Fashion and Lookbook Video Prompts

TL;DR: Fashion and lookbook video prompts work by naming the fabric, not just the garment: silk, denim, knitwear and leather move completely differently under wind or a walk, and lighting for texture reads differently than lighting for skin. Below: 26 copy-paste prompts organized by shot type, the two legal risks to know first, and which Prompt Architects plan actually generates video prompts.

What Makes a Video Prompt Actually Read as Fashion?

Four things, and none of them are about picking a "fashion" style keyword. Fabric behavior is the whole game. Silk pools and ripples; raw denim holds its shape and creases at the knee and elbow; a chunky knit sways as one soft mass instead of rippling; leather swings and creaks rather than flowing. A prompt that names the garment ("a dress") gives the model nothing to work with. A prompt that names the fabric ("a bias-cut silk slip") tells it how the thing should move before a single camera term shows up. This is the craft that separates a fashion clip from a generic person-in-clothes clip, and it belongs in the first sentence of every prompt below, ahead of the shot type and the camera move.

Second, lighting choices split by subject, and treating them as one setting is the most common miss in a first draft. Lighting for texture (the weave of a tweed, the grain of leather, the ribbing of a knit) wants a raking, off-axis source that throws small shadows across the surface. Lighting for skin wants something softer and more frontal, closer to how a portrait is usually lit. A single setup rarely does both jobs at once, which is why a detail pass and a portrait-style editorial shot in this same post call for two different light directions below. Our lighting vocabulary guide has the fuller term list if you want more than "raking" and "frontal" to work with.

Third, pace. A fashion cut is slower than an ad cut, and that difference is deliberate rather than a byproduct of budget. The point of a lookbook sequence is letting a viewer actually register drape and proportion, so specify a slow, deliberate walk or an unhurried turn directly. Don't assume a model defaults to editorial timing on its own; its uncorrected default usually leans toward the faster pace of a generic ad cut.

Fourth, color accuracy matters here in a way it doesn't for most other video categories, because color is a purchase decision for a shopper deciding between two near-identical shades. State the exact shade and finish (matte, satin, dust-rose, oxblood) rather than a vague color family, and treat any generated color as a starting reference for mood, not as a swatch a print shop or a customer's screen will reproduce exactly.

What Can't You Legally Generate for a Fashion Brand?

Two constraints, before anything else, because fashion is a category where getting this wrong has real consequences.

Generated video with people in it also raises synthetic-media disclosure questions on several platforms, which we cover in its own section below rather than guessing at a rule that varies by platform and keeps changing.

What Goes Into a Fashion Video Prompt?

Our anatomy-of-a-video-prompt guide breaks a video prompt into seven parts. Mapped onto fashion, that's the wardrobe (garment plus fabric, named specifically), the action (which shot from the list below), the camera (position and movement), the lighting (texture vs. skin, per above), the mood and color palette, the setting, and duration or pace. Cover all seven in a prompt and the shot type barely matters; skip two or three and even a good camera term won't rescue a vague garment description.

One mechanical note worth flagging before the prompt library: aspect ratio, clip length, and whether audio generates at all are usually request parameters on the vendor's own API or app, not something you can reliably set in prompt text. Google's own Veo 3.1 documentation states its aspect ratio is 16:9 or 9:16 and no other value. That's a parameter you set outside the prompt, not a sentence you write inside it. Set those controls where each vendor documents them, and use the prompt text below for everything else.

How Do You Prompt a Walk-Toward-Camera Shot?

Name the fabric first, then the pace, then how the fabric responds to the walk itself.

Silk, slow walk toward camera:

A [MODEL DESCRIPTION — build and styling only, not a named person] walks slowly toward camera down a plain [SETTING] runway, wearing a bias-cut [COLOR] silk slip dress. The silk ripples and pools with each step, catching light along its folds. Steady dolly-in matching the model's pace, [LIGHTING] key light, [DURATION] seconds, [ASPECT RATIO].

Tailored wool coat, slow walk toward camera:

A [MODEL DESCRIPTION] walks at a deliberate, unhurried pace toward camera on a [SETTING] backdrop, wearing a structured [COLOR] wool coat that holds its shape and only sways slightly at the hem with each stride. Locked-off camera at chest height, soft directional key light, [DURATION] seconds.

Raw denim, slow walk toward camera:

A [MODEL DESCRIPTION] walks toward camera on a [SETTING], wearing raw [COLOR] denim that creases sharply at the knee with each step and holds its structure rather than flowing. Camera tracks backward at walking pace, even studio-style lighting, [DURATION] seconds.

How Do You Prompt a Turn That Reveals the Back of a Garment?

State what the turn is meant to show, namely the back of the garment, so the model treats it as the point of the shot, not an incidental pivot.

