Film Scans setup with ACES workflow


(Stevebearman) #1

Hi
I work in film restoration and am looking for some advice in setting up the project. We have 16bit Log DPX files coming in and we are working in ACES. The only options in resolve in the IDT for film seem to be ADX 16 which skews the image forcing a tremendous amount of light into the picture.

Can I work with no IDT as I can get much better results without an IDT or is it essential.
Many thanks
Steve


(Alex Forsythe) #2

Steve

ADX16 encodes additional range and precision as compared to ADX10. The density range encoded by ADX16 is about 8 vs about 2.046 for ADX10. Since no film will have a density of 8, this means scans in ADX16 will appear very dark if you just look at the DPX files directly. The ADX16 IDT takes the range into account when converting ADX16 images to ACES.

It sounds like your 16-bit DPX files are encoding less range than is specified by ADX16. This is likely why you’re getting very bright ACES images out of the IDT.

The IDT is needed. If your scans encode the same range as ADX10 but use all of the additional 6 bits for precision, try using the ADX10 IDT.


(Jim Houston) #3

The fix for it is (if set to ACEScc color science with an ADX16 IDT and a Rec709 ODT )

apply a gain of about 0.25 in the primary wheels. This will get you in the ballpark and
you can time from there. This will get rid of the overbrights.

It looks like your 16-bit scans came from the scanner as ADX10 but in 16-bits. There is a 4X boost from ADX16 that is normally needed which Alex talks about.


(Stevebearman) #4

Hi Alex

Thanks for getting back to me. That makes sense to what I’m seeing. What would be great to confirm is whether the ADX 10 is just a matrix transform and won’t affect the 16bit workflow? Is this correct.

Many thanks
Steve.


(Alex Forsythe) #5

What would be great to confirm is whether the ADX 10 is just a matrix transform and won’t affect the 16bit workflow?

Steve, not sure I’m following the question … the both the ADX10 to ACES and ADX16 to ACES transforms include a couple of matrices and a “Log to Lin” conversion.


(Stevebearman) #6

Sorry I’m not explaining it very well, I’m trying to find out that if by using the ADX10 it would somehow constrict our 16bit DPX files to 10bit. Our pipeline has to be 16bit from scanning up until our deliverables.


(Alex Forsythe) #7

ah … I’m not sure off the top of my head but I don’t think it would. What’s really constricting the range is dedicating the entire 16-bit code space to the same density range as the 10-bit code space. Practically speaking, however, there aren’t a ton of negative stocks that have a density range that exceeds the range encoded in ADX10. Even if you are using one that does, the scene itself needs to have take advantage of that additional stock density potential.