Hiding jump-cuts thanks to AI
Quick example to frame this: a handheld medium close-up of someone talking to camera. An interview for exemple. As soon as you start trimming words/pauses and editing the shot, jump-cuts pop up everywhere.
The classic fixes are zooming slightly on the cut to sell it as an intentional reframe, or a funky transition (glitch, flash, whip, whatever). They work, but they bring a “YouTube craft” look I’m not going for. What I actually want is something that reads like a fiction scene shot in one clean take.
I know about others options:
Premiere’s Morph Cut, DaVinci Resolve’s Smooth Cut, Boris FX Continuum’s Jump Cut Fixer ML. But they are based on "optical flow warp/dissolve" between the two frames. Has anyone here pushed past that and tried something more generative?
The idea would be to actually generate true in-between frames — instead of warping/dissolving between the out-frame of shot 1 and the in-frame of shot 2, generate new intermediate frames with an AI model, closer to a full resynthesis than a warp. Any models that work well for this specifically? A ComfyUI/Stable Diffusion workflow that handles it properly? Maybe even before/after examples you can share?
Thanks for any feedback.