Custom pixel format options and Possibility for true lossless export via Premiere

[Bonny]

Is it possible to choose a custom pixel format option instead of the ones already available in frameserver? (RGB24, 36, YUV2, v210)

I've done a bit of testing with frameserver, trying to achieve true lossless conversion. but after running PSNR and SSIM checks, the original footage and the exported footage always have some differences, without modifying them inside premiere.

This could be due to alot of factors that i'm unaware of so i'm hoping to get some insight here.
i've tried matching the sequence settings as best as i could to original footage, or instead convert the original footage to a matching sequence before importing it to premiere. but the export results are the same. exporting to an image sequence has given better quality matching results, but still not identical results.

I understand this may be an unusual approach, I'm just testing a proof of concept to find a method that preserves the highest possible quality for future projects.


On a side note,
not sure if this issue is on ffmpeg's side or frameserver's (or mine), but when exporting footage as video and NOT as image sequence, ffmpeg will post some errors when comparing original footage to export:

"not matching timebases found between first input: 1/24000 and second input 125/2997, results may be incorrect!"

"Application provided invalid, non monotonically increasing dts to muxer in stream 0: 237 >= 237"

when exporting to an image sequence and comparing, the mismatching timebases error still shows, but the latter doesn't and the test is a lot more accurate.

[XBOX CDK]

When it comes to video transcoding or editing, the principle is to play the video in its original lossless form and then record it. For editing software, the preview is based on the preview image. When you mention "lossless", do you mean the image quality or the time frame rate axis?

Let's talk about the time frame rate axis. You need to use (mediainfo) to check if the source material is VFR (variable frame rate) or CFR (constant frame rate). At least five or six years ago, it was like this. For PR, if your video is two seconds long, with 20 frames per second in the first second and 30 frames per second in the second second, averaging 25 frames per second, it would decode at 25 frames per second. This causes audio-video desynchronization, and the problem becomes more obvious towards the end of the timeline. VEGAS doesn't have this issue. You might need to use HANDBRAKE to transcode the material into a CFR (constant frame rate) and then import it, which will make it consistent. I'm referring to the preview effect here. Frameserver usually uses MEGUI to call AVISINTH to read the preview image. When I used it before, there was no problem, and the frame rate matched, and the preview image in the editing software would change accordingly. If you are referring to the image quality, let me know, and I'll explain that as well. My native language is Chinese, and I used a translation software for this. Please take it as a reference.