Verified against Veo 3.1 · 2026-07-28
Generate a fashion lookbook motion clip that describes fabric, not just the model
A Veo 3.1 prompt for an apparel motion clip built around explicit fabric-physics language, since garment movement is one of the most common places generated video shows visible cloth-simulation artifacts if the movement is left unstated.
The prompt
Ready to copy — highlighted parts are example details you can swap.
Write a Veo 3.1 prompt for an 8-second fashion motion clip showing a garment in movement, structured to describe how the specific fabric behaves rather than only naming the garment and the model wearing it. SUBJECT AND GARMENT a woman in her late 20s with a straight, confident walk, wearing a floor-length silk slip dress in deep emerald. Name the fabric explicitly — silk, structured cotton twill, heavy wool — not just the garment category, since the fabric's weight and stiffness is what determines how it should move, and an unnamed fabric leaves Veo guessing at cloth physics that then frequently render as either stiff, unmoving fabric or an unnaturally fluid drape that does not match the material described. MOVEMENT turning a quarter-circle away from camera and back, with the hem lifting slightly and settling half a beat after the turn completes, showing real weight as the garment responds to that movement. Describing the fabric's actual response — "the hem lifting and settling half a beat after the turn, not simultaneously" — gives Veo a physical lag to render, which is what separates cloth that looks like it has weight from cloth that looks pasted onto the body and moving in lockstep with it. CAMERA A single controlled arc, a slow arc from front-facing to a three-quarter side profile, at a pace slow enough to actually resolve the fabric's movement rather than blurring through it. A camera arc that outpaces the garment's own movement is a common cause of the fabric appearing to warp or smear, since the model is rendering two different speeds of motion in the same frame. ENVIRONMENT AND LIGHTING a plain studio backdrop, one soft side light raking across the silk to catch its sheen. Name where the light catches the fabric specifically, since fabric texture — the sheen of silk, the texture of wool — only reads clearly under a described light interaction, not from the fabric name alone. STYLE AND MOOD clean editorial, muted color grade, high-fashion catalog feel. AUDIO a faint room ambience, the soft sound of fabric moving, quiet, no music unless no music — this will be scored separately after export specifically calls for a track — most lookbook clips are re-scored after export, so leaving the native audio minimal avoids generating a track that then has to be stripped out. WHAT TO AVOID Do not describe more than one garment changing or layering mid-clip — a single eight-second clip showing one outfit change on top of the described movement is asking for two separate physical events at once, and the transition between them is the most likely place to see morphing or an unnatural jump cut inside what was meant to be one continuous shot. OUTPUT The finished prompt, followed by one line naming the specific fabric-behavior detail chosen, so it can be checked against the garment's real material before generating.
Customize
Optional — swap in your own details for the highlighted parts above.
Why this works
Cloth simulation is one of the harder physical behaviors for a generative video model to infer from a garment category alone, which is why naming the specific fabric and describing how it physically responds to the movement — not just naming the movement itself — is the load-bearing part of this prompt. A garment named only as "a dress" gives Veo no signal about weight or stiffness, and the model's default guess tends to land in one of two failure modes: fabric that looks glued to the body and moves in perfect lockstep with it, or fabric that flows with an unnaturally uniform fluidity that no real material actually has, both of which read as artificial the moment a viewer who has ever worn clothing looks closely. Stating an explicit lag — the hem lifting and settling half a beat after the turn completes, rather than simultaneously — gives the model a concrete timing offset to render, and that offset is what visually communicates mass and drape; without it, movement and fabric response happen in the same instant, which is physically impossible for any fabric heavier than a plastic sheet and is a fast, subconscious tell that the footage is synthetic. Matching the camera's own speed to a pace that can actually resolve the fabric's motion addresses a separate but related failure: when the camera arcs faster than the garment's described movement, the model ends up rendering two different velocities inside the same frame, and the fabric is usually where that mismatch shows up first as a smear or a warped silhouette, since it is the most physically soft and least rigid element in the shot. Restricting the clip to one garment event — one movement, one fabric response, no mid-clip outfit change — is not a stylistic limitation but a direct response to how morphing artifacts happen in fixed-duration generation: two separate physical events inside eight seconds force the model to render a transition between them, and a transition between two different garments on the same body is exactly the kind of discontinuity that produces a visibly unnatural blend rather than a clean cut, because nothing in the prompt told the model a cut was supposed to happen there.
Verified against
Veo 3.1 3.1 · 2026-07-28
Changelog
- 2026-07-28 — Initial publish, verified against Veo 3.1.
Need this built into your business?
If a prompt isn't enough — what Scult builds, built and maintained for you — that's Scult's day job.
EXPLORE WHAT SCULT BUILDS
