Color levels a bit off in final DCP

[peter]

Hi all,

Running into an issue with color levels being a bit off in the final DCP.. vs the source Quicktime that was used to make the DCP. I'm pretty sure that in the past when I used DOM to make DCPs.. I didn't see this issue.

Here is my workflow. Graded everyting in DaVinci Resolve (v18.0.3). Exported a DNxHR HQX Quicktime to use to import into DOM to make the DCP. I pretty much used the default settings. In Resolve I selected 'DNxHR HQX 10-bit'. I believe by default.. it exports at video range (vs full range).

Brought that Quicktime into DOM and mostly left things on the default settings. Color-rec709, Range-Video. Made a 2K Scope Interop DCP. The DCP played in a few theaters. Played ok.. but the Director commented that the picture looked a bit lacking in saturation and contrast. I explaned that each theater and projetor might be a bit different and we didn't have issues in the past.

But I did bring into Resolve the DCP.. and also the Quicktime that was used to make the DCP. I put both into a timeline and switched between the two sources. They were not the same.. I would have expected them to be. The DCP looks to have a slightly raised gamma... vs the source Quicktime file. That seems odd.

Any idea why that might be? Any of my workflow seem wrong or anything I might have done wrong? In DOM.. I did look at the Properties of that source Quicktime file. Most looked ok.. but these caught my eye:
Colour range - Limited (4-58)
Colour primaries - BT709
Colour transfer characteristic - Unspecified
Colourspace - BT709

What is 'Limited (4-58) Colour range? I had not heard of that before. Shouldn't that say Video (64-940)? Or does that all seem ok from what you see? Any thoughts on this would be helpful. Oh.. I used DOM 2.16.28 to make this DCP. Thanks!

[Carsten]

Does it really say '4-58'? That's odd. Although the video properties display in DCP-o-matic is not always accurate. Doesn't necessarily mean that DCP-o-matic is misinterpreting actual values during conversion.


I think this could be a gamma 2.2 vs. 2.4 issue. Try the BT.1886 colour conversion setting in DCP-o-matic instead of rec.709.

Yes, 'limited' is the same as Davincis 'video' range (as opposed to full/data range).

[carl]

That incorrect range in the properties is a bug in the properties window.

[peter]

Thanks guys for checking into it. Would that bug have affected the color space conversion and caused the final DCP to have different video levels from the source Quicktime? Or is that just a display issue on the clip properties? Once the next version is out.. I can create the DCP and see if the video levels (gamma) is different from the DCP that I have already made.

-Peter

[Carsten]

That's just a display issue. A color conversion actually based on 4-58 would be completely broken.


Try BT.1886

[peter]

Ok gotcha. I'll try that BT1886 setting and see how the resulting DCP compares.

thanks.

-Peter

[peter]

That's just a display issue. A color conversion actually based on 4-58 would be completely broken.


Try BT.1886

Sorry. To clarify. Do you mean 'Rec. 1886'? I don't see an option for BT.1886. This is the 'Video - Colour' setting right?

thanks,
Peter

[Cfoxv1980]

Yea. I had similar issues with it being washed out looking

Set it to Rec. 1886. It seems to be correct if your in a gamma 2.4 session

[peter]

Thanks All. So setting it to Rec. 1886 for Video-Colour fixed the issue. The resulting DCP matches the original Quicktime source. Or at least it's very close. There are some very small variations when looking on the waveform scope.. but visually they look the same.

I will likely use this setting moving forward. Seems like something might have changed in DOM. I'm pretty sure I always set it (or just left it on) Rec. 709 and the exported DCP matched the source Quicktime file.

Question. Should there be a separate Video-Colour and Video-Gamma setting? Isn't the color space still technically Rec. 709 and the gamma curve is the BT.1886? But maybe I'm wrong on that.

I'm not sure if there is a better workflow I should be following (like exporting a DNxHR .MXF file.. or using DPX files instead of Quicktime files?), but I'll stick to Rec. 1886 for my color setting for most of my projects which follow this more standard color workflow. Thanks again everybody.

-Peter

[Carsten]

The color conversion of Rec.709 vs. Rec.1886 is the same. The gamma is the only thing that changes. There is a bit of confusion in the camera/display/editing/color correction world about Rec.709 and gamma. Formally, Rec.709 never had a spec'd display gamma. It was later assumed to be 2.2. Rec.1886 (BT.1886 is the same) explicitly added a gamma of 2.4. Rec.709 was conceived in the early nineties for analog HD video. Practically speaking, one would nowadays suggest Rec.709 with a 2.2 gamma for web video. content that is e.g. watched on computer monitors, tablets, smartphones, etc., in a lit environment. Whereas in a home theater , 'movie watching' environment, the gamma 2.4 of Rec.1886 is better suited. And in optimized environments like a dark cinema auditorium, DCI 2.6 is used.

Per default, DCP-o-matic uses Rec.709 for HD sources. One could argue wether that should be changed to Rec.1886. I would think that e.g. screen ads would usually work better with Rec.709, vs. shorts and features using 'dramatic' lighting and exposure that would probably more often suit Rec.1886 better.