Rec 2020 PQ : right color space in Resolve ?

Hi

Here's a sample of Canon R10 HDR PQ 1080p footage (not mine)
https://we.tl/t-wouw2W0Emz

If i import it in Final Cut Pro, it automatically sets the color space of the footage to Rec.2020 PQ.

In Resolve, i use Color Space Transform, but there's no Rec.2020 PQ option in "input color Space", only Rec.2020. I've tried a variety of different options, but none of them look like what i get with Final Cut Pro.

Any idea how to set the color space ?

Here's what it looks like in final cut pro
https://wetransfer.com/downloads/c55917 ... 502/70fcdb


Thanks

Try using project level Resolve Color Management with the following settings:
Input Color Space Rec.2100 ST2084
Timeline Color Space DaVinci WIde Gamut Intermediate
Output Color Space Rec.2100 ST2084

That works fine for me. Looks pretty much exactly the same in FCP - Rec2020 PQ project

I'm sorry but i don't find these options. Would you mind making a screenshot and sharing it ?
Thanks

Take a look at this, Robert.

HDRProjectSettings.jpg (95.37 KiB) Viewed 15365 times

The settings are very simple as above. If Resolve does not automatically set the Input Color Space to Rec.2100 ST2084 you might need to do that manually but it does for me with your file,

GeneralPrefs.jpg (61.21 KiB) Viewed 15365 times

You also need to make sure your general prefs are set as above for HDR editing and to conform with FCP and Quicktime.

If you are not doing serious color grading you might be better just to stick with FCP - it is far simpler because there are only a few options. That said, the above HDR settings work perfectly.

Take a look at this, Robert.

He doesn't specifically address HDR settings so, while the video is good, it does not answer the OP's question.

It was meant more as an overview of RCM, which Robert seemed to be unaware of. A supplement to your more specific answer.

Thank you Mick !

I forgot to say i want to convert in Rec709. I'm not using an HDR display, i just want to use the extended dynamic range and 10 bit mode that HDR PQ offers.

Also, when i applied your settings, the colors look quite saturated and maybe a bit clipping compared to what i see in Final Cut Pro. See my screenshot. What do you think ?

The attachment HDRProjectSettings.jpg is no longer available

The settings are very simple as above. If Resolve does not automatically set the Input Color Space to Rec.2100 ST2084 you might need to do that manually but it does for me with your file,

The attachment GeneralPrefs.jpg is no longer available

You also need to make sure your general prefs are set as above for HDR editing and to conform with FCP and Quicktime.

If you are not doing serious color grading you might be better just to stick with FCP - it is far simpler because there are only a few options. That said, the above HDR settings work perfectly.

It was meant more as an overview of RCM, which Robert seemed to be unaware of. A supplement to your more specific answer.

Thanks. I think i already saw that vid, that dude is the best.
I have no issue using sample Canon CLog3 footage i've found online. HDR PQ is something else though, and there doesn't seem to be a lot of info online on how to convert it to rec 709 without screwing the image

You probably had tone mapping turned off.
These are the settings I'd use.

2023-07-01 17_27_24-Window.png (31.9 KiB) Viewed 15297 times

Thanks Shebbe, but the results are a bit darker this time.

You can't judge the accuracy of the transform by relatively small variations in the initial appearance of the ungraded image. Which is "right"? Unless the transform is actually clipping data which should be available, you really can't say.

In any event, I would use the settings shebbe provided.

Sorry but i disagree, converting HDR PQ to Rec709 isn't supposed to be a subjective conversion.

When you're converting Clog3 footage to Rec709, it's not subjective either : apply the right color space transformation and you'll have exactly the same result in Final Cut Pro or Resolve or Premiere. I don't see why it would be different with HDR PQ.

"When you're converting Clog3 footage to Rec709, it's not subjective either : apply the right color space transformation and you'll have exactly the same result in Final Cut Pro or Resolve or Premiere. I don't see why it would be different with HDR PQ."

Um ... not my understanding. All transform processes, whether LUT, manual, or algorithms such as tonemapping, apply essentially curves to the transformed image.

There's many ways to do a transform from any X to Y in Resolve. There isn't a perfect or always "absolutely correct" one, just a number that can work depending on your actual field produced footage, and maybe ones you prefer in general or for X specific file in Y project.

This is what I've been trained to understand from the wide array of colorists I've learned from. And makes logical sense from personal experience and testing.

