Hey everyone,
I’m running into a consistent issue with the Unifab upscaler and wanted to see if anyone else has noticed it.
Every time I upscale a video or clip in Unifab, the exported result is missing the final 1.5 seconds, regardless of file type, duration, frame rate, or resolution. This means the last line of audio or final visual beat gets chopped off every single time.
At this point I’ve tested:
MP4 and MOV
Multiple resolutions
Multiple frame rates
Both long videos and short clips
The outcome is always the same: exactly the last 1 to 2 seconds are gone.
I’ve been padding the tail end manually before importing just to make sure my final clip isn’t missing anything. But that becomes really tedious when processing multiple short clips for shorts, speeches, or music drops.
From what it looks like, the model or encoder pipeline is closing early — as if Unifab thinks it reached the end of the frames before the buffer finishes. It consistently stops processing just before the true end of the timeline.
Ideally:
Unifab should force the model to process every single frame up to the real end of the clip so nothing is lost.
Temporary suggestion:
Even a simple checkbox option like “Auto-pad end by 1.5 seconds (silent/black)” would be a life-saver until the underlying issue is fixed. Then the program could auto-trim the extra silence afterward.
This is absolutely repeatable. If needed, I can upload a sample source + export showing the exact missing frames.
If anyone else is experiencing the same thing, please chime in so the devs can see it’s not isolated.
Thanks!
