How to make a 30 second AI video in one shot
Most AI video models make a few seconds at a time, and longer clips are stitched from several shots. A few now make a full 30 seconds in one continuous shot, which keeps the camera, the light and the people consistent from start to finish.
Which models make 30 seconds in one shot
In ReelCritix Studio: Seedance 2.5, Kling 4.0 and Wan 3.0. The other models make shorter shots and Studio links them, each new shot starting from the last frame of the one before. That works well, but a single shot is smoother.
Write the prompt as a scene, in order
Thirty seconds is long enough for a small story. Describe what happens in the order it happens, and start each new beat with "Cut to:". Describe your main character once, clearly, and keep calling them "the same woman" or "the same man" so the model keeps them consistent.
First scene: a woman in her 40s grips the steering wheel in rush-hour traffic, frustrated. Cut to: the same woman wanders a crowded gym, unable to find a free machine. Cut to: the same woman smiles as she sets up a laptop next to a yoga mat in her sunny living room.
Three or four beats is plenty. More than that and each one gets too little time.
Settings
- Duration: 30 seconds.
- Shape: 9:16 for TikTok, Reels and Shorts; 16:9 for YouTube and the web.
- Sound: on, if the model offers it. Put any spoken lines in double quotes.
What it costs
A 30 second Seedance 2.5 clip costs $4.20 at 480p. On Wan 3.0 it is $2.10 at 480p. Prices are per second, one credit is one US cent, and credits never expire.
Test short, then go long
Before spending on 30 seconds, make 5 seconds of the first beat at 480p. It costs a few cents and tells you whether the look and the character are right. Then make the full clip.
Watch the last few seconds closely. Long generations sometimes wobble at the very end; if yours does, trim the ending in any video editor.