Color accuracy using SMPTE (ARIB) colors

[Cary_Knoop]

I created a video with SMPTE (ARIB) bars.

When I import it in Resolve (Version 14 beta 2 on Windows) and use ACES (input and output transform Rec.709) as the Color Science the Red bar will have 4% green.

Is this as designed?

Link to the test video:

[sdyer (Scott Dyer)]

Well it’s not “as designed” because it was not intended, but it is a known issue. There is an error in the RRT that makes it not fully invertible due to the “red modifier.” This is on the list of issues to fix in an update to the RRT.

Previously mentioned here:

ACES in commercials, white titles, supers and logos

I don’t remember seeing anything not invertible in the ODTs (the segmented splines should be as most of the transformations), however the RRT code definitely has some non invertible code, i.e. Red Modifier. Edit: Actually, there is a clamp in for example transforms/ctl/odt/rgbMonitor/ODT.Academy.RGBmonitor_100nits_dim.ctl: linearCV = clamp_f3( linearCV, 0., 1.);

and here:

Preserving the look of sRGB graphics

I will be pedantic but it will be unchanged in the limit of the inversion capabilities of the system, i.e. the red modifier is not fully invertible. Thomas

(you can click through for full details)

[Cary_Knoop]

Ok great, so there is a bug database where the error in the RRT is recorded as a bug?

[jim_houston]

There is an internal bug tracker, but not at the moment generally accessible.

[Zeb (Zeb Chadfield)]

I’m not able to view the previously attached video but I think this is related to something I’m running into right now in various applications with ACES. I’m curious about two things, can Colour Bars work with ACES? My research so far says no, grade in ACES then output to something else to deliver. The other thing is for restoration work with a REC709 copy of the original film, working in ACES on the rushes to match grade to the source. What is the accuracy like? If the bars from REC709 don’t display accurately (yes even with the correct IDT etc) then surely the reference file is also not accurate enough to grade against. Any thoughts would be much appreciated.

[nick (Nick Shaw)]

@Zeb

If the bars from REC709 don’t display accurately (yes even with the correct IDT etc) then surely the reference file is also not accurate enough to grade against.

Although that is strictly correct, in fact it is not normally relevant.

The reason the bars don’t pass through correctly is that the RRT is not 100% invertible. Therefore the combination of the inverse Rec.709 ODT, the inverse RRT, the RRT and the Rec.709 ODT clip a bit of the Rec.709 gamut. However the vast majority of colours occurring in normal images pass through intact. It’s just that certain saturated colours (particularly yellows) are outside the range that can be inverted.

[stevebearman]

I’m getting desat yellow and cyan but not on my reference so until Zeb noticed this issue I have been pushing yellow and cyan back in. This will result in overly saturated yellow and cyan on output no?