Gen 4 color science

With resolves new beta, Gen 4 color science has been released. However, I've been reading that the best way to apply it is not to select Generation 4 from the RAW tab, but use the Color Transform tool.


I've been going back to a few clips and simple selecting color 4 science from the raw tab and I am losing a lot of color in the greens and Blues.

[Peter Chamberlain]

Do you have clips recorded with the generation 4 color science?

Do you have clips recorded with the generation 4 color science?

I recorded RAW using the latest firmware for the UM4.6k, am I supposed to have flip a setting on in camera?

sorry, gen 4 of what?

Gen 4 of color science. Shows up in recent betas of DR 15, but is only in the broadcast camera until now. Supposed to come to the UM46P with a new firmware. John Brawley is already showing some very impressive examples here: http://www.bmcuser.com/showthread.php?2 ... discussion

[CaptainHook]

I've been reading that the best way to apply it is not to select Generation 4 from the RAW tab, but use the Color Transform tool.

The CST plugin assumes the camera was set to 6000K so transforms from BMDFilm Gen 1/3 to Gen 4 at temperatures around tungsten will especially not be as accurate. I would stick to the RAW tab.

I've been going back to a few clips and simple selecting color 4 science from the raw tab and I am losing a lot of color in the greens and Blues.

Overall there will be less saturation in Gen 4, so you would need to increase it to compare to say Gen 3. Plus the gamut is just different. Do you have some images you can share?

I'll copy my guesses here, so maybe developers can make things more clear to us.

I read Resolve 14 Manual starting at page 135. There is a lot of info about Color Space Transform but seems no clear info about Color Science versions and their relation to Timeline Color Space. Also in manual developers name "Color Science" everything related to color transforms, not only Color Science setting in RAW tab. So probably we need some deeper lesson from Resolve developers to explain what going on under the hood when we disable Resolve Color Manage input and use "Color Science" options in RAW tab.

I usually use RCM + Color Checker correction, because when try to use Color Science it feels like always adds contrast and slightly clips extreme colors, so it slightly less flexible for future grading. But if you don't have Color Checker, using Color Science may help a lot to fix known dimmed saturation and brownish greens colors from BM cameras.

From my personal unconfirmed guess Color Science (v1, v3, v4) acts like input LUT with build-in color correction. Designed for specific camera and specific timeline color space. It is not able to transform colors to any desired color space same as RCM aka Color Space Transform do. If it produce ok result with some different timeline color spaces it only means that it just lucky similar to color space and camera that Color Science was designed for.
If shoot to ProRes, Color Science correction is baked in footage so you can not disable it or switch different version. That's why ProRes footage always looks more vivid compare to RAW.
Is this all correct?

I done a lot of tests this weekend and compare different combinations of Camera + Color Science + Timeline Color Space. The results are totally different and also very with different cameras.

Same tests with BMMCC footage. I only add here contrast 1.1 node and Color Space Transform node (Timeline to Rec709 With Luma/Saturation mapping ON) in the end.

BMDfilm (no Color Science) Colors converted from input to timeline with build-in Color Spase Transform. Uncorrected colors, brownish greens, less saturated look, 100% no any clipping, perfect starting point for ColorMatch tool correction.




Color Science v1, Timeline set to BMDfilm Nice looking image vivid corrected colors, seems no any clipping. Color Science perfect matches to camera and timeline.




Color Science v4, Timeline set to Pocket4K. More contrast and more vivid look, may start people to believe that it is "better" Clips extreme red colors with Pocket/Micro footage.




Color Science v4, Timeline set to BMD Broadcast. Seems no clipping but slightly washed blacks.




Color Science v4, but Timeline set to BMDfilm. Oversaturated Reds, something wrong in this setup.

I've been going back to a few clips and simple selecting color 4 science from the raw tab and I am losing a lot of color in the greens and Blues.

CaptainHook,

So am I to understand that Generation 4 color science isn't part of the UM 4.6k and only a part of the UMPro?

I thought that if you recorded CDNG RAW on the UM4.6k and applied the Gen 4 color science in Resolve you'd have Gen 4 color science.