Matching BMPCC 6K and Lumix G7 B-cam

If you had to do your best to color match multicam footage from a BMPCC 6K shot in BRAW and from a Lumix G7 (good looking footage, but with much higher contrast and darker, and H.264 at like 100 Mbps), where would you start? Maybe convert the Lumix footage to log to have a more common base with the BMPCC 6K footage, even if it's not the same?

CST both of them to DaVinci Wide Gamut.

Or use something like Cinematch.

where would you start?

Here...

Color management will not solve it, and neither will a Wide Gamut work space. No idea how well Cinematch would work in this case, but there's a free trial. Filmlook and similar film emulation packages with both BMPCC 6K and G7 profiles might accomplish something similar, in a roundabout way.

And of course you could try to do it by hand, but that's likely to be frustrating if you have no experience in the matter.

Color management will not solve it

Why not?

(Seriously. I don't know.)

Color management will not solve it

Why not?

(Seriously. I don't know.)

I guess he means that we're talking about a digital cinema professional camera that shoots in RAW and is far more sophisticated than the Lumix G7, which is a plain consumer camera shooting in h.264. I don't have any hopes that I will match them perfectly, but I would like to get as close as possible. It's like in movies and TV series when they have something that obviously they shot on iPhone or a drone that is not the latest and greatest, and you can tell there's a difference, but they still did a pretty decent job at matching them.

But color depth is not the only thing, you have those cameras that artificially sharpen the picture, and if that's too drastic then you can try to minimize it, but you won't get it perfectly identical to a camera that doesn't do that.

Thanks Jim for the video, I've been watching it since last night and it helps to have the perspective from someone at BMD, because I watched tons of tutorials and it can be pretty confusing when you're new to the world of color grading, because in the other NLEs you have one choice for Rec.709, in Resolve you have like 4 or 10,000.

Then you watch a tutorial on achieving a cinematic look, and they all do the convert to Cineon because the Film Look LUTs expect a Cineon Log profile, but then other tutorials mention setting the color science to DaVinci color managed, so it's rather confusing even for someone like me who's been tinkering with video settings for 20+ years.

Another question related to this... I know that to do proper color grading you need professional and very specific monitors, obviously a pro monitoring card as opposed to your graphics card or Mac's HDMI port, etc, etc.

But say you don't have access or a budget for any of those things. You do have a Mac Studio with a perfectly functional HDMI output connected to an LG OLED TV set, and two thunderbolt monitors that are not professional but are a couple of steps above basic consumer level, and by this I mean two Samsungs LF32TU87, which I hate for another reason but they have great PQ.

In the macOS settings app, in displays, you have the choice of setting the color profile to a number of things, the default is always the color profile that came with that monitor's driver, but you also have several others, like sRGB and Rec.709, just one profile called "Rec. ITU-R BT.709-5".

Now, if I leave it at the default monitor profile, it looks fine. That's how I always had it. If I set it to sRBG (which as I read is about the same as Rec.709 with a minimal difference), the contrast is higher and the colors are more saturated.

And if I set it to that Rec.709 profile, then the contrast and saturation are even higher. To me it looks great, but what's the most accurate as far as Rec.709 is concerned? I mean, can I even trust that as being in the ballpark of Rec.709, or is it just some arbitrary gamma curve?

Darren also has a video on using the Film Look LUT's. I asked him about combining those with Color Management, he basically said it can't be done.

If you want to use those Film Look LUT's you'll need to go the CST route.

Darren also has a video on using the Film Look LUT's. I asked him about combining those with Color Management, he basically said it can't be done.

If you want to use those Film Look LUT's you'll need to go the CST route.

So if I understand correctly, you either do the "DaVinci Resolve Color Managed" setting, or you do the standard one and use the CST filter to convert to Cineon, then another node to apply the film look, correct?

You're both on the wrong track. The purpose of color management isn't to make footage from all cameras look the same. Color managed Arri footage will still look like Arri footage, as will Canon, BMD, Sony, etc. The differences aren't eliminated on a color managed timeline. Every camera is normalized to a new color space according to its own particular math. The characteristics of each are preserved.

Similarly, a CST routine from camera log to cineon to rec. 709, allowing for the use of Resolve's film look LUTs, is unlikely to accomplish shot matching between cameras, even if both sort of look like a certain film print stock in their own way.

What you *could* try is a film look emulation package which has profiles for the two cameras. In theory, that should produce a result which meets a single "film look" standard. But no guarantees there either, I wouldn't have high hopes. The obvious choice -- at least something to try first -- is an application designed for this exact purpose. It's mentioned above.

If you had to do your best to color match multicam footage from a BMPCC 6K shot in BRAW and from a Lumix G7 (good looking footage, but with much higher contrast and darker, and H.264 at like 100 Mbps), where would you start?

