Film Restoration

Discussions - Using ACES
1

I’ve been testing ACES in DaVinci Resolve 12.5 on classic motion pictures that my team and I color restore. Source material is 35mm scanned to 2K DPX uncompressed.

My conclusions thus far, is that ACES is a superb workflow for digital material shot on today’s cameras. My team and I enjoy using ACES for productions shot in the 21st century. Color restoration of classic motion pictures is something else entirely.

For today’s post-production eco-system the IDT list in Resolve is already quite extensive, with a wide selection of cameras the colorist can choose from as we move the captured digital image into the ACES color space and encoding specifications. So far, so good.

I believe this scenario poses great challenges to the colorist who uses ACES to color restore classic motion pictures. While it is true the movie to be color restored was shot on film, it now resides in digital form as an intermediate. IDTs do not seem to address this ‘middle ground.’ However, the testing by my team and me indicates that Resolve’s IDTs do produce some interesting results. For instance, one IDT will give us a LOG look, while another will introduce saturation and overly vibrant reds. These and other IDT results we have tested are based off digital cameras that were not used in the production of the film. For this reason I am suspicious about the accuracy of any transformed image produced.

I would be interested in hearing from ACES representatives or other restoration colorists about their experiences with ACES, and what the future may hold for ACES vis á vis color restoration of classic motion pictures.

Kindest regards,

Jim Wicks
Senior Colorist, Manager of Film Restoration
Olympusat, Inc.

2

Jim,

I would suggest trying the ADX10 to ACES transforms if you have material that was previously scanned from film negatives. Your DI files are probably encoded in a manner consistent with the defacto standard known as “Cineon". Cineon, which is a printing density metric, was never standardized so it’s hard to validate whether your material does or does not conform that defacto standard. ADX (See SMPTE 2065-2 and SMPTE 2065-3) is also a printing density specification. Strictly speaking the ADX to ACES transform is intended to convert images from ADX to ACES but given ADX is an encoding of printing density it’s probably the best candidate for what you’re trying to achieve.

Alex

3

I’ve tried using the ADX to ACES transforms as part of ingesting some of our old effects assets, it worked pretty well given the issues Alex raises (though I know some of the scans really were some form of CPD as I was involved with calibrating the scanner). I’m not sure how well they hold up from a preservation perspective though as for that you might need something more specific.

Kevin

4

Hi Alex,

Thank you for the response.
Apologies for the late reply, but it has been busy - which is a good thing.

Yes, am aware of ADX. It’s a bit buggy in Resolve 12.5.
Testing so far has provided mixed results. It really depends on the condition of the film, type of film, scan settings, etc. If I were to be critical at all it would be that there is only ADX-10 as a solution, would like to see a number of options to chose from.

Kindest regards, Jim