Generate image
Generate an image from a text prompt, optionally guided by reference images and actor, product, or visual-style entity ids. When reference images are provided, the prompt describes the desired transformation. VideoGen automatically routes each request to the most effective state-of-the-art image model for your prompt, reference images, entities, and quality tier, so you don’t pick a model.
Authentication
Request
Optional file ids of reference images (e.g. ["vg_file_..."]). Upload files first via POST /v1/files/upload, then pass the returned ids here. Maximum 4 images. When provided, the model uses these as guidance for generation.
Optional actor, product, or visual-style entity ids (e.g. ["vg_enti_..."]). The model uses each entity as identity/reference the same way in-app image generation does. Can be combined with imageFileIds. A missing id returns not found; an inaccessible id returns a permission error.
Aspect ratio for the generated image. Defaults to 16:9 when omitted.
Image generation quality tier. Optional; when omitted, your account’s Default AI quality for images is used (change it at https://app.videogen.io/settings/account).
Controls how content-policy rejections are handled during generation.
Controls whether the VideoGen watermark is applied to the output. AUTO applies the watermark unless you have a Pro plan. VIDEO_GEN always applies it. NONE removes the watermark (requires Pro; returns an error if you don’t have it).
When true, generated files are temporary. Temporary files are guaranteed to be available for 24 hours, after which they may be archived at any time. Temporary files are not analyzed (no description, transcript, or embedding will be generated), so they will not appear in search results. Defaults to false.
Response
Execution accepted; poll until complete.
Execution id (e.g. vg_tool_...).