durationinteger · optionalLength of the generated media.
minimum: 4 · maximum: 30
Generate video with duration, resolution and native audio controls.
by ByteDanceReady to run$0.0150 per 1k tokens
“Turn this burger photo into a 4-second video, with the camera arcing slowly around it in soft light.”
Your agent works out the exact inputs, such as IDs, dates and settings, from what you ask.
{
"tool_id": "bytedance.seedance-2-5-image",
"input": {
"prompt": "the burger remains centered while the camera makes a slow studio arc and soft light moves across the sesame bun",
"duration": 4,
"image_url": "https://storage.googleapis.com/falserverless/example_outputs/flux-2-flash-edit-input.png",
"resolution": "480p",
"generate_audio": false
}
}{
"note": "video returned as an artifact",
"video": {
"audio": false,
"content_type": "video/mp4",
"duration_seconds": 4
}
}Copy the prompt and paste it into your agent, then tell it what you need.
Read https://scrollport.com/start. Use my existing Scrollport connection, or help me connect if needed. Help me use Seedance 2.5 Image-to-Video (tool ID: bytedance.seedance-2-5-image). Ask me for the details you need, then show me the result.
Provider access is included. If your agent isn't connected to Scrollport yet, it will help you set that up.
Scrollport has no free tier: it costs $0.0150 per 1k tokens, paid per use from a prepaid wallet with no subscription. Connecting your first agent adds $1 of trial credit, so you can try it before adding funds.
No. Your agent runs it through your Scrollport connection (MCP, the Scrollport CLI or the HTTP API), and Scrollport handles access to ByteDance, so you don't sign up with ByteDance or manage an API key.
Create a 4–30 second Seedance 2.5 video at 480p or 720p using the prompt and references shown in the input schema. Native audio is included. Reserve the maximum output-token charge, then pay only for the generated output tokens after the video has been preserved as a Scrollport artifact. Failed requests are uncharged. The same job resumes after interruption. The model rejects reference images containing real human faces. Video references are excluded.
durationinteger · optionalLength of the generated media.
minimum: 4 · maximum: 30
end_image_urlstring · optionalformat: uri
generate_audioboolean · optionalWhether to generate an audio track with the video.
image_urlstring · requiredPublic URL of the source image.
format: uri
promptstring · requiredInstruction describing the requested generated output.
minLength: 1 · maxLength: 5000
resolutionstring · optionalOutput resolution of the generated media.
enum: 480p, 720p
notestringNote about how the result was produced or limited.
videoobjectGenerated video file.
invalid_input: image_url must be a public JPEG, PNG or WebP URL and duration must be 4–30 secondsexample: {"prompt":"the burger remains centered while the camera makes a slow studio arc and soft light moves across the sesame bun","image_url":"https://storage.googleapis.com/falserverless/example_outputs/flux-2-flash-edit-input.png","resolution":"480p","duration":4,"generate_audio":true}