Blackmagic Forum

I'm very confused working with UMP 4.6k RAW clips 4608x1920 24fps. The UG is not very helpful. This is in 12.5.5.026 Studio:

If I set what would seem to be correct from the UG:
Color Science to DaVinci YRGB Color Managed,
Camera RAW decode using Camera Metadata
Input Color Space to Blackmagic Design 4k Film v3 (Everything shot Film)
Timeline Color Space to Rec 709

I get recoverable highlights blown out to smithereens on the timeline and in the Media Pool.

If I set Input Color Space to Bypass, the highlights are corrected and not blown out where recoverable. But according to the UG 'Bypass' means that the input color space is the timeline's. That makes no sense as it would seem to imply that all input sources are transformed with the same math.

Similarly, with an R3D Dragon clip, all the different Dragoncolor input color spaces show the same blown out readings on the parade scope. Setting Bypass brings everything into range.

And the same thing happens with BM Prores files. They are also not blown out on 'Bypass' nyt are on 'Blackmagic Film'.

So it would seem that either the description in the UG is somehow wrong, OR the workflow is buggy and the input color space is not yet working properly, OR the intended workflow is exactly as I'm seeing it, working magically for for all varieties of source files, in which case why have different input settings at all?, OR I'm just just understanding the correct workflow for DaVinci YRGB Color Managed.

Do you have to use color management?

I have a project with all kinds of source material: Red Dragon, Varicam35 VLog, GH4, BM DNGs, and EXRs. BM Color Management as described in the UG would be ideal, if it works.

Have you considered switching your color science from DCM to ACES? I've had good results with that when working with mixed camera projects. The controls don't respond in quite the same way, but overall I found it beneficial.

I suppose I could switch to ACES but the BM Color Management simplifies a lot of the problems with input and output ACES transforms. OTOH the two responses so far seem to be saying that BM Color Management doesn't work (yet). Is that right?

David Cherniack wrote:I'm very confused working with UMP 4.6k RAW clips 4608x1920 24fps. The UG is not very helpful. This is in 12.5.5.026 Studio:

If I set what would seem to be correct from the UG:
Color Science to DaVinci YRGB Color Managed,
Camera RAW decode using Camera Metadata
Input Color Space to Blackmagic Design 4k Film v3 (Everything shot Film)
Timeline Color Space to Rec 709

I get recoverable highlights blown out to smithereens on the timeline and in the Media Pool.

If I set Input Color Space to Bypass, the highlights are corrected and not blown out where recoverable. But according to the UG 'Bypass' means that the input color space is the timeline's. That makes no sense as it would seem to imply that all input sources are transformed with the same math.

Similarly, with an R3D Dragon clip, all the different Dragoncolor input color spaces show the same blown out readings on the parade scope. Setting Bypass brings everything into range.

And the same thing happens with BM Prores files. They are also not blown out on 'Bypass' nyt are on 'Blackmagic Film'.

So it would seem that either the description in the UG is somehow wrong, OR the workflow is buggy and the input color space is not yet working properly, OR the intended workflow is exactly as I'm seeing it, working magically for for all varieties of source files, in which case why have different input settings at all?, OR I'm just just understanding the correct workflow for DaVinci YRGB Color Managed.

Have you tried other tool (Scratch etc)?
Maybe clips are badly recorded. If not some Resolve math is wrong.

There's nothing the matter with the files. There may not be anything the matter with BM color math. It could be I'm not doing something right with the complexity of the set-up.

David: I forgot to say earlier that I had an identical experience with properly exposed Alexa mini footage and DCM with blown-out, but recoverable highlights. When I posted about it on this forum the general response was that it was behaving as expected, which I found pretty confounding. Sorry that isn't helpful, but you're not alone in this experience.

