Ursa Mini 4.6k G2 and Davinci YRGB Color Managed

I recorded some footage on the Ursa Mini 4.6k G2, just simple 4k UHD using Prores 422.

I brought it into Resolve (17.4.6) with Davinci YRGB Color Management. I just noticed that Resolve sets the input color space as "Blackmagic Design Pocket 4k Film Gen 4" rather than any of the "Blackmagic Design 4.6k" options.

The metadata for the clip states that it is "Blackmagic URSA Mini Pro 4.6K G2, Color Science Gen 4" which isn't listed as an option within Resolve.

Is the "Pocket 4k Film Gen 4" color science the same as the G2? Is it just defaulting to something similar since the specific 4.6k G2 gen 4 color science is missing from Resolve?

Erik

The thing they have in common is they both can shoot Gen 4, but most people using the BMPCC4K would be on release 7.3 and shooting Gen 5. The sensors are quite different. If you switched from RCM to ACEScct version 1.3, I wonder what you’ll see.

The description of colour management in the Resolve 18 manual is good and you might want to download the manual to see if that helps.

That’s likely a bug. RCM should ID those G2 clips as 4.6K Film Gen 4. You can simply set it manually to that by right clicking on them in the media pool or in the color page thumbnail timeline. The automatic identification of input gamma + gamut is still a work in progress. (And BMD really should replace the use of “gamma” and with the more accurate term transfer function.)

Is not the automatic identification of camera only for raw (any camera raw?) Then that can be over-ridden in the raw panel or individually for clips by selecting their color space. Anyways the OP seems to be using Prores so the color space should be selected manually as from appropriate camera for the input. At least this has been my understanding. (I've only been using Braw so my memory as to the other stuff might be fuzzy.) And in any event best to use DaVinci Wide Gamut for the intermediary (timeline.)

No, even with ProRes and other YCbCr/RGB codecs Resolve does attempt to guess at what the input color space should be if there’s metadata in the file that Resolve can read.
It doesn’t mean Resolve always gets it right though, so you always have to check and change it if needed.
For raw files in RCM and ACES, there is no input color space option to select since Resolve handles the decode for raw sources directly to the timeline color space, and you’ll see the gamma and gamut options in the raw controls greyed out in ACES and RCM for that reason.

That’s likely a bug. RCM should ID those G2 clips as 4.6K Film Gen 4. You can simply set it manually to that by right clicking on them in the media pool or in the color page thumbnail timeline.

Well, here's the rub on that... Resolve 17 only lists "Blackmagic Design 4.6k Film Gen 1" and "Blackmagic Design 4.6k Film Gen 3". There is no option in the input color space for a Gen 4 version to choose. And from screenshots provided in the Resolve forum, it seems that v18 also lacks a 4.6k Gen 4 option.

Ah yes, I shouldn't have mentioned ACES, as I forgot that BMD still hasn't updated the IDTs there.

In Davinci YRGB Color Managed mode, there is an input color space setting for the 4.6K G2, but you've got to choose the custom option and check the "use separate color space and gamma" box which will allow you to set the gamma to Blackmagic Design 4.6K Film and the color space to Blackmagic Design Wide Gamut Gen4/Gen5.

BMD hasn't made it so easy to understand what's going on by defaulting to combing the gamma (which is more technically a transfer function) and the gamut into one "color space" setting. They are much more flexible to manage and easier to understand when approached as separate attributes. The 4.6K sensor has gone through three iterations of "color science" (just think "color processing" and the term makes more sense), all using the "4.6K Film" gamma, but changing the processing of the color gamut from Gen 1 to Gen 3 to finally in Gen 4 to "BMD Wide Gamut", which BMD has settled on as the gamut used on all the cameras with Gen 4 and Gen 5 "color science". If you shoot BRAW on the G2 rather than ProRes, you get the option to decode it in the raw tab to Gen 5 which has the same BMD Wide Gamut as Gen 4, but with a new flatter Gen 5 gamma curve which, as with "BMD Wide Gamut" is what BMD has settled on as the single gamma curve across all their current model cameras, rather than what was done previously where each camera sensor used a separate gamma curve.

