[Kevron]

Hi, Is it possible to render in 16 bit half float with ZIP (1 scan line) from Maxwell, or is it "only" 32 bit Exr?

Current workflow is to open in Nuke and render it out again to 16 bit half with 1 scan line compression. But I seem to loose the connection to the Maxwell Nuke plugin when doing so. (Mxi Mixer will not load information from the re-rendered files.

I guess this is both a Maxwell and a Nuke question, so sorry if this is the wrong forum.

Any thoughts on this subject?

Regards,
Tor M

[renbry]

I've been waiting for a less 'brute force' approach too, Tor.

My compositors complain endlessly about the over-kill 32bit float packing method. They even say that 32bit is only needed in a very few niche cases (beyond 16bit half float)

[Kevron]

Yes, Nuke operates a lot faster with image sequences when compressed in either 1 or 16 scan lines.
Also nice to know that you guys do most of your comp work in 16 half float.

I reckon it's on next limit's radar to implement more options with exr. They seem to care about not only beeing a single frame render software, and some more options with exr would remove some steps in post vfx pipeline.

I think you have manage to do some really beautiful work with Maxwell, always inspiring to see your work.

[renbry]

Thanks very much

We have been runnning a post-render program which does a domain of definition detection (DoD) and recompresses/reformats the EXR into our preferred packing method (zip scanline) but doesn't change the bit depth for safety reasons. It would be great to one day deprecate that post-render stage as there's some bugs relating to resumed renders.

[Mihai]

But I seem to loose the connection to the Maxwell Nuke plugin when doing so. (Mxi Mixer will not load information from the re-rendered files.

Which information would you like to retain in this case? Would Multilight not be necessary?

[Kevron]

I would really like to retain the possibility to use Multilight after converting the output to 16 bit half float with scanline compression.
Current workflow is to shuffle the different light passes and add them together. It works for now.