Avidemux can also do, but lossless join can fail because segments must be also done with same encoding settings and lossles cuts must happen on Keyframes. Other option is TVAI output as FFV1, import and merge FFV1 segments into your video editor and output result lossless for further processes, or do the final encode into video editor. Audio tracks can be added als final step from original video file by remuxing.
In any case at least one lossy encode happens, if not in TVAI then somewhere else I assume you donât want to keep all your final upscales uncompressed, but you can do if you have enough storage space all processing steps lossles and delete all process data at the end.
Or what you alo can do is ignore lossels cuts, cut scene based into a video editor, export segments lossless for use into TVAI and do the join into video editor.
Under âAll Sourcesâ I âselect allâ (function in TVAI) loaded videos I imported AND then I select a preset, which is then linked to all videos.
The I press âQuick Exportâ.
Itâs strange that even applying a preset doesnât work.
So I wrote a PowerShell script to process all files in a folder, even fixed a quirk of TVAI FFMPEG.
I noticed that if you start SLm processing and donât open the preview, the video is generated a little faster. Perhaps this is due to the fact that I have a generally weak PC.
So I renewed my subscription for VAI but am starting to wonder if I wasted my money. We used to see a beta every 2 weeks. Lately though the updates have been more infrequent. As of now we are at 17 days post 7.0.2 release and no beta in sightâŠ
There are a few optimizations that can be made to the current model to improve itâs speed without losing any quality. Weâre waiting on those optimizations now.
Every 2-weeks a beta version? Ha! I also recall that every 2-weeks, a full release versions should be deployed too (so far, since January '25, TL has only released 1 version every 4-5 weeks⊠). I also recalled a much more detail roadmap, yEaRs back, that had a list of upcoming models/features on the horizon for the current year.
Now, SL & SL Mini is said by staff, âSL is the roadmapâ for now. Huh?! So, for the 2H25 (and maybe for the rest of the year), only one local model (maybe one more) before the year ends?
Normally the team members/the stuff attend active the thread(s) and give (sometimes) answers/information.
Perhaps it is on the end not possible to increase the render speed for amateur users without losing quality.
I also donât know, how fast they could render on their own cloud render machines.
As I said before, I donât want to make a very high invest in cloud render costs per minute.
I guess, this is also the case for all other âamateurâ users.
But, we all would be happy to get in time a âhonestâ information about the status. This is better than hide
I tried SLm on a few old videos and concluded that the model is very raw and unsuitable. Problems: it distorts the background and makes the grass look unnatural. Peopleâs faces often donât look like themselves.
Glasses, beards, and jewelry on people often disappear. Texts turn into alien hieroglyphs. Small elements like pairs, stubble, and scratches disappear. Considering how much time it takes, Iâm disappointed.
). I also recalled a much more detail roadmap, yEaRs back, that had a list of upcoming models/features on the horizon for the current year.
