Weird colour issue on BMPCC

Hi guys, wondering if you can help me. I'm really new to video production but due to getting some funding for a project I bought a BMPCC which I'll be doing a load of filming on - interviews and such mainly. I have probably bitten off more than I can chew, but what the hell, right?

I did my first 2 camera shoot yesterday, and I noticed a weird thing with the BMPCC footage I can't seem to fix with my limited knowledge of Resolve. Below is a video comparing shots

- first is a Lumix G6/Lumix PZ 14-42 on full auto
-no processing has been applied
- second is the BMPCC/speedbooster/canon EF-S 17-55 plus a Tiffen T1 IR Filter, ISO 800 f/1.7
-turned up the saturation by 150 just so you can see the colours



For some reason I'm getting a very dark result when uploading to vimeo (fine on youtube) - that's not the issue at hand though, just mentioning it so you know I'm aware - the oriiginal footage exposure in fine in FCPX.

As you can see the G6 footage shows a bright gold buddha - the look is a bit over the top but the colours are broadly accurate. The BMPCC footage shows the buddha as a much more bronze colour. I can't correct it in resolve, though I'm sure it's possible for a pro, but my question is what is happening here? Am I missing something? I mean the colours should roughly match even if a bit of grading is required (I'm aware there is a bit of a green cast from the filter, I can correct for this fine - just thought I'd leave the footage clean for the purposes of seeking advice).

Sorry for asking what is probably a stupid question, but I really need to know if I'm doing something that will spoil all my footage before I embark on tons more shooting over the next couple of months and I figure you guys might be able to help.

All the best

Well, welcome to the world of Blackmagic Color Science!

BMD seem to have their own opinion about how red and orange should _really_ look like. The bronze/golden buddha is the prominent feature in your clip, but you may also notice some differences in the reds. Part of it may be IR-pollution (the bmpcc has a very inefficient IR filter), part of it might also be BMDs color science, but in any case: that's known behavior, and deemed not a fault but a trait of design.

It is possible to counteract by shifting the hue a few degrees, but you need to filter and apply the shift only to the brown/red/orange range.

Mac, you're right, but note that David says that he is already using an IR filter. The result he is getting is exactly what i get from my bmcc and pocket. I really hope that they will change their color science, because I prefer to start grading from a neutral log rather than a brown-ish one.

If anyone has a lut to correct this I would be really thankful because I'm having hard times to grade in Da Vinci to achieve what my clients want

Mac, you're right, but note that David says that he is already using an IR filter. The result he is getting is exactly what i get from my bmcc and pocket. I really hope that they will change their color science, because I prefer to start grading from a neutral log rather than a brown-ish one.

If anyone has a lut to correct this I would be really thankful because I'm having hard times to grade in Da Vinci to achieve what my clients want

The odd colours only seem to happen for me when using the "video" profile, the "film" profile for me does yield a neutral log image.

As long as you have fully white balanced the shot, the colours seem quite accurate to me after just applying some contrast and saturation.

Failing that - the Rec709 LUT in resolve with the v2 suffix is very very close to perfect - when compared with a calibrated profile I made myself.

Example:

How the image would look with the old Rec709 LUT in resolve



How the image should look:



How the image looks now with the new Rec709 v2 LUT in resolve


Its a bit more saturated but hue wise its very very close to perfect.



For best results - I would always avoid using the "video" lut in the camera, if you want a quick grade, apply the Rec709v2 LUT in resolve after perfecting the white balance and you will get accurate results.

I always shoot raw so no, I can't use the "video" mode. I import the raw in Resolve, set "BMD Film" in settings and start to grade from the log image. I come from r3d workflow and it was a bit easier to get the colors I had in mind, maybe it's just because I'm not used to BM log yet. Tom I will try the new rec709 Lut and let you know, thank you

I always shoot raw so no, I can't use the "video" mode. I import the raw in Resolve, set "BMD Film" in settings and start to grade from the log image. I come from r3d workflow and it was a bit easier to get the colors I had in mind, maybe it's just because I'm not used to BM log yet. Tom I will try the new rec709 Lut and let you know, thank you

No problem,

so with Raw - set it to BMDfilm on the Raw tab and then add the new rec709 lut to a node

If you set the Raw tab to Rec709 (the default) then it will be using the older version.

Panasonic colors are not as good in my opinion. The reds tend to blue, and saturared reds, greens and blues tend to be oversaturated. I think it has to do with rendering of skin color of asian people.
A coca cola or santa clause red is to blue and to saturated. I always need to correct this in Photoshop.

Quite opposite the BMPCC. European skin color is perfect. Wheres as roofs and the coca cola red tends to be too orange.

Thats different ways to handle the same problem:
I you look at some specs of sensor spectral sensitivity you can guess the problem.
Any 3 color colorspace device that tries to mimic the human eye or give an illusion of colors by "fake stimulating" with metamer colors has to make sure that the spectral sensitivity of the three input sensors for red, green and blue channel are exactly the same as those of our cones in the eye.
If not there are more or less problems. Since you cannot correct the errors by simple matrix calculation. Matrix colorspace transformation are valid inside a 3-channel representation of spectral distribution of light. But the problems appear during this lossy transformtion of a continous spectrum of light into a three value representation.

Film has some problems with violett. Digital seems to have some problems with reds. Digital behaves a little bit like old unmask negative film.