We've been using Brenda for quite some time now for rendering architectural stills. Since our editing stations are doing nothing at night, I started making a desktop app that allows you to pick jobs from the AWS Queue.
It works, sort of, well on Linux, but on Windows it looks like it's forgetting to run the python script. The python script does a number of things: it specifies which subframe should be rendered, makes sure the tile size is set to 32x32 (cpu) and sets a proper output location for mask outputs from compositing.
It outputs 4 files:
subframe.py is the actual job for rendering the subframe and settings the blender settings. data.txt shows the blender output. err.txt shows any console errors if they happened. the frame (and masks if any) are PNG images.
The command being executed looks like this:
blender -b *.blend -P subframe.py -F PNG -o $OUTDIR/frame_######_X-0.0-0.2-Y-0.2-0.4 -s 1 -e 1 -j 1 -t 0 -a
As you can see the err.txt's are both empty, which makes it look like everything went smoothly. On the data.txts however, you do see a difference. At the bottom of the first one you see the line:
This is a testprint from the subframe.py file
This is an output from the subframe.py file. This line is missing from the bad run data.txt file however. Leading me to believe that the subframe.py file got ignored in this case.
I fail to diagnose why this file is not being used when starting the execution of that work snippet. Running the same thing on linux does produce the suspected output. Any idea on why this is happening or how to troubleshoot it?