PDA

View Full Version : Mental ray deep compositing images


gauranga108
08-11-2011, 12:00 AM
I am looking for info on creating images in mental ray to be used in nuke for deep compositing.

It sounds like to future of compositing.

Seems like render man and houdini are the only renders out there generating images for nuke deep compositing.

Wondering in there's more info out there for mental ray users.

Thanks

beaker
08-24-2011, 07:46 AM
Nuke currently only supports deep image files created with prman(dtex files). In the future though, once OpenExr 2.0 comes out, that will be the more standard deep file format which it will support. It will then be up to the renderers to implement writing deep image files into an Exr container.

gauranga108
08-24-2011, 08:37 PM
I see.

I heard exr 2 is coming out soon.

more waiting . . .

thanks mate

beaker
08-24-2011, 08:45 PM
Even after Exr 2.0 comes out your going to have to wait for Mental Ray to have deep support in the renderer because not all renderers support it by default.

Last I knew only prman, 3delight and Vray support it. There may be others though that I am missing.

theotheo
08-24-2011, 10:06 PM
Don't forget Mantra Deke, my personal favourite :)

beaker
08-25-2011, 02:50 AM
right, forgot about vMantra.

RadVfx
09-16-2011, 04:16 PM
The basis for "deep compositing" is having methods for isolating voxel depths from camera independent of the depth image. In that way, you can affect a blur on something that is behind something else.

But since the image format is not available in Mental Ray for this yet, some of the compositing you might wish to do using depth renders might get you partway there.

Simply isolate depth edges of objects much like you would a dof in Nuke, and you should be able to composite without using object's matte (alpha) channels.

I will give this a try in a simple example and post.

beaker
09-17-2011, 01:53 AM
The current Nuke deep tools don't have support for dof yet.

The initial deep support is for doing holdouts and depth compositing which is quite shoddy with zdepth files. This is a huge deal with motion blur and semi transparent objects can be perfectly composited together where using a zdepth for this is completely.

CGTalk Moderation
09-17-2011, 01:53 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.