If changing one line of a script regenerates an entire video, I’ll hesitate before changing that line. The price of the operation has changed how I use the editor.
I’d rather review a script and rough storyboard before paying for the final clips. Once a clip is approved, keep it. If I change the voiceover, explain whether that also requires new timing or new visuals. “Regenerate” shouldn’t conceal how much work will be thrown away.
There are engineering choices underneath this: caching, asset reuse, model routing and smaller intermediate outputs. But the person using the tool needs choices they can understand, such as a cheap preview or a final render.
Waiting has a similar problem. Tell me which step is running and whether I can cancel it. Don’t show a percentage that has no relationship to the work. If three of five clips are ready when a provider fails, show me those three and let me retry the missing ones.
I want to experiment without treating every click as a small financial decision. Clear estimates and recoverable steps help more than hiding the bill until the end.