Are you sure it’s 2.2 gamma? I thought it was sRGB gamma.
But aside from that, if it works in AE it should work in Premiere.
If you go to your sequence settings all the way at the bottom there’s:
which needs to be checked but I think it is by default.
I also see a lumetri effect above OCIO, is that doing something?
Also did a check up regarding that gamma curve.
This is what I get so I think it’s actually sRGB gamma. (Left side = Premiere / Right side = AE)
Judging by the reference bottom right which is the “proper” way to handle it in AE.
Well I thought it was, but now you have be questioning myself (which is always good!). Hopefully @Thomas_Mansencal or @nick can confirm it’s correct.
Yes! When I move the OCIO stuff above Luminare the clipping is gone!
Thanks for that. I see right above that in the Sequence Settings is Max Bit Depth. With that on the clipping goes away and you can do stuff like exposure downstream. Here’s the setup I have for this in Premiere. The stack is
OCIO convert footage to ACEScg working space (in: AP0 + 2.2 gamma, out: ACEScg)
Grade in Luminare (thanks to the 32 bit depth). I lowered the exposure to test here.
Keep in mind that Lumetri isn’t really designed to operate in linear so some controls might feel weird or give unexpected results. I think also the slider ranges are clamped on the Lumetri Panel. I never use the panel personally but rather just the values on the effect controls version.
Placing the fx/grade in a log sandwich of ACEScct or other log formats should feel a lot nicer for anything that isn’t exposure, blur, glow.
I don’t really have any big exr sequences to test, do you? I’m quite curious how well it performs vs AE.
I tried testing it with a RED file but if I set that to Legacy instead of IPP2 to access Linear gamma decode it looks a bit weird. Probably missing something obvious getting blind by all the testing hehe.
Indeed that would be good to test. I suspect it will be a lot slower than a Prores clip. I’ll need to leave that testing for someone else however. I’m a VFX CG guy.
I believe Premiere’s default timeline colour space is BT.1886 (the options for Working Color Space in Sequence Settings are Rec.709, Rec.2100 HLG and Rec.2100 PQ) so it will be adding 2.4 gamma to the EXR. That is consistent with my testing.
Good find on the OCIO plugin working with Premiere by the way.
/Library/Application Support/Adobe/Common/Plug-ins/7.0/MediaCore is the Mac path.
I just checked, and Premiere (v15.4.0 on Mac) does appear to be adding 2.4 gamma to the linear AP0 data. I’ve just added an “AP0 2.4 gamma” colour space to my experimental OCIOv2 config which from a very brief test does appear to work in Premiere.
Thanks for correcting that Nick!
It looks like AE does this for sRGB instead as that’s usually the default setting but has a separate switch for non-managed with 2.4 as second option.
I did notice whilst a 2.2 gamma didn’t fully null, the development OCIOv2 ACES config also didn’t show a full null when differencing against perserveRGB inside AE but was a lot closer using Display-sRGB. Is there a reason it’s not spot on? When I use the Color Profile Converter of AE itself it does null completely.
I suspect it’s down to the difference between pure 2.2 gamma and the two part sRGB (inverse) EOTF. I just pushed a commit to my repo which adds 2.2 gamma and sRGB EOTF AP0 colour spaces.
(Edit: error in initial commit. Amended version should be correct)
@Nick, I see you used ExponentWithLinearTransform for the EOTF in your OCIO v2 config. Since OCIO v1 does not support that transform do you know how to do this in OCIO v1? I tried inverting the linear_to_sRGB.spi1d but it clips.
What does seems to work is using the srgbf.spi1d LUT from the nuke-default config. That’s 2.2 sRGB however, not BT1886 2.4.
when I import EXR to Premiere, I found the equivalent to preserve RGB checkbox in Ae (that removes the automatic gamma curve) is the gamma correction effect maxed to 28
The equivalent would be 24, not 28, so it is removing 2.4 gamma. But in my experience the gamma correction effect clamps above 1.0, so is not suitable for this purpose.
For Premiere you would need to make a 2.4 gamma to linear LUT, like srgbf.spi1d from the Nuke config which covered a wider than 0-1 range. But why do you need to use OCIOv1 when the Adobe plugin supports OCIOv2?
That’s a great find @yooofi ! In that case one could just use the OCIO config normally with no need for extra input transforms.
TBH, at this point I’m questioning the wisdom of the approach of reading EXRs into Premiere honestly.
I’m thinking a proxy workflow with Premiere, where the ACES look is baked into clips in Rec.709 might be better for offline edit.
This OCIO workflow could work for online edit and conform possibly, but I must say I feel a bit hesitant with Adobe’s way of (not) doing color management, and would really prefer to do that in Avid or Resolve. You’d need to have another program to properly debayer anyway, so why not stay in that program using the EDL/XML/AAF from Premiere. Premiere can then do what it was built for: editing in Rec.709 space.
I am not sure I would try managing OCIO with premiere. but I know that I sometime need to use EXR’s in premiere for offline editing with EXR sources (without bothering with transcoding them first) and having a way to properly see it without forced gamma correction is important. I used gamma correction effect before set to 22/24 which appeared to do the same thing. but good to know that there is the Ae’s preserve RGB equivalent exists also.
From a performance point of view that is a much more sensible approach for offline editing. Haven’t tested it myself but I bet you can’t scrub an .exr clip with OCIO on top of it.
Great find!
Maybe for some cases using EXR in Premiere might still be useful.
Yes, in the very minimal tests I have done Premiere seems to hang a lot with EXR and OCIO, compared to AVID with linked EXR in ACES, which does not seem to have that problem.