That's at least my understanding of it, and if I've got it wrong, I hope Hook will chime in here to correct it.

I wonder if there are any gamut differences between the Wide Gamut selection and a numbered Colour Science Gen, those differences may show up on the CIE colour display from the Colour Page with HDR appropriately activated in your Project Settings, such as selecting Dolby Vision. Sorry don’t have computer access to verify but Jamie knows this stuff better than I.

I wonder if there are any gamut differences between the Wide Gamut selection and a numbered Colour Science Gen,

BMD Wide Gamut is used for both Gen 4 and Gen 5, standardized across all the cameras supporting Gen 4 or Gen 5. Before that the Gen 3 and Gen 1 color gamuts were different per sensor model. Again, that is as far as I know. Hook would know whether that is correct.

I have struggled with this too on the UMP 4.6K G2 and really appreciate the information Jamie provided here... it eyes opening for me. I hope BMD will make it simple, just like they did for the Pockets.

It seems like the UMP 4.6K G2 has been left in the corner like a bastard child in their camera portfolio.

Yeah, I usually use RCM these days, and have also found that the UMPG2 has been a little more challenging when using ProRes. BRAW it's wonderful. I can even match to the Pockets instantly with BRAW and RCM. But I have also had difficulty matching the UMPG2 to the P6KPro when using ProRes with RCM.

Back in February I had good success matching the UMPG2 in ProRes and the P6KPro in ProRes using RCM. I'll have to go back to that project tomorrow to double check the settings I used. But I had another project in January I remember looking at the interviews and having nightmare time matching. Enough so that I've determined ProRes is the worst for using the two cameras together.

Will share some things tomorrow.

Thanks for the help. I understand color management for the most part and read the sections in the manual. I was mainly just curious why BM left the latest version of their own flagship 4.6k camera out of their own software and if there was some logic in why it was picking the BM Pocket.
I did some comparisons and even configured the custom color management settings as mentioned above.
I set the clip to various different BM input color spaces and then compared stills. BM choice of the Pocket 4 gen 4 actually looks the most like I remember the scene and the colors look the most natural out of the choice with maybe just a touch too much saturation. So the BM Pocket 4k gen 4 color space does seem to work and the results look good...but from a purely technical standpoint I don't know how accurate it is to the actual color science of the actual 4.6k G2 chip.

We do shoot braw on occasion and that does take a lot of this burden regarding color spaces off since Resolve does all that work for you internally.
However, a lot of the content we shoot is just for simple talking head streaming and often we aren't the ones doing post. The folks we hand the footage of to often may not have the time, equipment or skill to work with braw and do their own grades.
And often we have tight turnarounds and other constraints that make recording to a common format like Prores more useful and not a proprietary format that needs additional software to view. So hence the recording to prores.

I was mainly just curious why BM left the latest version of their own flagship 4.6k camera out of their own software and if there was some logic in why it was picking the BM Pocket.

Calling the UMP 4.6K G2 BM's flagship camera is unfortunate. BM doesn't think otherwise.

... and often we aren't the ones doing post. The folks we hand the footage of to often may not have the time, equipment or skill to work with braw and do their own grades.
And often we have tight turnarounds and other constraints that make recording to a common format like Prores more useful ... So hence the recording to prores.

I share the same sentiment, and not speaking for Tim, but I believe he too shares the same experience as above. I wished with have in-camera Gen 5 for Prores firmware upgrade for the UMP 4.6K. But that is wishful thinking these days.

Okay, so here's what the project shows:

RCM Settings:

RCM Beach Boss Settings

Screen Shot 2022-04-26 at 11.17.37 AM.png (368.91 KiB) Viewed 3332 times

Then in the A-CAM URSA Mini Pro G2 Bin I selected all the footage and chose these settings:
Input Color Space - Blackmagic Design Wide Gamut Gen 4/5
Input Gamma - Blackmagic Design 4.6K Film

That gave RCM the info it needed.

