Does anybody have any idea why merging multiple Read nodes and copying them 
into a custom channel might cause a Nuke render to fail??

Specifically, I have script where 5 EXR mattes are being merged along with a 
key and then copied into a custom channel (Ex: matte.blahblah) at the top of 
the tree. The depth channel in those 5 EXRs is further merged with the depth 
channel in a 6th EXR and copied into another custom channel(Ex: test.depth). 
The rest of the script follows after that.

When submitted to the farm, the frames fail fairly quickly with an error for 
each Read node saying "Could not read image file 'xxxxxxxxx' Too many open 
files. Reader did not set input channels. Reader did not set bounding box."

The script can be rendered manually on my local machine through Nuke but when 
running a command line render locally to emulate the behavior of a node on the 
farm it fails in a similar way.

If I disable any one of the EXRs, the render goes through with no issue. 
Similarly, if I precomp after the custom channels are copied in or precomp 
after the EXRs are merged, the render has no errors. I have tried every 
possible combination I can think of Copies, ShuffleCopies, ChannelMerges, etc. 
to get the script to render without a precomp but nothing has worked so far.

Why would 4 EXR mattes plus a key piped into a custom matte channel and 5 EXR 
depth channels piped into a custom depth channel work, while 5 EXR mattes+key 
and 6 EXR depth channels won't work? Is there some magic number Nuke is 
expecting here? Does it have something to do with commandline renders?

I am currently at a loss as to what could possibly cause this. Precomping is 
not the end of the world but it would be an extra step in the pipeline which 
we'd like to avoid if possible. Even if it is necessary, I'd still like to 
understand why. I hope somebody can shed some light on this. Thanks!!



_______________________________________________
Nuke-users mailing list
[email protected], http://forums.thefoundry.co.uk/
http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users

Reply via email to