DCP has wrong gamma

Anything and everything to do with DCP-o-matic.
dhl
Posts: 24
Joined: Sun Apr 10, 2022 6:17 pm

Re: DCP has wrong gamma

Post by dhl »

Hi Carl and Carsten,

Our test in the cinema went well though there were some differences of opinion on which version looked best. I decided to dig a bit deeper and had a question about the DoM Gamma Output Correction checkbox:

Screen Shot 2022-09-11 at 7.50.52 PM.png

This checkbox specifies inverse gamma 2.6 correction on output.

I did a test comparing gamma settings in the Screen player and noticed some gamma differences between my ProRes masters and DoM's converted file:

DoM-output-gamma.png

The DoM file is gamma 2.6 (comparing Match Quicktime and gamma 2.6 shows the same gamma)
When I set the DoM file to gamma 2.4, it matches my masters which are at gamma 2.4

I see that 2.6 gamma is normal for DCPs but shouldn't the DoM conversion process result in a file that matches my gamma 2.4 master?

Thanks!
You do not have the required permissions to view the files attached to this post.
Last edited by dhl on Mon Sep 12, 2022 5:52 am, edited 1 time in total.
Carsten
Posts: 2663
Joined: Tue Apr 15, 2014 9:11 pm
Location: Germany

Re: DCP has wrong gamma

Post by Carsten »

The gamma 2.6 for the DCP is targeted at a theatrical viewing experience - a dark room, and low screen luminance (48cd/sqm peak white).

It does not make sense to compare the conversion results all on a direct view screen. One would need to compare e.g. a 100nits direct view screen at gamma 2.4 with an actual cinema auditorium at 48cd/sqm and gamma 2.6. This is of course not easy to do practically. One could only focus on certain luminance and color features of specific scenes in order to get an idea what might be lost or preserved.
dhl
Posts: 24
Joined: Sun Apr 10, 2022 6:17 pm

Re: DCP has wrong gamma

Post by dhl »

Carsten wrote: Mon Sep 12, 2022 10:24 am The gamma 2.6 for the DCP is targeted at a theatrical viewing experience - a dark room, and low screen luminance (48cd/sqm peak white).

It does not make sense to compare the conversion results all on a direct view screen. One would need to compare e.g. a 100nits direct view screen at gamma 2.4 with an actual cinema auditorium at 48cd/sqm and gamma 2.6. This is of course not easy to do practically. One could only focus on certain luminance and color features of specific scenes in order to get an idea what might be lost or preserved.
Thanks, makes sense!
jamielejeune
Posts: 1
Joined: Tue Sep 13, 2022 4:34 am

Re: DCP has wrong gamma

Post by jamielejeune »

Carsten wrote: Sat Aug 13, 2022 2:27 amThere is a long story about the relation and transition between rec.709 and rec.1886/BT.1886. We have both in DCP-o-matic, the essential difference is indeed the gamma 2.2 vs. 2.4. rec.709 actually does not have a spec'd display gamma at all (they were just assuming the default CRT gamma back in the early nineties). Some say, when today an imaging product indicates 'rec.709', it is actually BT.1886. BT.1886 finally brought the proper gamma spec, and one could say that the lower gamma (rec.709 and sRGB) are targeted towards standard daylight/office viewing environment (web video), while BT.1886/gamma 2.4 is targeted towards an improved 'home cinema' viewing experience. And finally, 2.6 is for a dark room, a cinema, an optimized theatrical environment. So, it doesn't surprise me that todays video editing applications use 2.4. Davinci, of course, allows for more detailed settings.
- cars-ten
While the sRGB spec describes separate encoding gamma and a display gamma (the latter being a 2.2 gamma), the BT.709 output metadata tag was a problem from the start since BT.709 does not specify a display gamma at all. It only ever specified the camera capture. As you noted, the display was simply assumed to be CRT. That's why BT.1886 had to be created and codified after display technology moved on to LCD screens that didn't have an implicit gamma.

