Should you use wide shots or close-ups when generating AI video?

Default to wide environment shots when generating AI video, and avoid close-ups of characters performing. Tight framing exposes uncanny facial and motion artifacts that wide shots hide — pull out rather than push in. Keep close-up acting beats on practical footage, and let story continuity, not visual ambition, decide each frame.

Pull out, don't push in. AI models render environments — oceans, landscapes, exteriors, weather — convincingly at wide framing, but character faces doing any kind of acting break down under a tight lens: micro-expressions, eye lines, and mouth movement are where the artificial look surfaces. So assign AI the wide establishing and environment shots, and shoot performance close-ups practically whenever you can. One documented hybrid production ran exactly this split: the filmmaker shot a practical emergence-from-water close-up in a 6x6 foot studio pool (too small to hide the set edges, and the actor could tread water for only one to two minutes per take), then generated wide red-ocean and shoreline shots to place that same action inside a larger world.

Prompt the camera position explicitly. Write the framing and movement into the prompt — "wide establishing shot," "slow pull-back revealing the coastline" — rather than leaving camera distance to the model. Inside invideo, treat this as a conversation with the invideo agent: generate a wide, tell it what to keep and what to change, and iterate until the framing sits right. If you have practical footage, upload stills from your own shoot as references first so the AI wides match your color, contrast, and look — the same production found reference-anchored generations dramatically more realistic than text-only prompting. Once the first wide establishes the look and effect, subsequent shots in the same scene generate faster and more accurately.

Use your close-up as the anchor and build the world outward. When your practical shot is confined to a small set, treat it as the locked shot and ask the invideo agent for wide environment shots that place the same action in a bigger space — this is how a 6-foot pool becomes an open ocean on screen. Add environmental continuity cues between the AI wides and your practical close-ups (matching haze, matching light direction) so the cut between them feels like one location.

Let story logic pick the wide, not visual impressiveness. Choose each AI-generated wide by what the narrative needs next — if your practical close-up shows a character crawling out of water, the matching wide is a shoreline, because the two shots have to read as one continuous scene. A striking wide that breaks continuity costs you more than a plainer one that matches.

Watch some of these to see what works for you:

If you push into this shot, or if the shot is too close, and you're going to get the actual character doing any kind of acting, that's when things look fake.

— an independent filmmaker documenting a hybrid practical-AI production