Page 1 of 1

gamma change within Resolve between Catalina and Big Sur

PostPosted: Tue Feb 02, 2021 11:23 pm
by metkat
I'm seeing a change in the way some (but not all) files are handled within Resolve when I switch between MacOS Catalina and MacOS Big Sur. I'm pretty confident I've isolated the problem to the two OS versions- if I do a clean OS install on a wiped external drive, then a clean install of the free version of Resolve 17.1b7, and do nothing else besides open the source file, the result is different depending on which OS version I installed.

With h265 encoded 4K video from a GoPro 8 using ProTune, here's the waveform of what I see under Catalina:
Image

vs under Big Sur:
Image

I see this difference both visually and in the scopes within Resolve, and (more of a big deal) in otherwise-unchanged exports to ProRes in each case. Those waveforms above are looking at the same frame of a pair of ProRes exports which should be identical other than having been run under different OS versions.

Likely-seeming variables I've eliminated:
- color managed or non-color-managed workflows
- studio / free versions of Resolve
- 'Use Mac display color profiles for viewers' on or off
- 'decode h264/h265 using hardware acceleration' on or off
- Rec.709 Scene vs Rec.709-A input color spaces

I don't see this difference importing h264 HD video, nor does it happen with video I've already transcoded to ProRes. The original file looks identical opened within QuickTime or FCPX on either OS version (they use a gamma curve roughly identical to what I see in Resolve under Big Sur, although don't introduce those bands in the waveform).

a) can anybody else duplicate this?

b) I know "right" is kind of arbitrary, but which is right? Compared below, the Big Sur curve looks visually flatter like I'd expect from a ProTune curve, and there's more visual detail in the highs. On the other hand, the banding in the waveform looks like lost data to me. I've tried fiddling with luma curves etc to smooth those bands out, unsuccessfully.

c) am I missing something obvious, or misunderstanding those bands in the waveform scope?


For visual comparison, Catalina:
Image

vs Big Sur:
Image

Re: gamma change within Resolve between Catalina and Big Sur

PostPosted: Wed Feb 03, 2021 4:56 am
by waltervolpatto
1) one seems flagged full range and the other video range
2) one seems 10 bits and the other only 8.

is that correct?

Re: gamma change within Resolve between Catalina and Big Sur

PostPosted: Wed Feb 03, 2021 6:10 am
by metkat
You're onto something with full range vs video range. Both of those are with Resolve interpreting the original file as video range. If I explicitly switch it to full range under Catalina, I see the same curve being applied as in Big Sur with video levels, but without that banding/striping in the result. If I explicitly switch it to full range under Big Sur, I get the same shape of curve but squished down into the ~100-900 range (with the banding still present).

My understanding is that everything becomes 10 bit as soon as it gets into Resolve, then I'm exporting into ProRes which is 10 bit. That artifacting does look like something might be being handled as 8-bit somewhere in the process? Switching the 'use 10 bit viewers' pref doesn't make a difference, is there some other setting for 8 vs 10 bit in Resolve I don't know about?