View Full Version : Mental Ray Render Passes

01 January 2007, 06:38 PM
Hey there --

Im currently doing an animation in 3dsMax as part of my degree and the scene is starting to get quite dense.

So while creating this i've been trying to think of rendering solutions in MR (MR is a requirement) and came to the conclusion that i will probably need to venture into the land of render passes.

I was wondering if there was any nice resources for this, seeing as MR is not very well documented im finding it hard to get any concrete information as such.

So a few questions to put out there:

Render passes; How do i achieve them and what types should i be looking at?
(Please to not think i havn't looked at the different types, i have tried to research this with varying degree's of success)

Im particularly interested in motion blur, and have heard of what is called a 'velocity pass' that simulates motion blur by vector points i believe instead of the change in image. -- Is this viable and how would i go around doing such a thing?

I'm using HDRI solution, will this change any of my needed passes and if so how and why?

What information should i know before commiting myself to this?

I am no stranger to compositing as i did it for 2 years as part of a multimedia qualification. So i am tempted by the flexibility render passes will grant me. I have also become fairly confident with mental ray as a renderer, so hopefully this wont be too daunting!

and as a last note... any help at all is much appreciated.


01 January 2007, 12:51 PM
Rendering in passes is pretty much a necessity in production as it is a major time saver and makes life a whole load easier with the extra control it gives you.
How many different passes you decide to output depends on the shot requirements really, and varies from scene to scene.

There are two different forms of pass:

The first is for scene management purposes, and involves rendering out your heavy scene in layers. So background hills may be rendered separately from foreground grass etc. In max you'll need to become used to using the "matte/shadow" shader for this.

The second type is for image control purposes, and involves rendering out separate passes for colour/shadow/reflection etc. Obviously each scene layer would need its own separate image passes, so you can see that its very easy to quickly become swamped in files if yuo're not careful when actually it would have been just as useful to only render out 1 or 2 passes.

In terms of which passes to separate, I would suggest trying to have as much as possible in the one render. For example its often not really necessary to output a separate specular pass. But it depends on how much control you want.
If you wanted to go into more detail you could have a shadow pass, a z-depth (if you need depth of field), a reflection, a velocity (for motion blur) etc etc.

There are a couple of different ways of getting these image passes. Firstly, if you're not already, you should be outputting your main renders as separate frame sequences in your file format of choice (something with lossless compression and alpha, such as .tga/ .tif, etc).

Some of the passes will have to be output as separate sequences by using the "render elements" tab in the render dialog in max.

For some passes though you can output 1 sequence into a file format that can record more information than the usual RGB and Alpha. Which format you use depends on which compositing package you use, and also what passes you need. For example, if you use combustion you can output the sequence as .rpf files, which will record z-depth, velocity, colour among others. RPF files tend to be quite large though (often over a meg each). I don't know for sure but i wouldn't expect many other compositing programs to support this file type though, but i could be wrong.
The other format you could use is ILM's Open EXR format. However this isn't recognised by every package (but probably more than rpf). Its not supported by combustion 3 for example, but is in combustion 4.

Once you've got your sequence you can use the separate passes to tweak individual elements (shadows/motion blur) separately without having to re-render the whole sequence again. The velocity information can be used to create motionblur in the compositor instead of at render time. For example combustion has an "rpf motion blur" operator.

Using HDRI won't affect using passes (as far as i know).

Hope that helps. :thumbsup:

01 January 2007, 01:54 PM
Thanks Matt, you've really helped alot, your post has norrowed down alot of questions for me which is really going to help me research this!

Again, thanks so much for the reply.

02 February 2007, 11:53 AM
Glad i could help :)

CGTalk Moderation
02 February 2007, 11:54 AM
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.