Three hundred and twenty-one characters, and it beats a much longer prompt that says "claymation style" ten different ways. Worth understanding why.
The first word of the prompt is the technique, and the fifth-to-last clause is the frame rate. Those two bracket everything else. "Stop-motion claymation" sets the search space; "stepped 12fps motion" sets the thing that actually distinguishes stop motion from a smooth render of a clay-textured character. Most people write the first and forget the second, and then wonder why the output glides.
The fingerprints are the second half of the same idea. A model deciding how to render clay has to choose between clay-as-material and clay-as-object-someone-touched. Fingerprints choose for it, and they cost four words. Felt walls do the same job for the set: felt is a material nobody uses to build a real room, so naming it says "this is a miniature" without spending a sentence saying it.
The action is written as three separate beats in order, and it is worth counting them against the runtime. Carries the stack across, sets it down, looks at camera and blinks. In five seconds that is a little under two seconds each, which is comfortable. Four beats would be tight and five would come back slowed down, because the model fits everything it was asked for into the seconds it was given.
The last beat is doing something specific. Looking at camera and blinking is a held moment, and a held moment is how you end a short clip without it looking cut off. It also puts the character’s face in the frame at the last frame, which is the frame anybody scrubbing a grid of thumbnails will see.
The sound is three items and it is mixed: pizzicato strings as the score, clay squeaks and a plate clink as the foley. The squeaks are the audio equivalent of the fingerprints — a sound only a physical puppet would make. On a model that writes audio with the picture, that is a texture cue arriving through a second channel.
Video