Free · copies cast, sets and prompts to your workspace · you pay only when you generate
About this template
Each character is first built as a model sheet image, generated with GPT Image 2.5 Flare at 2K. Pink's sheet places a title block with name, age, role and personality next to a color palette for hair, eyes, skin and outfit, plus labeled front and side bust portraits and a full-body walking view. James and his golden retriever get a matching sheet, and the street is a separate empty plate with no people or signage. In the video prompt, identity is taken from the attached sheet while the pink wardrobe is described in text, so the face holds steady even if you change the clothes.
The scene is written as a single continuous 20-second oner in vertical 3:4 with an explicit no-cuts instruction, drafted in Seedance 2.5 at 480p. A global block sets the look up front: harsh noon sunlight, bright pink tones that pop off white sidewalks and shopfront glass, and a camera that glides smoothly. The references are then listed in attachment order with @Image tags, covering Pink, James with the dog, the sunny street and the coffee cup, each with notes on what to match. A list of diegetic sounds such as heels, a light shoulder thud, a speed-ramp whoosh and splashing liquid gives the model cues for the moment the sunglasses, cup and phone hang in the air before landing on the sidewalk.
The canvas holds 7 reference image nodes, the video generation node and three 20-second instrumental music drafts in French house, retro funk and dreamy electronica styles. To remake it, duplicate the template, replace the character sheets with your own talent and swap the cup reference for your product. Keep the structure of the video prompt, meaning global look, ordered references, beat-by-beat action and sound list, and rewrite only names, wardrobe and prop details. Generate, choose a take, then add music and an end card on the timeline and export to mp4 or FCPXML.