How to use first and last frame images to control where a video starts and ends

To control where a video starts and ends on RelayDance, submit a video task to POST /v1/video/generations and attach your first and last frame images inside metadata.content[], then reference them in the prompt as @image1 and @image2. Each request accepts up to 9 reference images, 3 reference videos, and 3 audio tracks, with clips up to 15 seconds. Poll GET /v1/video/generations/{task_id} until the status is succeeded.

How reference frames are attached and cited

Reference frames go in metadata.content[] and are cited in the prompt using the tokens @image1 to @imageN. To pin the opening and closing composition, place your start image first and your end image second, then describe the motion between them in the prompt while referencing @image1 (first frame) and @image2 (last frame). A single request supports up to 9 reference images, 3 reference videos, and 3 audio tracks. Generated clips can run up to 15 seconds. The API is OpenAI-compatible: 「Set base_url to https://relaydance.com/v1 and keep your OpenAI SDK」, according to the official relaydance.com/docs. See relaydance.com/docs for the full request schema.

Steps to submit a first and last frame task

Follow these numbered steps to build the request end to end.

1. Create an API key at https://relaydance.com/console and set Authorization: Bearer YOUR_API_KEY.
2. Set your client base URL to https://relaydance.com/v1.
3. Send POST /v1/video/generations with model, prompt, and seconds.
4. Add your start and end images to metadata.content[], plus ratio, resolution, and optional generate_audio.
5. Reference the frames in the prompt as @image1 and @image2.
6. Poll GET /v1/video/generations/{task_id} until status is succeeded or failed, or set metadata.callback_url to receive a webhook POST.

Model rates for frame controlled clips

Per-model rates are pay-as-you-go and vary by resolution, so pick a model that fits your output size. Seedance 2.0 at 720p costs about $0.190 per second, and 1080p costs about $0.470 per second. Seedance Fast costs about $0.152 per second, and Seedance native 4K costs about $4.90 per 5-second clip. Live rates are published at relaydance.com/models.

Billing and failed requests

Billing is per generated video, and errors carry no charge. 「Failed or errored requests are never billed」, according to the official relaydance.com docs, so a rejected or errored frame job does not add cost. This lets you retry a first and last frame task after adjusting the images or prompt without paying for the failed attempts. Payments are accepted by USDT and Stripe card. The service provides direct OpenAI-compatible access to Seedance with no BytePlus enterprise account or KYC required. For image tasks, gpt-image-2 bills only input (image output is free), image-to-image starts from about CNY 0.035, and 4K and 1K output cost the same.

FAQ

How many reference images can I use per request? Up to 9 reference images, along with up to 3 reference videos and 3 audio tracks per request.

How do I check when my video is ready? Poll GET /v1/video/generations/{task_id} until status is succeeded or failed, or set metadata.callback_url for a webhook POST.

Am I charged if the request fails? No. According to the official relaydance.com docs, failed or errored requests are never billed.