Backless silk gown, reveal turn:

A [MODEL DESCRIPTION] stands facing camera, then turns a slow 180 degrees to face away, revealing a backless [COLOR] silk gown with a low draped back. The fabric shifts and settles as the turn completes. Static camera, [LIGHTING], [DURATION] seconds.

Tailored blazer with back vent, reveal turn:

A [MODEL DESCRIPTION] turns slowly away from camera to show the back of a fitted [COLOR] blazer with a center back vent, which opens slightly as the shoulders rotate. Camera holds a fixed mid-shot at chest height, [LIGHTING], [DURATION] seconds.

How Do You Prompt Fabric in Motion: Wind, Walk or Drape?

This is the section worth slowing down on, since it's where the fabric-first rule from the top of this post earns its keep. Pair these with the terms in our camera movement vocabulary guide if you want more control over how the camera itself moves while the fabric does its own thing.

Silk scarf in wind:

A lightweight [COLOR] silk scarf lifts and ripples in a steady breeze against a [SETTING] backdrop, catching and releasing light as it moves. Slow-motion feel, static camera, [DURATION] seconds.

Heavy denim jacket, minimal fabric movement by design:

A [MODEL DESCRIPTION] stands in a light breeze wearing a heavy [COLOR] denim jacket that holds its structure and barely moves except at the open hem, in contrast to hair or a loose scarf moving freely nearby. Static camera, [LIGHTING], [DURATION] seconds.

Chunky knit sweater, single-mass sway:

A [MODEL DESCRIPTION] turns gently in place wearing an oversized [COLOR] chunky-knit sweater that moves as one soft mass rather than rippling, the ribbed hem swaying slightly behind the turn. Slow orbit camera, warm directional light, [DURATION] seconds.

Leather trench coat, swing and settle:

A [MODEL DESCRIPTION] walks briskly across a [SETTING], wearing a full-length [COLOR] leather trench that swings open at the hem and settles heavily rather than flowing, with visible sheen shifting across its surface as it moves. Camera tracks alongside at matching pace, [DURATION] seconds.

How Do You Prompt a Detail Pass on Texture, Stitching or Hardware?

Detail passes are lit for texture, not skin. Say so directly, and specify the macro framing so the model doesn't default to a wider shot.

Denim stitching, macro pass:

Extreme close-up, macro lens, on the contrast-stitched seam of raw [COLOR] denim at the hip pocket, texture and thread visible under raking side light. Slow push-in, [DURATION] seconds, no visible face.

Hardware close-up (buckle, zipper, clasp):

Extreme close-up on a brushed-metal [HARDWARE TYPE — buckle, zipper pull, clasp] against [COLOR] leather, catching a hard specular highlight as the camera moves in a slow arc. [DURATION] seconds.

Knit texture, close pass:

Close-up, shallow depth of field, on the ribbed texture of a [COLOR] chunky-knit sleeve, raking side light picking out individual yarns and shadow between ribs. Slow lateral pan, [DURATION] seconds.

How Do You Prompt a Flat-Lay Coming Together?

Flat-lays are usually asked for as a static shot with objects settling into frame. Treat it as items landing into place on their own, not a person handling them, unless you specifically want hands in frame.

Outfit flat-lay assembling itself:

Top-down static camera on a plain [SETTING] surface. A [COLOR] top, [COLOR] trousers, and a pair of [SHOE TYPE] settle into a flat-lay arrangement one at a time, each item landing softly into frame. Even, shadowless overhead light, [DURATION] seconds.

Accessories landing into a flat-lay:

Top-down static camera. A watch, a pair of sunglasses, and a folded scarf drop gently into frame one at a time onto a [SETTING] surface, settling into a clean flat-lay arrangement. Soft overhead light, subtle shadow, [DURATION] seconds.

Editorial Mood Shot Prompts

Editorial mood shots are where lighting choice matters most. Pick a direction and commit to it rather than defaulting to flat, even light, which is the safest choice and also the most forgettable one.

Single-source moody editorial:

A [MODEL DESCRIPTION] stands against a dark [SETTING] backdrop, wearing a [COLOR] [GARMENT], lit by a single hard side light that leaves the far side of the frame in shadow. Slow push-in, [DURATION] seconds, muted color grade.

High-key soft editorial:

A [MODEL DESCRIPTION] stands against a pale [SETTING], wearing a [COLOR] [GARMENT], lit by a soft, near-shadowless frontal light. Static camera, gentle handheld drift, [DURATION] seconds, bright airy color grade.

Window-light editorial:

A [MODEL DESCRIPTION] stands beside a large window in a [SETTING] interior, wearing a [COLOR] [GARMENT], lit only by soft daylight falling across one side of the face and the fabric. Static camera, natural color grade, [DURATION] seconds.

Street-Style Candid Prompts

Ask directly for the informal handling (an unposed glance, an unhurried stride) since a video model's uncorrected default leans toward posed and locked-off.

