BMD's "lurid" color

I just read Alan Robert's BBC test report of the original UMP in 2016, and his description of the camera's color as "lurid", and with a color chart to prove it.

The same is certainly true of the BMPCC 4K when using either a normalizing LUT, RCM or curves to normalize. A large saturation reduction is always one of the first steps.

I wonder what the BMD thinking behind this design is, apparently across all camera models?

The standard transforms are all just decisions about what might be a good default look. I haven't encountered anything inherent in the signal for the 4.6K that requires it to be more saturated when transformed for display. There's no reason that you have to use those standard transforms. If you don't like the standard looks, just make your own.

The 4K/6K Pocket has less dynamic range than the 4.6K, so saturated colored light is more apt to clip at a given exposure, but as far as I know that would be true for any camera with less dynamic range.

EVERY single time a bring in BMD footage i Color Conform to Arri LogC.
It gives me consistency across the board.

I just read Alan Robert's BBC test report of the original UMP in 2016, and his description of the camera's color as "lurid", and with a color chart to prove it.

The same is certainly true of the BMPCC 4K when using either a normalizing LUT, RCM or curves to normalize. A large saturation reduction is always one of the first steps.

I wonder what the BMD thinking behind this design is, apparently across all camera models?

This is what happens when users think they can just slap on a LUT without understanding the following:

• Color Space
• Gamma
• Tone Mapping

There is a much better approach, but I won't rob you the joy of discovery (and true learning) by simply telling you the answer.

There is a much better approach, but I won't rob you the joy of discovery (and true learning) by simply telling you the answer.

If you're confident you can instruct color graders the world over, who routinely use normalizing LUTs and the likes of RCM, by all means, enlighten the ignorant with your "true learning".

And examples of your grading, please, from which we all can learn!

The issue is what the camera produces (you know, like, camera tests?), not what occurs in post-production or post-production workflows.

There is a much better approach, but I won't rob you the joy of discovery (and true learning) by simply telling you the answer.

If you're confident you can instruct color graders the world over, who routinely use normalizing LUTs and the likes of RCM, by all means, enlighten the ignorant.

The issue is what the camera produces, not what occurs in post-production or post-production workflows.

There are plenty of professional colorists on YouTube that use Resolve. I'm sure you have some free time, go take a look. I gave you the clues....

Personally, I prefer BM colors to Red/Sony/Panasonic, etc... The only color science that I like better is from Arri.

You don't understand the initial post. The issue isn't how to color correct BMD footage.

You still don't understand the initial post.

How is that? You mention some guy who doesn't like BM colors, then talk about using a LUT, Resolve Color Management or curves to normalize...

Are you shooting in BMD Film? It seems that you think some standard that worked for other cameras can be applied the same way to a BMD camera. Otherwise, you'd just adjust the "normalizing" LUT to fit BMD color science or adjust the curve to appease your eye. Maybe I don't understand your post, because I've never had this issue and I've been shooting with BMD since 2012.

I've seen what you talk about, but only when slapping on a LUT.

Is this person shooting BMD Film or some other format?

Explain it to Alan Roberts. I'm sure he'd appreciate your help:

https://tech.ebu.ch/camera_reports_tech3335

Explain it to Alan Roberts. I'm sure he'd appreciate your help:

https://tech.ebu.ch/camera_reports_tech3335

I don't give a you know what about these camera tests... who cares...

"The colour performance is rather lurid, with a distinctly pink look to many colours, and the camera responds to infra-red..."

WE ALL KNOW THAT!!

There are very few cinema cameras that you can shoot with and would use the image straight out of the camera. In practice, I'm sure the number is 0.

...and the thinking behind BMD's design... I don't think there is any at this point. Why change it? Although I'm sure there will be a Gen 5, Gen 6, etc of the color science. We're buying it, so it's good enough. The deeper they go, the closer they get to Arri territory as far as color science. What would that cost? It's good enough already... As a user, sure, make it as good as you can... But will you be getting a 4K cinema camera for $1300 at that point? Probably not.

In addition to different color science versions, BMD RAW source can be processed in two different ways with slightly different results:

If you use YRGB Color managed project setting, your input is defined only by simple matrix based profile (same as CST node) and produce slightly less saturated, less corrected "brownish" colors. This look doesn't match ProRes colors from camera.

