The videos

One idea per clip, carried out until it is finished. No narration, no text on screen, no music bed.

The format

Aspect ratio

9:16 vertical, 1080 × 1920 after scaling.

Length

5 to 15 seconds, built from consecutive blocks.

Audio

Generated with the picture. No licensed music, no voice-over.

Camera

Usually a single static close-up. The action moves, the frame does not.

What is in them

The recurring subject is a familiar everyday object whose surface behaves differently than expected — a glass-like glaze over something soft, a frozen interior, a crystalline shell. Something acts on it: a blade, heat, pressure. The clip shows the whole sequence of that action, several completed passes rather than one slow movement, with each pass leaving a visible result.

The sound follows the material in layers: a bright crack where the hard surface gives way, then the muffled sound of the soft interior, then the piece coming to rest. If the sound does not match the material, the clip is not used.

Rules the channel holds itself to

Every clip is AI-generated and marked as such where the platform offers a disclosure.
No real people, no recognisable likenesses, no impersonation.
No text, logos, watermarks or brands inside the frame.
No copyrighted music and no material taken from anyone else.
Nothing depicting harm, and nothing presented as if it were a real recording of a real event.
Clips that fail the technical check are discarded, not fixed up and posted anyway.

Publishing rhythm

A small number of clips per day, limited by how long generation takes on a single machine — a few clips, not a feed filled automatically. Each one is looked at before it is released.