One prompt, many formats: batch producing social media variants of the same video

To make several social media versions of one video with RelayDance, send the same prompt to POST /v1/video/generations once per format. In each request, change only metadata.ratio, metadata.resolution and seconds. Each request returns a task_id. You can poll that task_id or receive the result by webhook. Clips can run up to 15 seconds. Seedance 2.0 costs about $0.190 per second at 720p, and failed requests are not billed.

How does one prompt become several formats?

You submit one video task for each format you need, with the same prompt and reference media every time. RelayDance is OpenAI-compatible, so you can run the batch from your existing client. According to the relaydance.com docs: “Set base_url to https://relaydance.com/v1 and keep your OpenAI SDK”. Each task goes to POST /v1/video/generations with model, prompt, seconds and a metadata object. For each version, set metadata.ratio to the platform's aspect ratio, metadata.resolution to the quality tier and seconds to a clip length of up to 15 seconds. Supported ratio values are listed in the docs. Every call needs the header Authorization: Bearer YOUR_API_KEY, using a key created at relaydance.com/console. A minimal request body:

{"model": "doubao-seedance-2-5-1080p", "prompt": "Product spin on a desk, @image1", "seconds": 10, "metadata": {"ratio": "RATIO", "resolution": "RESOLUTION", "generate_audio": true, "callback_url": "https://your-server.example/hook"}}

What does each variant cost?

Each version is billed as its own generated video, at the per-second rate for the model and resolution you choose. The rates below come from relaydance.com/models. The 10-second column is the rate multiplied by 10, so it is approximate. For example, three 10-second versions on Seedance 2.0 cost about $5.70 at 720p and about $14.10 at 1080p. If you produce many drafts, you can use Seedance Fast at about $0.152 per second and then render the approved cuts at a higher tier. You can pay with USDT or a Stripe card. Current rates are on the models page.

How do you collect batch results?

You can get results by polling each task or by receiving webhooks. Webhooks mean you don't have to keep polling. Every request returns a task_id when you submit it. To poll, call GET /v1/video/generations/{task_id} until the status is succeeded or failed. A succeeded result contains the video url. To use webhooks, set metadata.callback_url and RelayDance POSTs the final state to your server, so one endpoint can receive every version in the batch. Store each task_id with its target platform so you can match each result to the right format. Retrying a failed version costs nothing for the failed attempt: “Failed or errored requests are never billed”, according to the relaydance.com docs. Billing is pay-as-you-go per generated video.

How do reference media keep variants consistent?

Reference media keep the versions looking the same because every request can carry the same assets in metadata.content[]. You cite those images in the prompt as @image1 to @imageN, so the prompt text stays the same across formats. Each request accepts up to 9 reference images, 3 reference videos and 3 audio tracks. For still images such as thumbnails or covers, use POST /v1/images/generations. The relaydance.com models page also lists gpt-image-2 billing: you pay for input and image output is free. Image-to-image costs from about CNY 0.035, and 4K output costs the same as 1K output. You can use Seedance through RelayDance without a BytePlus enterprise account or KYC.

FAQ: batch variants on RelayDance

These are short answers to common questions about producing several formats from one prompt.

Can I reuse the same prompt for every format? Yes. Keep the prompt and metadata.content[] the same and change just metadata.ratio, metadata.resolution and seconds in each request.

How long can each variant be? Each clip can run up to 15 seconds. You set the length with the seconds field.

Am I charged if one variant in the batch fails? No. Failed or errored requests are never billed, and billing is pay-as-you-go per generated video. Current per-model rates are at relaydance.com/models.