B-CAM P6K Pro settings:
Input Color Space - Blackmagic Design Wide Gamut Gen 4/5
Input Gamma - Blackmagic Design Film Gen 5

That got the footage closer to matching together. I then did some finessing to match better. Obviously lenses and the different sensors do result in slight color variations even with RCM. But RCM got me to a really good point to begin from.

Another option that I think would work is to use "DaVinci YRGB" then use the Color Space Transform tool to convert:

Input Color:Blackmagic Design Wide Gamut Gen 4/5
Input Gamma: Blackmagic 4.6K Film
Output Color Space: Blackmagic Design Wide Gamut Gen 4/5
Output:Blackmagic Design Film Gen 5

You can also select the other Blackmagic Output Gamma options that are part of the BRAW decode like "Blackmagic Extended Video Gen 5" and "Blackmagic Video Gen 5."

I've experimented with some of my URSA 4.6K with GEN 3 to GEN 5 with the CST tool and it seems to work really well.

I've experimented with some of my URSA 4.6K with GEN 3 to GEN 5 with the CST tool and it seems to work really well.

There are caveats with both Gen 1 and Gen 3 that the transforms are color temp dependent. The CST transform does the math as if the source clip is 6500 Kelvin. If the clip is tungsten (or closer to tungsten) the color, especially skin tones, can end up looking a bit wonky. Hook has posted about this (can’t find the link at the moment, but when I do, I’ll add it). This color temp dependency is not the case in Gen 4 and Gen 5. And it is not an issue for cDNG raw source clips from Gen1 and Gen 3 cameras since Resolve uses the raw data to do the math right. It's only an issue with ProRes files from Gen 1 and Gen 3 cameras in RCM or using CST (so to be clear it does not apply to ProRes files from the Ursa Mini Pro G2 since those are Gen 4).

EDIT: I found the link. Here's what Hook wrote about that:
"The difficulty with that is that Gen 3 is white balance dependant (essentially two different primaries or gamuts interpolated for the given CCT - one at 6500K and the other 2856K and this approach goes back to the original colour science in BMCC etc), so it requires a different transform for every WB setting. Being practical, making a 'daylight' and 'tungsten' version could be enough to be useful in practice or even just a compromise of the two in a pinch, or if you happen to know you'll always be working with daylight-ish etc."

All Hook's post in that thread are worth reading: https://forum.blackmagicdesign.com/viewtopic.php?uid=16&f=2&t=130645&start=50#p731649

Jamie, thanks for posting that explanation, Hook seemed to be responding to someone creating a lut vs the CST tool. But that does seem to speak to the CST tool too.

I tested a few clips recorded with both tungsten and daylight recorded in ProResHQ with 4.6K gen 3 and it does bring the clip’s colors closer to that of the 12K without seeming to ‘break’ it. That’s in a clip without skin tones though where I have a shot of both the 12k and the 4.6K pointed at the same objects.

It’s not that anything breaks per se, instead what you’ll see is a hue and saturation skews from where it would be otherwise, in some colors more than others. Shoot a chart for a reference comparison between source color temps and you’ll see it clearly.

It’s not that anything breaks per se, instead what you’ll see is a hue and saturation skews from where it would be otherwise, in some colors more than others. Shoot a chart for a reference comparison between source color temps and you’ll see it clearly.

Yeah, I will take a look. I think, from what you are saying, I’ll shoot both DNG and ProRes from the URSA 4.6K since the DNG can transform correctly.

from what you are saying, I’ll shoot both DNG and ProRes from the URSA 4.6K since the DNG can transform correctly.

Yes, that's right. If you intend to work in ACES or RCM from the original 4.6K UMP, then DNG is the better way to record. And on the G2 that's BRAW, by far. I mean seriously, there really is zero reason to ever shoot ProRes. Educate your clients, and if they simply refuse to listen to reason, shoot BRAW anyway and just pull the BRAW into Resolve and export ProRes **AFTER** you've tweaked the white balance and ISO if adjustment is needed.