JPEG2000 from Davinci10 VS JPEG2000 from EasyDCP Creator

[Peter Chamberlain]

Hi Robin, can you confirm the exact version of V10 that you tried, and what colour space was selected in the look up tables tab in the project settings. This will define the LUT used for timeline to XYZ conversion prior to JPEG encoding.
Peter

[Robin Erard]

Thank’s for your answer.

I have just checked output lut and I had two option Rec709 2.4 to XYZ or Rec709 2.2 to XYZ… I changed for REC709 2.2 and now it’s similar between DaVinci and EasyDCP. But why this difference, Rec709 isn’t 2.2 anyway ?

Best
Robin

réalisateur, scénariste, monteur, étalonneur
http://www.robinerard.ch

[Andi Winter]

as far as i know in the recommandation 709
there is no strictly defined gamma, therefor
one has to chose the gamma of the monitor
which was used during grading.

so i think the gamma options are quite
necessary, but poor if there are only this
teo options…

[Juan Salvo]

[Andy Winter] “as far as i know in the recommandation 709
there is no strictly defined gamma, therefor
one has to chose the gamma of the monitor
which was used during grading.
“
There is an implied gamma, which has been interpreted as 2.2 for non-crt display. For a long time this was the defacto standard.

Now there is an explicit standard Bt.1886 which sets the gamma at essentially 2.4. Of course getting the whole world to switch gamma standards isn’t easy. And the issues of what gamma in which situations is itself somewhat tricky.

So your advice Andy is actually the right way to go. Find out what the gamma was set at in the grade… and it helps to know explicitly what the room was set to as well.

At the end of the day, the best results for theatrical exhibition is to grade in a theatrical environment.

https://JuanSalvo.com
https://theColourSpace.com

[Evan Kultangwatana]

i’ve heard folks make references to “surprises” when converting a rec709-graded master to a DCP shown in a cinema environment. but it’s not always clear to me if this was because of a technical error in the conversion or a matter of actual colorspace/gamma/brightness issues.

anecdotally, what differences can one generally expect to see going from a 35fL rec709 2.2/2.4-gamma environment to a 14fL DCP projected environment? assuming that there is no technical issue in the conversion, of course.

given that you need a linux box and the right i/o and display hardware to even properly see P3 colorspace, i’m not sure what the “second-best practice” is for those of us working with decklinks on macs and PCs.

[Juan Salvo]

Doing a pass in a theatre might be a good idea.

Surprise wise assuming a technically correct conversion, there’s usually potential issues with black levels, and sharpens, detail, over all brightness. It’s a really different experience watching on direct view display and reflective projection, so lots of things can feel different.

https://JuanSalvo.com
https://theColourSpace.com

[Peter Chamberlain]

Glad you see the difference is illuminated at 2.2 gamma.

The original easyDCP was set at 2.2 gamma as many monitors used for creating DCPs were computer monitors. We added the 2.2 option to permit consistency for those who still use the computer monitors but 2.4 gamma is correct if you are grading on a correctly calibrated Rec.709 monitor.
The other conversion options relate to other monitor config/calibrations.

Peter

[Joakim Ziegler]

There are many potential issues, and this conversion is generally not one that can be done 100% with a LUT without manual adjustments.

The difference in brightness and gamma is the most obvious, but the other big difference is color temperature. rec.709 is 6500 Kelvin, while DCI is 5800 or thereabouts. If you convert using a LUT with no white point compensation, the result will look fairly similar to your video version, but your pure whites will no longer be pure white in the P3 space, they’ll be quite magenta compared to the actual pure white of the space (and also compared many other DCPs). If you use a LUT with white point compensation, the pure whites will remain pure, but things tend to feel a bit greenish. We generally convert using a LUT with white point compensation, then do some manual color adjustments.

On top of that, things tend to require a slight saturation adjustment, depending on the material and how it’s graded.

In summary, you can’t expect the built-in LUTs in Resolve to fix this for you. You need to monitor in DCI-P3 or DCI-XYZ somehow, apply a LUT, and adjust, to get the best possible results.

—
Joakim Ziegler – Postproduction Supervisor

[Terrence Meiczinger]

Another complicating issue is that there are different ways to calculate the gamma linearization. Some applications use a simple calculation, while others use a complex calculation. Mixing the two methods can result in black levels being increased or decreased at the low end, mimicking a gamma shift.

[Joakim Ziegler]

Well, gamma is gamma, and there’s only one way to calculate actual gamma. However, some people get confused and use the rec.709 camera encoding curve to decode material, or assume that’s what should be used to linearize rec.709 material, which will indeed change things up. This is completely wrong, though, you should never use the rec.709 curve unless you’re a camera manufacturer. Everything else is (de facto) straight gamma 2.4.

—
Joakim Ziegler – Postproduction Supervisor