Troubleshooting

If requests fail, run okou doctor check-connector --env-name SEEDANCE_TOKEN or okou doctor check-connector --url https://ark.ap-southeast.bytepluses.com/api/v3/contents/generations/tasks --method POST

Video preview

Before generating a video, prepare a few keyframes matching the user's subject, style, and aspect ratio. Reuse suitable supplied or already approved images.
Show the actual images through accessible links, briefly describe the intended motion, and ask the user to confirm. End the turn and wait; do not start a video job before confirmation. Revise the preview if requested.
After confirmation, generate the video. Use the approved images as first/last frames or references when supported by the selected model; otherwise follow the approved visual direction in the prompt. Reuse existing confirmation for an unchanged preview.

Keep the preview message short: the images, a brief motion description, and one confirmation question.

How to Use

Video generation is asynchronous: submit a task → get a task ID → poll until succeeded → use the video URL (valid for 24 hours).

1. Text-to-Video (Standard — 1080p)

Write to /tmp/seedance_request.json: