products on the market, it will be interesting to see if they will decide to contribute to this framework or if the spec will eventually become obsolete...
 
Old 01-03-2013, 07:15 PM   #9
wancow
A Noid
portfolio
Christian Holm
Livermore, USA
 
Join Date: Nov 2009
Posts: 208
A little OT: but OpenSubDiv looks exciting!
http://www.youtube.com/watch?v=Y-3L9BOTEtw
 
Old 01-03-2013, 07:15 PM   #10
thatoneguy
Interesting User Title
 
thatoneguy's Avatar
portfolio
Gavin Greenwalt
Exec. VP of Pixel Mergers and Acquisitions
Straightface Studios
Seattle, USA
 
Join Date: Jan 2005
Posts: 2,332
Renderman's popularity in feature films is kind of an anomaly IMO. The reason you've seen Brazil/Vray/MentalRay more popular is because they are/were better renderers for 99% of use cases. For instance it wasn't until I think IronMan 2 that ILM started using area lights. Why? Because they thought they were too slow to render. Compare that to a small studio which might be rendering 15s of animation for a TV spot and that performance hit from using area lights is more than made up by the ease of lighting and the quality of the results. I actually think that Renderman went about it completely backwards, Arnold, Vray, Brazil etc I think have the right philosophy "Computers are cheap, artists are expensive". It's way more cost effective IMO to just throw CPU cycles at a problem if it means less incredibly expensive artist time. Back in the day PRMan made lots of sense, motion blur was essentially free (compared to being extremely expensive for ray tracers) ditto with displacement. However, that only applies if you don't need any raytraced effects, which is largely "never" now.
 
Old 01-03-2013, 08:22 PM   #11
Oddgit
synthetic photographer
portfolio
Robert Turner
Sacramento, USA
 
Join Date: Mar 2004
Posts: 344
Send a message via AIM to Oddgit
I learned PRMan in school and was using it for my masters thesis before some personal stuff got in the way. But what could be done with PRMan is outstanding, the pointcloud occlusion and colourbleed, the really cool delayed read archives and LOD functions, but my favorite, even though it has its quarks SLIM is very cool. Building a procedural shader in it made so much sense to how my mind works. You could create fantastically complex procedurals very easily. Like what was mentioned above, it does need more support and knowledge then the integrated renderers. Fortunately, RSL (Renderman Shader Language) is pretty simple. There is a very cool node in SLIM called SLBox which is just an empty box and you can input RSL so if a prebuilt SLIM node does not do what you want or work as expected you can make your own.
__________________
I like to tell people my scars from my CTS surgery are actually stigmata scars.
 
Old 01-04-2013, 03:15 AM   #12
earlyworm
car unenthusiast
 
earlyworm's Avatar
Will Earl
craftsperson
Grizzly Country, Canada
 
Join Date: Mar 2005
Posts: 1,685
Quote:
Originally Posted by thatoneguy
For instance it wasn't until I think IronMan 2 that ILM started using area lights. Why?


Yeah but in certain areas ILM haven't always been ahead of the game. They didn't start using SubDs for their creatures pipeline until Eragon (2006). Heck I've been using point-clouds to simulate area lights for the last couple of years using a fairly simple PRMan trick.

Quote:
Originally Posted by thatoneguy
I actually think that Renderman went about it completely backwards, Arnold, Vray, Brazil etc


Motion blur and displacement are also huge contributors to the photo-realism of an image (in fact motion blur was one of the first things Pixar aimed for in terms of trying to nail photo-realism). Consider that a significant amount of films got by and look great without raytracing, global illumination or even some of the more fancy shading techniques available today.

Pixar's raytracing speed is also pretty decent - especially for ray-intersection hits. I've only used depth-map shadows for lighting a few times in the past five years.

Quote:
Originally Posted by thatoneguy
I think have the right philosophy "Computers are cheap, artists are expensive". It's way more cost effective IMO to just throw CPU cycles at a problem if it means less incredibly expensive artist time


I completely disagree with this philosophy/sentiment, which sounds great on paper, but only scales so much - when you experience the issues associated with even a hundred users overloading a large render farm with various simulations and renders, you'll realise the problem isn't as simple as adding more CPUs and HDD space nor is it by any means cheap - it's compounded when you start getting into the hundred+ artists hammering an even bigger render farm. Adding more machines you'll often have to deal with finding space to put them, power and cooling requirements, networking, render queue management software (some don't scale well to large number of jobs), ongoing hardware maintenance - none of these issues are cheap nor simple to implement or maintain.

Quote:
Originally Posted by thatoneguy
...because they are/were better renderers for 99% of use cases...


I'd say that number is more likely 80% of use cases. There are still things that PRMan does as a renderer that still really stand out - motion-blur and displacement have already been mentioned, but hair, skin and volume rendering all shine in Renderman. You cannot discount these aspects as they make up a very large part of the rendering that goes in on in feature film work - primarily animated creatures/characters and large scale destruction effects.

In terms of customising, Renderman is able to solve more esoteric rendering and shading problems without needing to know C++ and some of the more complex programming concepts required to write good working custom shaders in some of the other renderers out there.
 
Old 01-04-2013, 03:42 AM   #13
Oddgit
synthetic photographer
portfolio
Robert Turner
Sacramento, USA
 
Join Date: Mar 2004
Posts: 344
Send a message via AIM to Oddgit
Good points Will, you forgot to mention PRMan can render HUUUUUUUGE scenes.
I am glad you mentioned displacement and blur. When i got to use Renderman and saw how well it handles displacement i was blown away. You can have very complex displacement maps with a minimal hit to render time. Same with motion blur and DoF. I remember hearing, not sure if it is true or not, that the renderer figures out what will be blurred by the depth and motion and then lowers its antialiasing and sampling in areas that will be blurry to save time. Though that may be an urban myth.
__________________
I like to tell people my scars from my CTS surgery are actually stigmata scars.
 
Old 01-04-2013, 04:14 AM   #14
wancow
A Noid
portfolio
Christian Holm
Livermore, USA
 
Join Date: Nov 2009
Posts: 208
Quote:
Originally Posted by Oddgit
Good points Will, you forgot to mention PRMan can render HUUUUUUUGE scenes.

So are you saying that PRMan handles large scenes better than most other render engines? I remember scenes from A Bug's life and, was Weta using PRMan for the Fellowship of the Ring and their Massive scenes? (I mean those scenes made with the Massive software).
 
Old 01-04-2013, 10:45 AM   #15
earlyworm
car unenthusiast
 
earlyworm's Avatar
Will Earl
craftsperson
Grizzly Country, Canada
 
Join Date: Mar 2005
Posts: 1,685
Most renderers can handle fairly large and complex scenes these days - Renderman* has been able to do it for a while through various methods, but I wouldn't necessarily say it was a huge selling point these days.

*I'm referring to PRMan here, but a lot of this also applies to 3Delight.

Massive originally had it's own renderer called GRUNT (Guaranteed Rendering of Unlimited Number of Things) - from what I understand it was a fairly basic renderer, but it could handle anything you threw at it. Now people tend to render massive agents using other renderers.

Quote:
Originally Posted by Oddgit
Same with motion blur and DoF. I remember hearing, not sure if it is true or not, that the renderer figures out what will be blurred by the depth and motion and then lowers its antialiasing and sampling in areas that will be blurry to save time.


It can reduce the shading rate in areas undergoing any motion-blur or depth-of-field via the motion and focus factor settings - the idea being if it's going to be blurred it doesn't need to be as finely shaded. It doesn't affect anti-aliasing which is controlled via a different set of controls.
 
Thread Closed share thread


Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

vB code is On
Smilies are On
[IMG] code is On
HTML code is Off
CGSociety
Society of Digital Artists
www.cgsociety.org

Powered by vBulletin
Copyright ©2000 - 2006,
Jelsoft Enterprises Ltd.
Minimize Ads
Forum Jump
Miscellaneous

All times are GMT. The time now is 12:20 AM.


Powered by vBulletin
Copyright ©2000 - 2016, Jelsoft Enterprises Ltd.