Sponsored brands video is a SKU plate plus a hook, not a film
Seedance body, hook pack openings, captions. Movie Mode is the wrong object.
Sponsored Brands video is a SKU plate plus a hook, not a film. The body is one locked product still put into motion. The openings are a pack. Captions go on after. Movie Mode is a chained-scene film from a brief. That is a different object.
Agencies keep routing SKU ads through a movie pipeline because the client said "video." The client meant a packshot that holds, a first-second that survives mute scroll, and words on screen — not a logline.
The body is one still, held
Start with the plate you would print on a listing: square-on SKU, label readable, no invented type. Animate that file. Do not prompt a new bottle.
The image-to-video row on Seedance 2.5 is built for that pass: a required still, first frame (and an optional last frame), native audio, 4s–30s, 480p or 720p. Catalog copy: one frame into continuous motion without stitching shorter clips. Listed credits sit at 47; billing is resolution-based per second. Iterate the still. Then spend the take.
720p is the family ceiling. If the label is wrong in the still, every second of motion is a wrong label with a soundtrack.
The generator surface is image-to-video: upload is frame one, the prompt describes motion, not a new product.
Openings are a pack, not five more bodies
The first three seconds are a different SKU from the body. Regenerating Seedance five times to "try a new hook" is five identity passes you did not need.
Create a branded hook pack plans N distinct openings, charges once for the set, and renders each clip as 1080p portrait on Vidu Q3 with the line burned on as large TikTok-style text. That is the test set. Attach the winner to the Seedance keeper. Do not rebuild the bottle to match the energy of each open.
Pack billing is the product: one charge, shown before confirm. Five separate generates are five charges and usually five cousins of the same line. The hook pack charges once for the set is the bill.
Captions after the take
Native audio on the Seedance body is a mix, not a subtitle file. Speech-to-captions is add captions to a video — auto-transcribed, styled, burned in. A claim nobody says (price, ASIN-shaped line, "shop the bundle") is not transcription. Write it. Burn one fixed line with text overlay, or several timed lines with timed text overlays.
Do not ask the video model to typeset the offer. On-screen text in generated video is why: a still has to be right once; a clip has to be right every frame.
Movie Mode is the wrong object
AI Movie Maker is Movie Mode: a logline into a multi-scene short — script, shots, voice, music, edit. Typical outputs run 30 seconds to 3 minutes. That is a brand film. It is not a Sponsored Brands unit.
A film pipeline will invent a kitchen and a second SKU. Keep Movie Mode for jobs that are actually films. For this unit: lock the plate, I2V the body, pack the opens, caption after.
FAQ
Can I run the SKU body inside Movie Mode if I keep it to one scene?
No. Movie Mode is a chained-scene recipe. One scene inside a film tool is still a film tool. The Seedance I2V row is the body. Use it.
Does the hook pack replace the product take?
No. The pack is N openings with burned hook text, Vidu Q3, 1080p portrait, one charge. The body is the Seedance still-to-motion pass. Cut the winning open onto the keeper. Do not ask the pack to hold a label across 15 seconds.
Why not generate the offer text in the Seedance prompt?
Because the offer has to be editable tomorrow. Overlay text is typed, timed, and re-exported without regenerating the bottle. In-frame type is a new generate every time legal changes a word.
Is 720p a problem for a shopping ad?
On this Seedance family, 720p is the listed ceiling. If you needed a 4K cinematic plate, you needed a different row and, usually, a different job. Sponsored Brands is a SKU that has to match the listing, not a 4K film.