Image to video request rejected: fixing invalid image URL, size and format errors

When an image to video request is rejected for an invalid image URL, size or format, verify that each reference image is reachable, cite it correctly in the prompt as @image1 to @imageN, and place the media in metadata.content[]. RelayDance allows up to 9 reference images, 3 reference videos and 3 audio tracks per request, with clips up to 15 seconds. Rejected requests are not charged, so you can retry after correcting the input.

Why the image to video request is rejected

A rejection usually means a reference image is unreachable, exceeds the allowed count, or is not cited in the prompt. RelayDance accepts up to 9 reference images, 3 reference videos and 3 audio tracks per request, and each clip can run up to 15 seconds. Reference media are supplied in metadata.content[] and must be referenced in the prompt text as @image1 through @imageN. If a URL returns an error or a required index is missing, the task fails validation. According to the documentation, 「失败或报错的请求一律不计费」(据 relaydance.com 官方文档), which means you can correct the input and resubmit without cost. See https://relaydance.com/docs for the full task schema.

How to fix an invalid image URL, size or format

Fix a rejected request by validating the media, matching prompt references, and staying within the documented limits.

1. Confirm each image URL is publicly reachable and returns the file, not an error page.
2. Add every asset to metadata.content[] in order.
3. Cite each asset in the prompt using @image1 to @imageN.
4. Keep totals within 9 reference images, 3 reference videos and 3 audio tracks.
5. Ensure any clip runs no longer than 15 seconds.
6. Resubmit with POST /v1/video/generations, then poll GET /v1/video/generations/{task_id} until status is succeeded or failed.

Because failed or errored requests are never billed, iterating on the input carries no per-request cost.

Endpoint and authentication for image to video tasks

Submit image to video tasks to the OpenAI-compatible endpoint using a Bearer key. Send POST /v1/video/generations with model, prompt, seconds, and metadata (including ratio, resolution, generate_audio, callback_url, and content[]). Authenticate with Authorization: Bearer YOUR_API_KEY, using a key created in the console. The base URL is https://relaydance.com/v1. According to the documentation, 「将 base_url 改为 https://relaydance.com/v1 并保留 OpenAI SDK 即可调用」(据 relaydance.com/docs 官方文档), so existing OpenAI SDK code can reach the service after a single base URL change. You can either poll the task ID or set metadata.callback_url to receive the final state by webhook. Full request details are at https://relaydance.com/docs.

Pricing after a corrected request succeeds

Once a request passes validation and generates a video, billing is pay-as-you-go by output, while rejected attempts remain free. The table below lists published per-model rates.

For image generation, gpt-image-2 charges only for input: image output is free, image to image starts at 約 ¥0.035, and 4K and 1K outputs share the same price. Payment is by USDT or Stripe card. Live rates are at https://relaydance.com/models.

FAQ

Am I charged when an image to video request is rejected? No. Failed or errored requests are never billed, so you can correct the image and retry at no per-request cost.

How many reference images can I attach? Up to 9 reference images, plus 3 reference videos and 3 audio tracks per request, with clips up to 15 seconds.

Do I need to rewrite my OpenAI code? No. Change base_url to https://relaydance.com/v1, keep the OpenAI SDK, and authenticate with Authorization: Bearer YOUR_API_KEY.