Sent from my SM-S908U using Tapatalk

Yes, you definitely can choose not to use the lut provided by Canon to transform Clog3 to Rec709. I understand. BUT you have the option to if you want, and i think 99% of color graders use this conversion created by Canon.

I only want that SAME option with HDR PQ in Resolve. From all what i've seen so far, the built-in Final Cut conversion is my favorite, but i can't get the same result in Resolve. Now i wonder if it's even possible.

Sorry but i disagree, converting HDR PQ to Rec709 isn't supposed to be a subjective conversion.

When you're converting Clog3 footage to Rec709, it's not subjective either : apply the right color space transformation and you'll have exactly the same result in Final Cut Pro or Resolve or Premiere. I don't see why it would be different with HDR PQ.

It's not a question of being subjective. The end result of the transform depends on the exposure and the nature of the shot. The math of the transform does not take into consideration the individual variations of the exposure or the content. Which transform looks best initially is irrelevant.

And you can't view the result of different pipelines and assume one is more accurate than the other based on what you get in another system, and least of all if you're viewing the material in the windows of different applications. You do know, of course, that the Resolve window is not meant to be color-accurate?

Based on your stated requirements, the only correct settings above are shebbe's. If that doesn't match what you're getting in other systems, then either the preview is not accurate, there are errors in your settings or their processing inconsistencies.

Please, i don't want to turn this into a 10 page thread.
I just like the built-in REC 2020 PQ to rec709 conversion in Final Cut. To my eyes it looks the most faithful, and it probably is since Apple know they stuff.
I just want to know how to achieve this in Resolve
Shebbe's settings are good, but use Rec2100 instead of Rec 2020, and i fear this is what is causing the darker image.
Thanks

For the last time, a "darker" image in itself says nothing about the accuracy of the conversion. And you cannot judge the quality of the conversion in the Resolve program window.

You are mistaken about how this works.

My bad, just checked Canon’s website and as stated by Canon, HDR PQ is « ITU-R BT.2100 & SMPTE ST.2084. »

Final Cut defaults to Rec2020, i thought this was correct as the picture looks great, but if Canon is to believed, it’s wrong and Rec2100 should be used.

Therefore Shebbe’s settings are accurate. Sorry again, but i thought i could trust apple on this one. Though i am not an expert and maybe there’s not difference between Rec2100 and Rec2020 .

Rec2020 is a colorspace and doesn't directly reference a transfer function, you can have Rec2020 colorspace with 2.4 gamma. ST 2084 is a transfer function (gamma). They're combined under Rec2100.

An accurate conversion would result in clipping as PQ is a larger space than 709. Once tone mapping is invoked the idea of correct goes out the window. DaVinci tone mapping itself is only a preset. The closest thing you'll get to a cross platform standard is probably ACES. The fact that one of DaVinci's tone mapping functions is called "DaVinci" should be a clue that it's the software developers interpretation and not any external standard.

Good Luck

An accurate conversion would result in clipping as PQ is a larger space than 709.

Sure, if you leave it outside the bounds of rec. 709 when you export. But who works that way? The transform to rec. 709 ouput doesn't break the out-of-bounds data. It's still there. It just needs to be brought down into the more restricted color space.

Sorry but i disagree, converting HDR PQ to Rec709 isn't supposed to be a subjective conversion.

When you're converting Clog3 footage to Rec709, it's not subjective either : apply the right color space transformation and you'll have exactly the same result in Final Cut Pro or Resolve or Premiere. I don't see why it would be different with HDR PQ.

I just like the built-in REC 2020 PQ to rec709 conversion in Final Cut. To my eyes it looks the most faithful, and it probably is since Apple know they stuff.
I just want to know how to achieve this in Resolve

Others already mentioned this but if you want the non-subjective version of PQ to Rec.709 Gamma 2.4 set the tone mapping method to none and cry at the blown out highlights. What you will then see is the pure mathematical transform from the color primaries Rec.2020 to Rec.709 which will not induce any chroma clipping if al information inside the image was already within Rec.709. But the dynamic range of the HDR signal to a pure 2.4 gamma SDR display will clip anything brighter than '100nits'. What you see in FCPX is simply Apple's version of (subjective) tonemapping. Similarly Premiere Pro recently introduced their own tone map system albeit a very rudimentary one at the moment.

What Resolve does with DaVinci tone mapping can be different just like camera vendor supplied conversion LUTs.

Trust me in that the image is properly converted to the target display and you want the tonemapping for a matching appearance. You will just have to apply some extra basic grading before or after the transform to get it to sit at a point where you are happy with it.