The Panasonic Lumix G7 was a first-generation 8-bit 4K camera. Why was this used as a b-cam in the first place?

The best you can do in terms of matching is to leave the 8-bit camera footage from the G7 as is and try to make the BMPCC look like it as much as possible probably by boosting the contrast.

There is simply not a lot of room for an 8-bit first-generation 4K video color correcting-wise, it will break down with anything but the smallest adjustments.

Second that, apart from the fact that they probably don't want to re-shoot with a better B-cam

Get the G7 footage as close to the intended look as you can, but only with subtle adjustments. To make sure, encode a small part with your adjustments to your final distribution format, to see if it doesn't fall apart from further compression. For a start, its settings under "Input Color Space" for RCM should be Rec.709.

Then adjust the BRAW footage to bring it as close as possible to the G7. There is no way to do it the other way, even if the BMPCC looks much better. Just yesterday I had to work with a very similar situation, one UMP 4.6K in BRAW and two amateur Sony cameras, one even set very wrong and underexposed. I first balanced the worst camera and was even able to denoise it with NeatVideo, then I went for the others. BTW, the new Color Slice in 19 was very helpful, and 19 was stable so far under Ventura, but I didn't do any fancy stuff.

Finally, the free Colorist Guide by BM is very helpful, even if you're familiar with other software.

The purpose of color management isn't to make footage from all cameras look the same.

Here's what I'm reading in the Reference Manual.


"DaVinci Resolve 12 introduced a color science option called “DaVinci YRGB Color Managed,” or
more simply “Resolve Color Management” (RCM). This introduced a so-called “Scene Referred” color
management scheme, in which you have the option of matching each type of media you’ve imported into your project with a color profile that informs DaVinci Resolve how to represent each specific color from each clip’s native color space within the common working color space of the timeline in which you’re editing, grading, and finishing."


Now to me, that means RCM was designed to get disparate camera color spaces into one common working space. In colloquial (if perhaps not technically correct) terms, to 'look the same'.

So it seems to me a good starting point for Sebastian in his stated goal. Darren, as well, makes this point in the video I linked. He shows how to use RCM to get media from different cameras into the needed Rec.709 space.

If you guys are making a different point, what am I missing here?

You are misreading the manual. "Matching" in the above example means matching each type of footage to its particular transform (its "color profile") for that color space. Not matching them to each other.

The fact that the timeline represents a "common working space" says nothing about how the footage from different cameras actually looks. Each has a separate transform and distinct characteristics. Why would they look the same at the end of very different transforms?

Sure, I don't read it as matching them to "each other".

I read it as matching them to a common Space. Which seems to be what Sebastian needs here.

RCM would put the P6K into Rec 709. The G7 is already there. That common Space would make the two cameras (again colloquially) "look the same".

Why would they look the same at the end of very different transforms?

Because those different transforms have the same end point. (Let's assume Rec.709 for simplicity.)

It's like two different roads that get you to the same city. Each road comes from a different starting point, but they both end up in the same place.

If I'm still misunderstanding, please enlighten. (Because I use my current understanding all the time with the expected results.)

To repeat, Jim: a common color space, which can be huge, is independent of the characteristics of one camera or another. They all happily co-exist inside it, with their variations.

The "same end point" (the color space) means they're operating in the same universe, not that they look alike.

If you had to do your best to color match multicam footage

I mean he clearly wants them to match, not just be in the same color space - that may make adjusting easier but won't automagically match them,

But they DO "look alike". They both look like Rec.709.

I don't say they look identical, that no further ]adjustments will be needed to get them closer. Just they when they're both in Rec.709, they are a much closer match, and that's a good place to start tweaking.

They both look like Rec.709.

How does rec709 look ?!
That's like someone asking for a description of the bank robber's car and you'd reply: it looked like a car

And if I set it to that Rec.709 profile, then the contrast and saturation are even higher. To me it looks great, but what's the most accurate as far as Rec.709 is concerned? I mean, can I even trust that as being in the ballpark of Rec.709, or is it just some arbitrary gamma curve?

That is a very deep rabbit hole - so many variables, norms and pitfalls. I'm afraid you'll need to dig deep to get answers.
Your best bet to use your Mac Display might be this workaround by Cullen Kelly:

This is NOT to solve common gamma shift issue after rendering, this is to get your Mac Display to something you can fairly trust if you don't have a clean feed to a calibrated reference monitor.

Rec. 709 simply means they both look like TV now. To stick to bank robbery: if all observers agree it was a car, at least we know it was not a truck, like Canon RAW, or a motorcycle, like BRAW
But every manufacturer has a different opinion on how a nice car, ahemm, video should look like. It’s one of the main jobs of a colorist to match different ones. And sometimes one of most difficult ones.