How to change someone's face in a video without the voice drifting out of sync
Ask Vizard Agent to test the change on a single shot with two different models before touching the rest, and to keep the original audio separate so it can be re-aligned afterwards. Changing a face is not a grade or an overlay: the frames are remade, and remade frames do not necessarily arrive on the same timing as the ones they replaced.
What do you type into Vizard Agent?
Ask Vizard Agent for the test before the job. A change judged on a single shot at full size costs a small fraction of doing the whole video, and it tells you whether the approach works at all before anybody spends time repairing sync. Say how strong a change you want, too, because a light retouch can leave the timing alone.
Prompt
What does Vizard Agent actually do?
Softening a face, removing a beard or changing someone's apparent age all mean the picture is generated again, while the voice isn't, so the two tracks can end up describing slightly different moments. The comparison comes before the commitment, and the sync check comes after every pass, not at the end. In the session behind this article it:
- Reviewed the video, extracted frames of the subject and read the transcript so the speech timing was known.
- Tested the facial change with two different models on one shot, then compared the results side by side and zoomed into both faces, because two models fail differently.
- Tested consistency between blocks of the regenerated footage.
- Verified the result was still in sync with the audio, then checked separately that the gestures still matched what was being said.
- Applied the chosen treatment to every shot of the subject and compared consistency across shots, not just within one.
When does this not work well?
Some changes cost more than they buy. If the subject talks in close-up throughout, every frame of every shot has to be regenerated, and the accumulated drift and per-shot inconsistency can be worse than the thing you were fixing. Long takes and many shots of one person add to it, and when lip sync is critical a dub or a cutaway may be safer.
Common questions
Why does the sync break at all? Because the picture is regenerated, not adjusted. Vizard Agent keeps the audio aside for that reason.
Is a light retouch safer? Much. Vizard Agent will tell you when the lighter route gets you what you want.
Will he look the same in every shot? Only if it is checked. Vizard Agent compares across shots as well as within one.