Upscale task errors on a 480p source video: common seedance-upscale failures and fixes

Upscale task errors on a 480p source video usually come from resolution mismatches, unsupported metadata, or reference media limits, and the fix is to align your request with the documented Seedance parameters. When a request fails for these reasons, you are not charged:

「Failed or errored requests are never billed」据 relaydance.com 官方文档

Why 480p upscale requests fail

Most 480p upscale failures trace to a resolution or metadata field that does not match the target model tier. Seedance 2.0 supports 720p (about $0.190 / second) and 1080p (about $0.470 / second), while native 4K is billed at about $4.90 per 5-second clip, so requesting an output tier your parameters do not describe can return an error. Check that metadata.resolution and metadata.ratio are set, that reference media stay within limits (up to 9 reference images, 3 reference videos, and 3 audio tracks), and that clip length does not exceed 15 seconds. See per-model rates at https://relaydance.com/models.

Steps to diagnose and resubmit

Resolving an errored upscale is a sequence of validation checks before you POST again. Follow these steps:

1. Confirm auth: send Authorization: Bearer YOUR_API_KEY from a key created at https://relaydance.com/console.
2. Set the base URL correctly. 「Set base_url to https://relaydance.com/v1 and keep your OpenAI SDK」据 relaydance.com/docs 官方文档.
3. Submit with POST /v1/video/generations, including model, prompt, seconds, and metadata (ratio, resolution).
4. Verify reference media in metadata.content[] are cited in the prompt as @image1 to @imageN.
5. Poll GET /v1/video/generations/{task_id} until status is succeeded or failed.

Choosing the right output tier

Picking a documented tier for your 480p source avoids parameter errors and clarifies cost. The table below lists current Seedance rates so you can match the request to the intended output:

Billing is pay-as-you-go per generated video, and rates are published at https://relaydance.com/models.

Handling costs during retries

Retrying a failed upscale carries no charge, which lets you iterate on parameters safely. Because failed or errored requests are never billed, you can adjust metadata.resolution, ratio, or reference media and resubmit without incremental cost for the failed attempts. For image workflows, gpt-image-2 output is free and only input is billed, image-to-image starts from about CNY 0.035, and 4K and 1K output cost the same. You can enable webhook mode by setting metadata.callback_url, and the final state is POSTed to your server. Payments are supported via USDT and Stripe card, with no BytePlus enterprise account or KYC required. Full details are in the docs at https://relaydance.com/docs.

FAQ

Am I charged when an upscale task errors? No. Failed or errored requests are never billed, so retries after a fix do not add cost for the failed attempts.

What clip length and reference limits apply? Clips can run up to 15 seconds, with up to 9 reference images, 3 reference videos, and 3 audio tracks per request.

How do I check task status? Poll GET /v1/video/generations/{task_id} until status is succeeded or failed, or use webhook mode by setting metadata.callback_url so the final state is POSTed to your server.