Demo — inputs & output

Ref2VA15s4× Picture

Prompt structure inspired by Inspired by fal.ai H3 Guide

Desert Highway Fashion Campaign

Multi-image luxury brand film — mood, talent, product, and logo each get their own reference job before the timeline runs.

Ref2VA15s16:9fashionluxurybrandeditorial

Who this is for

Fashion or accessories brands with model shots, product stills, and a logo end card — typical 15s campaign cut.

How this prompt is built

Read this first — once you see the pattern, you can write your own without starting from zero.

[References](Official field: reference material notes)
Assign each upload a job before any story beats — mood vs face vs product vs end card. This is what stops the bag from morphing between shots.
[Core idea]
One paragraph pitch: format (16:9), tone (premium, cool), and what to avoid (horror, cheerful lifestyle).
[Process](Official field: integrated_multimodal_description)
Timed ad beats (0–3s, 3–7s…) with camera verbs and hard cuts — editors think in blocks, H3 follows blocks.
overall_soundscape
Diegetic sound only: wind, trunk latch, footsteps — things on set.
non_diegetic_music
Score with BPM and swell points tied to cuts — separate from ambience.

Adapt it for your project

Swap @image3 for your SKU still; repeat “same bag as @image3” in every product beat.
Change desert highway to coastal road or city alley in [Core idea] — keep timing structure.
Replace logo @image4 with your PNG; hold last 2s on brand card.
For fewer assets, drop the male glance beat and extend product close-ups 7–11s.

Where people mess up

Uploading four images but only mentioning two in [References].
Letting dialogue run long — 15s fashion spots rarely need more than one line.
Writing “cinematic” instead of cut times — you get one slow drift.
Merging music description into soundscape — muddies both.

Your reference files (what to upload)

File	Job in the video
Image 1	Mood, desert light, film grain
Image 2	Female lead — face & coat
Image 3	Hero bag — shape & logo
Image 4	Closing logo card

Build official Ref2VA export in Prompt Builder →Generate video →

Full prompt (copy & edit)

[References] @image1 sets mood, desert light, and 35mm film grain. @image2 locks the female lead's face, hair, and camel coat. @image3 locks the black structured bag — handle, hardware, and logo placement. @image4 is the closing logo card.

[Core idea] A 16:9 high-fashion brand campaign beside a vintage cream convertible on an empty desert highway at golden hour. Premium, cool, agile editorial energy — not horror, not cheerful lifestyle fluff.

[Process]
0–3s: Wide static — woman opens the trunk; heat shimmer on asphalt; wind moves her coat hem. Hard cut on downbeat.
3–7s: Medium tracking — she lifts the black bag from the trunk; man in frame shares a brief quiet look, no smile. Truck left slowly.
7–11s: Close-ups — bag hardware glints; her hand adjusts the strap; logo readable. Cut to product on trunk lid.
11–15s: She walks away down the highway; car and man hold frame; fade to @image4 logo on black.

Dialogue: none.

overall_soundscape: Dry desert wind, distant tire hum, fabric rustle, trunk latch click, subtle gravel under boots. No crowd noise.
non_diegetic_music: Minimal electronic pulse with muted kick, 90 BPM, tense luxury mood — swells at 7s, drops for logo at 14s.

New to Ref2VA? Read the MiniMax H3 Ref2VA prompt guide