Seedance2 Multimodal Reference Video Page
Built for highly controllable generation. Users provide reference images, videos, and audio, then combine them with a prompt for more stable results.
This page is designed as a more controllable creation entry. It suits brands, ecommerce teams, and content teams that use existing assets to constrain results instead of relying only on pure prompt trial and error.
Support reference images, reference videos, and reference audio, while clearly showing asset limits and format rules at the page level.
Guide users to explicitly mention image 1, video 1, and audio 1 in the prompt so the model understands asset relationships.
Constrain generation with product images, sample shots, and background music, suitable for brand ads and product showcases.
Use cases
I have prepared typical scenarios that can be used directly based on the target users of this page.
Brand product ad film
Brand marketing team / ecommerce design teamUpload product images, sample camera shots, and background music to generate assets closer to an internal brand proposal film.
Product detail page video upgrade
Cross-border ecommerce / Amazon store teamCombine static product images with existing demo videos to quickly create a more rhythmic product showcase video.
IP character consistency short film
Comic IP team / game launch teamUse character images and action reference video to keep visual consistency and reduce character drift.
Holiday marketing custom video
Event operations / private community content teamUse existing posters, event materials, and holiday music to quickly generate a video version suitable for holiday promotion.
Upload reference assets and then control the generated result. Supports images, videos, and audio as reference assets.
Reference assets
Upload images, videos, and audio directly. After copying a screenshot or image, click this area and press Ctrl/Cmd + V to upload it too. The system uploads to R2 first, then submits the URL as a reference asset.