Seedream v4 Edit for AI influencers
ByteDance · Image generation · bytedance/seedream-v4-edit
The same model, but it takes input images. This is what makes a face stay a face.
Not currently routed on InfluAI Labs. The endpoint answers on our key; the product runs Seedream v4 today.
- Input
- Up to 2 reference images
- Output
- Single image
- Typical render
- 15–40 seconds
- Used for
- Identity lock, product shots
- Cost here
- 1 credit per generation
The identity anchor
Every photo a character produces is generated against its own reference frame through this endpoint. The scene, wardrobe, lighting and pose come from the prompt; the bone structure, skin, eyes and hairline come from the image. That is the mechanism behind the claim that a character does not drift.
Product shots that keep the label
Upload a product photo and it joins the reference set as a second image — the character first, the product second. The result is your character holding your actual product, with its shape, colour and branding intact, which is the version a brand will pay for. Describing a product in words gets you a different product.
Questions
Can I make an AI influencer hold my product?+
Yes. Upload the product photo in the studio and the character is generated holding it, with the label readable and the packaging unchanged. It is the fastest route to UGC-style ad creative without shipping samples to a creator.
Run Seedream v4 Edit against a face that never changes
Pick a character, describe the shot, and publish it to Instagram, TikTok or YouTube on a schedule. The model is one line of the pipeline; the identity lock is the rest.
Get started