Crosswalk candid:

A [MODEL DESCRIPTION] crosses a [SETTING] street at an unhurried pace, wearing a [COLOR] [GARMENT], glancing briefly away from camera rather than posing directly at it. Handheld-feel camera following alongside, natural daylight, [DURATION] seconds.

Café candid:

A [MODEL DESCRIPTION] sits at an outdoor café table wearing a [COLOR] [GARMENT], adjusting a sleeve while glancing off-frame, unposed. Static camera at eye level, natural window light, [DURATION] seconds.

Campaign Hero Prompts

A hero shot is usually the single frame carrying a whole campaign. Keep the composition simple enough to hold text or a logo if it's meant to run as a still-frame pull.

Studio backdrop hero:

A [MODEL DESCRIPTION] stands centered against a plain, unbroken [COLOR] studio backdrop, wearing a [COLOR] [GARMENT], in a strong, still stance. Static camera, single dramatic key light, [DURATION] seconds, negative space to the left of frame.

Outdoor golden-hour hero:

A [MODEL DESCRIPTION] stands in an open [SETTING] at golden hour, wearing a [COLOR] [GARMENT], backlit with warm rim light along the shoulders and hair. Static camera, slow ambient movement in the background, [DURATION] seconds.

Lookbook Sequence Prompts

A lookbook sequence is several looks shown back-to-back at a deliberately slower pace than an ad cut. Hold on each look long enough to register the fabric and proportion before the next one starts.

Three-look sequence, same setting:

A [MODEL DESCRIPTION] appears against a consistent [SETTING] backdrop, shown in three consecutive looks: first a [COLOR] [GARMENT ONE], then a [COLOR] [GARMENT TWO], then a [COLOR] [GARMENT THREE], each held for [DURATION] seconds with a slow turn before the next look. Static camera position across all three, consistent lighting.

Single look per scene, consistent studio:

A [MODEL DESCRIPTION] wears a [COLOR] [GARMENT] in a plain [SETTING] studio, walking a short distance and turning once at a slow, deliberate pace. [DURATION] seconds, even studio lighting, no fast cuts within the shot.

Accessories-in-Isolation Prompts

Accessories usually read best without a person in frame at all, so the object's own material and hardware carry the shot.

Handbag on a plinth:

A [COLOR] leather handbag sits on a plain [SETTING] plinth. Camera performs a slow 180-degree orbit at eye level, revealing the hardware and stitching on each side. Even studio light, [DURATION] seconds.

Jewelry or watch, macro isolation:

Extreme close-up on a [COLOR] [ACCESSORY — watch, ring, necklace] resting on a plain [SETTING] surface, catching a slow-moving specular highlight as a soft light source passes overhead. Static camera, [DURATION] seconds.

Shoes, slow reveal:

A pair of [COLOR] [SHOE TYPE] sits on a plain [SETTING] surface. Camera performs a slow, low-angle push-in from the toe toward the heel, revealing material and stitching detail. Even studio light, [DURATION] seconds.

Which AI Video Models Actually Document Fashion-Relevant Controls?

Not evenly, and none of them publish a "fashion" mode. What matters for this category is whether a vendor documents audio defaults, camera or motion control, and transition-shot guidance, since those are the controls a lookbook shoot actually leans on.

Don't carry one vendor's syntax to another. Kling's shot grammar and Seedance's bracket markers for sound are specific to those models; writing them into a Veo or Runway prompt does nothing. Our best AI video prompt tools guide ranks a wider set of vendors on documented control if you're choosing between more than these four.

Do You Need to Disclose That a Fashion Video Was AI-Generated?

It varies by platform, and every one of these policies keeps changing, so treat this as a starting point rather than a final answer. In the US, the FTC's 2024 rule on fake reviews and testimonials names "AI-generated fake reviews" directly and requires that an endorsement "reflect the honest opinion of the endorser," which is relevant if a fashion clip is ever framed as a customer wearing or reacting to a product. YouTube's own disclosure policy names this directly: "Making it appear as if someone gave advice that they did not actually give." TikTok's newsroom describes new labels for AI-generated content. Meta's own commitment is narrower than it's often summarized as: disclosure for political and social-issue ads "in certain cases," plus separate AI-info labels, not a blanket label requirement across every ad. Our UGC-style ad video prompts guide covers the fabricated-endorsement risk in more depth if you're building anything that looks like a customer speaking.

Worth knowing separately: Google's Veo carries a SynthID watermark on every generation. Not passing a generated clip off as unretouched footage is a provenance fact baked into the file itself, not just good practice.

Which Prompt Architects Plan Includes Video Prompt Generation?

Video Prompt Generation sits on the Advanced and Team plans, not on Pro or the free plan, per the feature comparison table on our pricing page. If you only need still-image prompts for a lookbook's product photography, that's available starting on the Pro plan; video prompting for the shots above needs the higher tier.