Thank you for your work @"mondomonger" , I really appreciate that.
I know that you prefer parsing movies, know your argumentation. I watched your face set and in my opinion the images are mostly dark and low-contrast.
Please check out my face set of the same actress:
https://mrdeepfakes.com/forums/thread-jessica-chastain-need-feedback?pid=6439#pid6439 Sorry that I couldn't upload it to the file hosting you mentioned, it's banned in my country. This face set is a perfect example to compare your and my approaches.
You could write a great article with approach comparison. If you take my face set into work and train another model and then apply it to the same target video, you could see if my source and face set is worse or not.
Also, I found several red-haired models with similar faces. I would be happy if you make some more clips based on these clips, please see corresponding topic:
https://mrdeepfakes.com/forums/thread-jessica-chastain
Let me know if I can do some more. For example, I could sharpen any of your face sets or intermediate aligned output faces so that you could check out if it helps making sharper videos. For me it's easier to sharpen a bunch of photos that sharpen a video. It also makes sense because only faces are processed, and not the background.