Thanks for all the info everyone.

Looking forward to get my Canon R10 soon and record some better footage for my tests.

If you want pure math then end result will be useless, unless you want to do a lot of work on it.

If you want better result than Resolve tone mapping (which is fixed math per timeline) run DV analysis and use DV tone-mapping which will be per shot (if you have cuts). End results are much better, but still not perfect and you need to work per shot to make it as you want. It's all subjective, not per 'fixed math' as eg. conversion from Rec.601 to Rec.709. There is no 'better' way.

Got my Canon R10 today. In my first tests, i compared HDR PQ video with a RAW photo file, same shot, same exposure, same light... I "printed" the raw file using the default Apple Preview settings.

With Shebbe's settings for HDR PQ, they're insanely close. For Tone-Mapping, turns out "Luminance Mapping" is much better than "DaVinci", which turns red into orange. "Saturation Preserving" looks too saturated to me.

The R10 is a great deal if don't want to spend the extra $$$ for C-Log but still want 10 bit.

Turns out applying these project settings gives even better results and no need for a Color Space Transform node...

There is no difference in Resolve Color Managed or manual with CSTs provided you'd use the same settings on a CST. The preset uses DaVinci as the tone mapping method by default and the file's input space gets automatically tagged as Rec.2100 I believe.

I would personally stay away from global project management since this can have potential issues with other workflows like compositing with Fusion or adding graphics/titles.

The preset uses DaVinci as the tone mapping method by default and the file's input space gets automatically tagged as Rec.2100 I believe.

It's possible -- Resolve can read the input color space of some log formats -- but I wouldn't bet on it. Here's guessing the OP is under the mistaken impression that selecting 2020 PQ as the color processing mode is accomplishing the same thing, which of course it isn't.

There is no difference in Resolve Color Managed or manual with CSTs provided you'd use the same settings on a CST. The preset uses DaVinci as the tone mapping method by default and the file's input space gets automatically tagged as Rec.2100 I believe.

I would personally stay away from global project management since this can have potential issues with other workflows like compositing with Fusion or adding graphics/titles.

Yes on my Mac the input space of the OP's sample gets set to Rec.2100 ST2084 automatically by Resolve. I would think that is the simplest way to go. It gives very close (identical) matching to what you get in Final Cut Pro where the color space is also set automatically to HDR. I personally really like using project level RCM but then I only use Resolve for grading.

Allright, i was testing on a fairly low contrast scene, and didn't notice the red shifting to orange. Therefore, i tweaked the project settings to "custom" and choose "luminance preservation".

Here's a new download :
https://wetransfer.com/downloads/328640 ... 247/bf505a

Using custom project settings gives a brighter image that seems to be clipping a bit in the highlights of the red channel, but it can be recovered by lowering the offset or gain, the information is still there.

Using CST clips at 0, especially on the red channel and it cannot be recovered, the information is lost (see my screenshots). Using a node before CST, i didn't manage to "recover" the information properly either. Or maybe the scope is fooling me ? I don't know.

Apart from these brightness and clipping differences, i guess the colors are indeed the same, but using project settings seems to be the better option.

What do you think ?

I don't know how you set your CST method up but if you were grading after the CST you are grading with WYSIWYG meaning if something is 'crushed down' to SDR it's near impossible to 'recover' anything. You should try your grading operations before the CST conversion to display happens.

Nothing wrong with using Color Managed though if you prefer that, just keep in mind the possible limitations I mentioned previously. If all you're doing is color grading it won't be much of an issue.

There is no difference in Resolve Color Managed or manual with CSTs provided you'd use the same settings on a CST. The preset uses DaVinci as the tone mapping method by default and the file's input space gets automatically tagged as Rec.2100 I believe. I would personally stay away from global project management since this can have potential issues with other workflows like compositing with Fusion or adding graphics/titles.

That is 100% our experience also. Well-said.

I often say, the power of Resolve is that there's often two or three different ways of accomplishing the same thing. In this case, you have ACES or RCM2 or CST nodes... and none of them is necessarily wrong. Just different. It's a question of individual choice and how you prefer to work.

I don't know how you set your CST method up but if you were grading after the CST you are grading with WYSIWYG meaning if something is 'crushed down' to SDR it's near impossible to 'recover' anything. You should try your grading operations before the CST conversion to display happens.

Nothing wrong with using Color Managed though if you prefer that, just keep in mind the possible limitations I mentioned previously. If all you're doing is color grading it won't be much of an issue.

Thanks !