[Marc Wielage]

Read these:

“Grading for Mixed Delivery: Cinema, Home, and Every Screen in Between” by Cullen Kelly
https://blog.frame.io/2019/10/14/grading-mixed-delivery/

and
“How to Deal with Levels: Full vs. Video” by Dan Swierenga
https://www.thepostprocess.com/2019/09/24/how-to-deal-with-levels-full-vs-video

and I think both cover the issues and the solutions very well. Understanding color management is also helpful:

“Color Management for Video Editors”
https://jonnyelwyn.co.uk/film-and-video-editing/colour-management-for-video-editors/

[Marc Wielage]

[matt schween] “I’m now sending a .DRP over to our editor to do the final mix and output but he’s noticing that there’s a colour difference between my final grade and his timeline. When he brings my 422 ProRes full resolution export into his system there’s a dramatic difference in colour between his timeline and the 422 file on a timeline.”
Read these:

“Grading for Mixed Delivery: Cinema, Home, and Every Screen in Between” by Cullen Kelly
https://blog.frame.io/2019/10/14/grading-mixed-delivery/

and
“How to Deal with Levels: Full vs. Video” by Dan Swierenga
https://www.thepostprocess.com/2019/09/24/how-to-deal-with-levels-full-vs-video

and I think both cover the issues and the solutions very well. Understanding color management is also helpful:

“Color Management for Video Editors”
https://jonnyelwyn.co.uk/film-and-video-editing/colour-management-for-video-editors/

[Marc Wielage]

What kind of timecode is in the clip? What specific codec does the clip use?

Also, check the EDL by reading it in a text editor, and see what the timecode numbers are and clip names are. This may give you a clue on the malfunction.

[Marc Wielage]

Yes, I agree with Michael’s suggestions above.

Some general tips for Face Refinement:

1) as Michael points out, you often get better results if you use PTZR to zoom into the face and THEN run the F.R. analysis mode. Once it’s tracked, go back to normal framing and the mask should still fit.

2) if there’s conflicts in the background, consider using a temporary black mask in a previous node to mask out everything else except the face. This will give F.R.’s internal engine fewer distractions to work with. This is particularly useful if you have 2 or more people in the same shot.

3) note that it’s possible to use a mask on top of F.R. in case it’s catching any details behind the actor’s face. Of course, you can also manually adjust the size of the F.R. mask area itself.

4) I find that F.R. alone is not enough to benefit actors of a certain age. I find you also have to pay attention to their neck, visible chest, and hands, and you’ve got to blur or denoise or use Minus Midtone or even Glow.

5) and I also think you have to use manual methods to get rid of crow’s feet and jowls, which often have to be windowed and tracked. The OFX Beauty plug-in is a good way to use frequency-selection to only soften the skin details and leave everything else intact.

6) the moment the actor turns sideways or otherwise gets covered up, F.R. can’t recover, and for that, you have to go to an all-manual method. It takes more than 9 power windows to duplicate what Face Refinement does, and while it can be done, it takes time and a lot of thought to pull it off effectively. Always remember that anything Face Refinement can do, you can also do manually… but it’s a lot of work.

If it’s popping off on the last frame, then the last-ditch suggestion I’d have is to start on the last frame and THEN begin the F.R. analysis, and sometimes (for mysterious reasons) it will work better starting at the end than at the beginning.

[Marc Wielage]

[Tim McCallister] “In the vein of this thread, I too had some old B&W 16mm transferred –> 5K (3840×2880, 4:3 Aspect Ratio, ProRes4444 10bit) but unfortunately due to my naivete, it was scanned into .709 rather than .2020.”

Rec709 or Rec2020 is not exactly a factor for film scans per se. The usual practice is to scan in 10- or 12- or even 16-bit Log, usually in a Lossless or near-lossless format like DPX or EXR or ProRes 444 (which can be 12-bit). Log can be corrected in a display-referred environment to Rec709 or HDR or whatever you want.

My experience is that 16mm tends to get really, really grainy when you stretch out the dynamic range to the 1000 nits or so of HDR or Dolby Vision, so be prepared for using some NR. One tactic is to apply the NR only to the highlights and be extremely careful about artifacts and loss of detail. There are a lot of factors for NR, so it’s fair to say it’s not a simple subject, and film restoration people often differ on how much NR is too much.

I would not worry about HDR unless you have an actual distributor demanding it, and unless you actually have a calibrated color-critical HDR display, like one of the top-of-the-line Flanders or Sony BVM monitors. HDR is challenging and not cheap or easy to do correctly.

[Marc Wielage]

[Robert Withers] “Premiere was just too flaky and I’m not so sure about Resolve. I would try Avid but doesn’t run on Catalina yet.”
Actually, Avid MC was just certified for Catalina this week, I believe. Whether it’s the right program for you is a matter of opinion, but users are talking about it on the Avid Editors of Facebook group right now.

[Marc Wielage]

[Juan Juig] “Would you say that the bitrate of DNxHD 185x is high enough for an archive master file? I’m thinking of also creating DNxHR 444 from now on.”
No, I think DNxHR 444 is a much better choice.

[Marc Wielage]

[Juan Juig] “Since I work on Windows machines most of the times, DNxHD and DNxHR are definitely better options for me, but I read some articles about CBR vs. VBR and many claim that VBR is always better, since it distributes bitrate among frames unevenly. Now, am I right to assume, that this mostly applies to smaller files, like h264 files with low bitrate? I tend to export those VBR 2-pass.”
The advantage as I see it in choosing Constant Bit Rate for ProRes (which is an option, at least on Mac OS), is that you can then use a utility like CineXTools to repair just part of the file rather than re-render the entire thing:

https://cinedeck.com/cinedeck.com/cinextools-2/

I don’t think QC companies will reject files for being CBR or VBR — they’ll go either way.

[Marc Wielage]

I’d have to see the keyframe window to know. My guess is you have an unwanted correction at some point, and it’s causing the shape to change or the correction to change. In a lot of cases, I wind up doing a lot of manual tracking on top of automatic tracking, but note that the Tracking keyframe window is different than the Color keyframe window. The Color keyframe window can affect the Tracking under many conditions.

[Marc Wielage]

You could do that if you pulled the entire file into Resolve, then on the Deliver page set in and out points for each segment you want to render, and enter a custom name for each segment. I would use the exact same file format used for the film scan without any color or sizing change, and the results should be “virtually lossless” depending on what file format as used.

What is the file format?