Blackmagic Forum

Hello all. I am delivering a series of short-ish HDR videos with a slightly odd review issue.

Since the client can't currently come do a session with us they've come up with system for doing color reviews that they would like us to follow. Basically I do the grade as per notes and then they all review separately on calibrated XDR monitors that they all have in the HDR Video reference mode. Then the idea is they would send me another batch of notes and we're all set.

The issue I'm having is i cannot get the quicktimes out of resolve to display correctly in this mode. I feel confident that the underlying data is correct and in fact when the XDR is hooked up via Blackmagic 8k to Teranex to the XDR in HDR video mode (P3-ST 2084 1000nt) it looks great in both Resolve and an exported tiff sequence to flame. The same project run out to the Sony BVM looks nearly identical with just better roll off at the high and low ends, color is pretty damn close really.

However the quicktimes in finder to XDR in this mode look like they're missing some sort of gamma curve, the low end is fine but then blows out at the high end in a fairly progressive fashion. The same quicktimes will display perfectly fine in the Pro Display XDR Mode (P3-1600nt) but that doesn't help me much.

The project is color managed and set for Rec2020 with the gamut constrained to P3 and the gamma to ST2084, Max Timeline Lum is set to 1000nt and for good measure i eventually threw in a color transform to cut everything off at 1000nt while i was experimenting.

I've tried tagging the quicktimes ST2084, ST2084 1000 nit, and Rec.2100 ST2084. No difference.

Has anyone seen this or have any ideas on how to proceed?

Not sure if I understand.
If you use custom XDR profile then QT preview is very wrong, but when you use built-in profile preset then it's close to be correct?

Put your exported file through mediainfo and check what HDR metadata does Resolve write.
From my quick test QTX will use HDR metadata for its mapping engine. If you have wrong HDR metadata written then this will affect final output.
I assume screen calibrate profile is done to 1000 nits?

They're actually both built in reference modes, nothing custom. One is supposed to be for using it as a reference monitor for HDR and one is for using it as a computer monitor. The quicktimes work in the computer screen mode, but not in the HDR Ref mode.

Mediainfo reports what I would expect, it looks tagged right. Primaries BT.2020, transfer is PQ, and matrix is BT.2020 non-constant. Which looks right to me, although I don't know what the "non-constant" part means.

Non-constant bit if fine.
Do you have mastering display metadata as well?
You need to have something like this:


If your blacks/color etc. looks okish, but highlights are blown then it's most likely wrong metadata.

If I take eg. 1000nits file and change metadata to eg. 500nits then QTX preview will have highlights clipped. This is exactly what you are seeing (I'm monitoring as SDR, but I think this is irrelevant).

You need to set 'HDR mastering for 1000 nits', enable HDR10+ in color management settings and during export tick Embed HDR10 metadata.

Ohh, that's really interesting. I've always been delivering dolby vision so i've never bothered turning on the HDR10+ stuff, I never would've guessed that's where quicktime would be getting that info from. HDR Mastering was already set for 1000nit.

So I now that I have all the appropriate mastering info in there but sadly the result is exactly the same.

The only thing that looks at all odd to me is for Mastering Display Color Primaries it has Display P3 and I can't imagine where it's getting that from, but i wouldn't think that's my issue.

GreyCarl wrote:The project is color managed and set for Rec2020 with the gamut constrained to P3 ...

This is where it gets P3 and this is the way how it should be.


PQ curve is fixed, so 1K nits is always 768 for 10bit signal. Depending how XDR engine works your 1K nits master may be displayed 1:1 for desktop mode or it's mapped to 1.6K nits.
Does XDR desktop mode stretches highlights to 1600nits for your file? Is it brighter than when monitoring over Teranex ?

"HDR Video (P3-ST 2084)
Use this mode for 4K or ultra high-definition video production workflows up to 1000 nits (full-screen sustained) using the wide color P3 primaries and the high-dynamic-range SMPTE ST-2084 EOTF. This mode is designed for controlled viewing environments set up per ITU-R BT.2100."

Also from FCPX guide:
"In the Displays pane on the Pro Display XDR, click the Presets pop-up menu and choose HDR Video.
When you choose this preset, Final Cut Pro turns off tone mapping to provide the maximum color and brightness accuracy for HDR video playback on the Pro Display XDR (but maybe not when used with just QTX?). The color and brightness is accurate for video content measuring up to 1000 nits."

Your video is graded to 1000nits, metadata correctly describes video as 1000nits peak, apparently any mapping turned off, so why is there any clipping going on? This is strange.

As a test download:
https://github.com/da8eat/VideoEditorIn ... atcher.dmg

load your MOV in Metadata Editor and overwrite max master luminance to 10K. Hit Apply at bottom right corner and then open file in QTX. Does it still clip whites? Try the same also with eg. 600 instead of 1000 (should clip way stronger).

What OSX are you on?

As workaround try using FCPX (but then everyone needs FCPX) to play file in HDR project, instead of QTX. This may work.

I'm not sure this solution will help anyone else, but here it is:

After doing some troubleshooting with our remote clients it turns out the issue was just in my setup, even though I had earlier tested that somewhat by plugging the XDR into a laptop and having the same issue.

So the fix was... unplugging the XDR from the 3rd port in the graphics card and plugging it into the 2nd. Since this issue is both intermittent and followed it from one box to another I'm guessing the cable is funky. In fact while talking to an apple engineer he mumbled something along the lines of "yeah the cables are a bit of weak point with the XDR."

So not an exciting resolution, but something.

Thank you for all the help Andrew!