If you use YRGB non color managed project setting, your input is defined by internal RAW input profile. It produce more saturated, more balanced and more "correct" colors because it based on some more complicated LUT-based input profile. This look 100% match ProRes colors from camera.

"lurid" BMD color also usually produced due IR pollution.

If you use IR filter and Color Checker you can get 100% color perfect image from any BMD camera without any additional manual color correction.
With additional 80A or 80B filter you you can get 100% color perfect image even under Incandescent light bulb:

[CaptainHook]

Explain it to Alan Roberts. I'm sure he'd appreciate your help

Alan made some assumptions in his analysis, just a couple of them :

• "It is not clear from the camera specification whether the recorded signals are coded to ITU Rec.601/709 formula, or sRGB. On test, on the initial assumption that the video signals would be 601/709"

Our "Video" option is not intended to be Rec.709 and is named as such deliberately because it is not Rec.709. If you are expecting "traditional broadcast video camera" colours then yes, it will not meet expectation.

• "Therefore, I have assumed that all signals are sRGB coded in Film and RAW modes, but ITU 601/709 in Video mode"

Again, not the case. It would be more appropriate to select Rec.709 from the RAW tab for "traditional broadcast video camera" colour.

There's a couple of other incorrect assumptions like this one:

• "The single sensor is specified as super35mm format with at least 4608x2592 photo-sites, and Bayer pattern filters. Super35mm in video cameras is usually 23.5x13.2mm, making the individual photo-site dimensions about 5.1μm square."

The sensor size is described on the product page on our website (25.34mm x 14.25mm) , to which you can calculate the correct 5.5μm.

Explain it to Alan Roberts. I'm sure he'd appreciate your help:

https://tech.ebu.ch/camera_reports_tech3335

Alan not shoot a color chart neutrally, but in his relation wrote that it light with tungsten light and correct wb in camera. From what i know bmd sensor are calibrated to neutral day light, when you change that and later correct color it's obviously that color is different and could be shifted...

Strange thing is that this annotation is only on two or three of cameras tested, most of other cameras not declare how light (color) and correct wb... (yes i downloaded all pdf and read it, just for curiosity).

i think he is more skilled on broadcast video cameras and not on cinema cameras.
Anyway. i learned from many years, from filmstrip to digital sensor that i not shoot with a color light and correct later, i add filter before hit filmstrip or sensor, or i shoot colored light be cuase i want that color.

Most of cameras if shoot with color corrected camera in wb could shift some color, and if you do with strong shifting like tungsten i can tell you that many cameras cannot correct sb only with wb balance, or better, from my experience : gh3,gh5, ump, pocket4k/6, c100mkI, c300, no other i did that kind of test.

If i shoot with tungsten light, and correct wb, i create a weakness on blu channel, by cause i amplify data to balance picture, often is a simply more noise on blue, but sometimes when you have strong warm light you can see a shifhting on pink, red, and blu, like he observe.

Dmitry, I have been grading lately using YRGB Colour Managed. Sounds like you would recommend non-coloured managed. I though Colour Managed was good as Resolve makes some choices for you. Simpler and more standard?

I’m mixing URSA Mini 4.6K ProRes 444 UHD and BMPCC4K BRAW Q0 HD and h.264 UHD from the Nikon Z6.

Dmitry, I have been grading lately using YRGB Colour Managed. Sounds like you would recommend non-coloured managed.

The manual is pretty clear on this one: the most accurate math is with RCM, if Resolve's normalizing LUTs are the alternative.

Thanks, John. I like the results I’m getting. But now I have the Apple Pro XDR Display, I’m starting to regrade the narrative film.

You might have a valued opinion on my current approach which may seem odd at least at first glance.

I’m setting my Timeline and hence output to the DCI P3 gamut and DCI gamma in preparation for a theatrical release that may never happen, but recognizing it will be a blessing just to have it okay on HDTV, I’ve set the Pro XDR Display reference mode to Rec.709 as that’s a standard for HDTV.

So I’m in the colour space I like but grading that for playing back on HD television. Is that a disaster or makes sense? Would you recommend I do that or go with Rec.709 gamut and gamma which I find harsh?

