Publishing an image sequence in Nuke deletes the renders?

OK, so this is super weird. We generally don’t publish directly from nuke. Usually the team will stick the output in RV and publish from there. I’m looking at the publisher now, as I essentially want to hack out some bits of it to make my own publishing and tagging system for a bunch of green screen stuff we’ll be ingesting soon - however I thought I’d just look at the normal publisher first, and I’m getting some crazy issues.

If you use the Shotgun Write Node to render some of your comp and then use the Publisher to publish the image sequecne (with out without the nk file being published too), it breaks, as it says it can’t publish empty files - and lo and behold, all the renders from the just rendered sequence are suddenly 0kb. That’s with our write ode having written Jpegs. If I have the write node doing EXRs instead, the publish doens;t fail, but it still turns the previously fine exr files to all be 0kb. I have absolutely no idea why this would be happening.

Has anyone else encountered this behaviour before?

I’m going to look through the actual publishing code and just strip out the code to publish the image sequence to use in my tool, but I fear that even when I’ve done that, this might still ocur as something in the Shotgrid publish process is causing this. I’m usng the latets version of tk-nuke ad writenode (and everything else).

Any help greatly appreciated!

1 Like

OK, on the vague (and hopefully very unlikely) offchance anyone runs into a similar issue and needs a solution - check your templates.

On rebuilding the project config I got a new error specifically about being unable to copy frames from the workfiels location to the publish location as it was the same location… Turns out the templates didn’t have a separate publish location, an oversight, as we haven’t published via nuke in years, one that I didn’t notice until now. Fixing the templates sorted the issue.

I think in the case of the original issue, somehow it was trying the copy, and whilst failing, not reporting correctly what had happened and the files getting zeroed out as a function of the mid point of the copy attempt, and then not reporting the error correctly. I really don;t know why htat would ahve been the case, but there we are!