A difficult wrinkle is that Apple's color management simply uses the inverse of piecewise camera capture gamma, so the result is an implied output gamma of 1.96 on any file with the transfer function tagged as REC709. And yeah, that's silly, but it is how all Apple devices color management reads the REC709 transfer function tag. If DCP-o-Matic is calculating REC709 tagged files from an assumed 2.2 gamma, that should be stated in the menu and documentation more clearly, since for those who are bringing in files correctly color managed for Apple devices (for example, those who color graded in the GUI of FCP), a 2.2 gamma interpretation of the source will deliver an output result that is different than it should be.
Last edited by jamielejeune on Tue Sep 13, 2022 5:40 am, edited 1 time in total.
Carsten
Posts: 2663
Joined: Tue Apr 15, 2014 9:11 pm
Location: Germany

Re: DCP has wrong gamma

Post by Carsten »

Yes, the Apple gamma handling is an ongoing source for confusion. Maybe it would help if the DCP-o-matic GUI would indicate the gamma with the color conversion preset, e.g. 'Rec. 709 (Gamma 2.2)'. For those who don't bother entering the custom conversion settings, it can been seen there.

- Carsten
Bildschirmfoto 2022-09-13 um 13.50.04.png
Bildschirmfoto 2022-09-13 um 13.50.17.png
You do not have the required permissions to view the files attached to this post.
Alexander055
Posts: 3
Joined: Tue Apr 11, 2023 6:03 pm

Re: DCP has wrong gamma

Post by Alexander055 »

I still don't understand if it is normal that the dcp should be clearer. I would like it to be exactly how they exported it to me from davinci.

Couldn't someone tell the whole workflow starting from the davinci exported proRes file?

Settings and color space to export to davinci
settings in DOM (color, range)

Basically the settings should always be the same.

Also, how come I don't see the option to set it to 2k full?
Thank you very much
Carsten
Posts: 2663
Joined: Tue Apr 15, 2014 9:11 pm
Location: Germany

Re: DCP has wrong gamma

Post by Carsten »

2k and 4k full containers are not 'allowed'. Use them only for test/setup purposes, not for productions.

DCP is strictly for cinema. You need to watch it in a cinema and check wether your levels are okay there. It doesn't make much sense to compare the same footage on a 100-300nits direct view display in rec.709/rec.1886 with a P3/gamma 2.6 projection in a dark cinema room.
Alexander055
Posts: 3
Joined: Tue Apr 11, 2023 6:03 pm

Re: DCP has wrong gamma

Post by Alexander055 »

Thank you very much for the quick reply.

1. I understand about the full format, whether or not you choose, DOM doesn't allow it. I was wondering why other software does. So I wanted to make sure it was a DOM prerogative.

2. I realize that, being PC monitors, there is a different color space. But it seems strange to me that different software to create dcp, have all given different results. Including DaVinci.
Given that I'm going to try it at the cinema, I wanted to know the correct DOM setting starting from a video file, with the specifications shown in the attachment.

I apologize if this may seem obvious as a request, but I really need your help. Thank you very much for your availability.
You do not have the required permissions to view the files attached to this post.
Carsten
Posts: 2663
Joined: Tue Apr 15, 2014 9:11 pm
Location: Germany

Re: DCP has wrong gamma

Post by Carsten »

You CAN use full container, ig you enable it in advanced prefs. But as I said, it is not 'allowed', that is, recommended, because, practically, no cinema has presets for it, so it doesn't make sense to use it for a DCP, unless you can make sure that the specific screen you are mastering for supports that format.


There is some ambiguity between rec.709/rec.1886 and gamma 2.2 vs. 2.4. Depending on your monitor setup, larger differences may be visible.

- Carsten
Alexander055
Posts: 3
Joined: Tue Apr 11, 2023 6:03 pm

Re: DCP has wrong gamma

Post by Alexander055 »

leaving aside my monitor and the fact that I can't know the truth, is there a specific workflow to follow, starting from a prores file like the one attached in the previous message?
What are the correct DOM settings when importing such a file?

A thousand thanks