Low budget filmmakers routinely grade in rec. 709 and leave the color space conversion to the post house making the DCP.

I'm not entirely clear on how/whether you've mixed display modes/color spaces here, but in your position, I'd stick religiously with rec. 709, because you know the chain is consistent and predictable, assuming you have the monitor accurately calibrated. Otherwise, there could be very unpleasant surprises in the course of later conversions. And meanwhile, you'll have something that will play correctly on any reasonably well calibrated rec. 709 monitor, professional or otherwise.

Okay, I’ll give it a go like that. Rec.709 all the way at this point. And if I need to do a DCP later, then I’ll be DCI P3 consistently.

Would be good to grade both ways now.

You might want to raise this question in the Resolve section, where the professional graders live.

I don't think you can do an accurate DCP grade at home, because you really need a projector. The monitor may be of very limited value here. Rec. 709, by contrast, is within the means of your setup. At least, it will be accurate within its limits.

Ah, yes, definitely need to test in a theatre. All theatres here are closed indefinitely due to COVID-19.

So might be easy to have someone approve a test as they may be dying of boredom. Just must practice ‘social distancing’ and that should be good with the health authorities. I’ll leave that to the producer to arrange. I’m just her worker bee.

I think you have to actually grade with projector output, hour after hour, if you want predictable results. Whereas it's always safe to grade rec. 709, and leave the DCP conversion to post-houses experienced at doing it.

Anyway, do inquire further before proceeding!

Dmitry, I have been grading lately using YRGB Colour Managed. Sounds like you would recommend non-coloured managed. I though Colour Managed was good as Resolve makes some choices for you. Simpler and more standard?

I’m mixing URSA Mini 4.6K ProRes 444 UHD and BMPCC4K BRAW Q0 HD and h.264 UHD from the Nikon Z6.

I like non color managed workflow because it is way more flexible and allow more control. Just render same frame with same settings from color managed and non color managed project and compare. "Non color managed" RAW exact match to ProRes colors from camera. "Color managed" RAW produce different look.
Here are my examples:

Interesting, Dmitry. You would think they’d be the same but there are differences. I’ll try that too.

I like non color managed workflow because it is way more flexible and allow more control. Just render same frame with same settings from color managed and non color managed project and compare. "Non color managed" RAW exact match to ProRes colors from camera. "Color managed" RAW produce different look.
Here are my examples:

Dmitry, I notice in your examples that you are setting the input color space to BMD Gen 1.
I wonder if that might be the reason for the difference.
Have you tested with Gen 4 on a newer BMD camera that processes the internal ProRes in Gen 4?

If I get a chance today, I'll test on UMP G2 to see if there's any difference.

[CaptainHook]

You won't get an "accurate" result with BMD Gen 1 (or 3) colour space and the CST plugin unless your white balance is set to 6000K (this is the colour temperature the Resolve team selected for the CST plugin).

Gen 4 is fine with the CST plugin no matter the white balance setting.

Regarding Alan Roberts's article: he tested colors with Tungsten lighting. Sorry, but there is a reason we are shifting more and more to artificial lights in daylight color temperature. The spectrum of Tungsten doesn't fit electronic sensors really well. Our Reds also suck under Tungsten.

You won't get an "accurate" result with BMD Gen 1 (or 3) colour space and the CST plugin unless your white balance is set to 6000K (this is the colour temperature the Resolve team selected for the CST plugin).

Gen 4 is fine with the CST plugin no matter the white balance setting.

Gen 4 produce same difference between color managed and non color managed. Sure Gen 4 looks more vivid and more balanced and accurate than Gen 1. But due smaller color space Gen 4 may produce ugly red channel clipping in some situations. Red channel clipping problem may be fixed by upscaling project to wider color space like "REDWideGamutRGB", this was described in separate forum thread.

P.S. If you use additional ColorChecker correction tool in Resolve - it adjusts color managed or non color managed, gen 1 or gen4 to exact same well balanced and well saturated look.

[CaptainHook]

Your example shot is close enough to 6000K that I wouldn’t expect much difference in the transform between Gen 1 or Gen 4 using the CST plugin. At 3200k (for example) I would however.