Thanks, Jordan. Let's see whether I get a similar response from anyone in BM support. It's hard to believe that it's behaving as expected but perhaps I'm expecting too much....that it should work as advertised

Ideally you start with native footage (no de-log LUTs etc) and do magic by yourself

IDEALLY, Blackmagic would have the color math working for their own raw and prores clips. I'm more inclined to believe that the problem lies in confusion about the settings

As a fresh Resolve user, I'm also confused with color management but not with RCM but with Timeline Color Space option being available for the "old" DaVinci YRGB color science...since it is display reffered, what would be the use of the defining a working color space when it is actually the color space of the monitoring device and you still need to set the output LUT that strictly defines the IDT-ODT transform?

Highlights will be blown out because they are highlights and we try to look at them on a low-dynamic range monitor. If we take that the 100% diffuse reflector is value 1.0, then everything shiny or emissive will be higher than that. Its just the way it works if camera is capable of recording more than 6-7 stops. It is also the same case with ACES, it is not a bug, it is a feature.

Bypassing the input transform means that you are looking at image data in form that is not actually meant for viewing. They are in scope range because they are stored as limited range integer values and that is fit into scope range with no clipping. Log files are not "flat", they are log encoded and log encoding is not a presentation form. But as it is highly desirable to have control over the machine not another way around, we can happily grade the log image as if it were an ugly-looking normal image without ever doing the appropriate transform from log to linear. This is how it works with no color management.

David Cherniack wrote:IDEALLY, Blackmagic would have the color math working for their own raw and prores clips. I'm more inclined to believe that the problem lies in confusion about the settings

Are you using Clip Metadata for RAW clips in RAW settings (or default ones) ?

David Cherniack wrote:I'm very confused working with UMP 4.6k RAW clips 4608x1920 24fps. The UG is not very helpful. This is in 12.5.5.026 Studio:

If I set what would seem to be correct from the UG:
Color Science to DaVinci YRGB Color Managed,
Camera RAW decode using Camera Metadata
Input Color Space to Blackmagic Design 4k Film v3 (Everything shot Film)
Timeline Color Space to Rec 709

I get recoverable highlights blown out to smithereens on the timeline and in the Media Pool.

You need to set the output color space as well. The timeline color space is more a matter of preference.

The fact that the high dynamic range of the UMP does not fit into the narrow Rec709 range should not come as a surprise. It is up to the color grader to make it look good in Rec709. It means that first you would have to set your black levels with Lift, pull in the highlights by using Gain, adjust Gamma and then you would have to decide on how to roll-off the blacks and highlights and fine tune.

That first step should take no longer than a few seconds once you are used to it.

If you do not want to do that process yourself you could use LUTs that make those decisions for you but you would lose control. Obviously each video is different and may need different adjustments. Also with LUTs there may be clipping concerns or full/limited range incompatibilities. Also unlike transforms LUTs use interpolations.

What you see is definitely a feature.

The RCM definitely does not work with RAW (at least CinemaDNG). If you own a BMPC4k camera do the following test to see it for yourself:

1. In your timeline, import a BMPC 4k Prores clip filmed at "Film" setting.
2. Then import a CinemaDNG clip from the same camera.
3. In RCM set any XYZ space as an input leaving the rest of the settings at Rec709 2.4: The Prores goes bonkers as it should, the RAW looks normal, but washed out (I believe the gamma is applied twice).
4. Set Input to "Blackmagic Design Film 4K" and leave the rest at Rec709 2.4: The RAW is still washed out (unaffected by the change in the input setting), while the Prores displays correctly.
5. Now select Rec709 (scene) as your output (Rec709 (scene), I believe, is a Rec709 gamut with linear gamma, but with lightness confined between 0 - 1 values, a "display linear" color space). The RAW is displayed correctly, the ProRes is too dark.

Go explain that.

PS. There are other bugs with RCM. Change the "Timeline Color Space" a few times and check how the "Max Timeline Luminance" responds: After a few changes, it stops reflecting the color space´s max luminance (and the monitored image stops responding to changes entirely)