Deep Image Compositing

   25111   23   6
User Avatar
Member
792 posts
Joined: April 2020
Offline
I believe nuke 7 already supported the deep exr's in openexr 2.0

The exiting feature in 8 is the multipart reading. Now you don't have to load the entire exr anymore if you only need one aov. (of course the exr needs to be written multipart as well, sidefx, please? , but you can alway convert right after rendering).

Cheers,
Koen
User Avatar
Member
92 posts
Joined: Aug. 2010
Offline
yeah koen's correct; Nuke has had this stuff for a while now. to generate the data in Houdini 13, you save your DCM as OpenEXR2 via the EXR file extension (instead of RAT). confirm it's a different path then your flat raster, otherwise they'll stomp on each other.

then in Nuke 7+, there is a whole suite of deep image tools (I could talk about deep images for hours)… but to get started, you'll have a deepRead for each deep image, and deepMerge to combine them. After combining everyone, you'll need to flatten the data back to 2D via deepToImage.

other handy tools are the deepSample (to pixel inspect the different channels' samples) and deepToPoints (to look at your data as a point cloud).

i hope that helps!
-jeff
Edited by - Dec. 9, 2013 01:57:17
User Avatar
Member
27 posts
Joined: June 2013
Offline
thanks jeff and koen.

i knew about nuke's workflow on deep compositing since nuke 7, but never got to render stuff out from houdini in this format. i'll give it a try.
User Avatar
Member
17 posts
Joined: Feb. 2006
Offline
Even if this might be solved, I am contemplating a custom Z pass creation, as exercised in Maya here:
http://asephei.blogspot.com/2011/10/maya-how-to-manually-generate-custom.html [asephei.blogspot.com]

I presume it can be implemented in vops, (for <H12 users). I haven't read it thoroughly yet, but seeing the network, it seems like a simple operation.

Nevertheless, the great work of SYmek has to be pointed out and acknowledged
  • Quick Links