排队视频生成
Queue a new video generation request.
Authentication: This endpoint accepts either a Bearer API key or a SIGN-IN-WITH-X header for x402 wallet-based authentication. The legacy X-Sign-In-With-X header is also accepted during migration. When using x402, a 402 Payment Required response indicates insufficient balance and includes top-up instructions.
/video/quote 获取价格估算,然后使用返回的 queue_id 轮询 /video/retrieve 直到完成。
私有模型还会为已完成的视频返回 download_url。这是一个短期分发 URL(下载失败时几次重试是可以的);详情和可选的 DELETE 隐私操作请参阅视频生成指南。
Seedance 2.0 与 2.5
对于seedance-2-0-*-basic 和 seedance-2-5-*-basic 模型(文生视频、图生视频、参考生视频,以及 Seedance 2.0 的 -fast-* 变体),请参阅 Seedance 2.0 与 2.5 指南 了解四工作流模型(Reference / Edit / Extend / Stitch)、与源匹配的 aspect_ratio / duration 值、家族特定的多模态输入限制、公共 API 媒体政策和定价详情。
公共 Seedance 媒体政策
公共 Seedance 模型不使用同意声明流程(consents.seedance / needs_consent)。包含可检测人物的媒体可能在上游被拒绝。要使用完整的 Seedance 功能集,请使用 Venice 应用或 Studio。详情请参阅 Seedance 指南。
视频放大
对于topaz-video-upscale 模型,请使用 upscale_factor(1、2 或 4)代替 resolution,并提供 video_url。时长和 FPS 会从视频文件中自动检测。完整细节和示例请参阅视频放大指南。
授权
Bearer authentication header of the form Bearer <token>, where <token> is your auth token.
请求体
Request body for video generation. Available fields and valid values vary by model.
The model to use for video generation.
"seedance-2-0-text-to-video-basic"
The prompt to use for video generation. Required for most models. The maximum length varies by model (default 2500 characters, up to 20000 for some models).
1 - 20000"Commerce being conducted in the city of Venice, Italy."
The duration of the video to generate. Available options vary by model. For Seedance 2.5 reference-to-video edit jobs, -1 or auto matches output length to the source clip (requires reference_video_urls on queue, or reference_video_total_duration on quote; source must be 4–30s).
1s, 2s, 3s, 4s, 5s, 6s, 7s, 8s, 9s, 10s, 11s, 12s, 13s, 14s, 15s, 16s, 17s, 18s, 19s, 20s, 21s, 22s, 23s, 24s, 25s, 26s, 27s, 28s, 29s, 30s, -1, 1 gen, auto, Auto "10s"
Optional provider-specific consent attestations. Required only for models that return a needs_consent response.
Optional negative prompt. The maximum length varies by model (default 2500 characters, up to 20000 for some models).
20000"low resolution, error, worst quality, low quality, defects"
The aspect ratio of the video. Available options vary by model. Some models do not support aspect_ratio. For Seedance 2.x reference-to-video edit/extend, adaptive or auto matches output aspect ratio to the source clip (requires reference_video_urls on queue, or reference_video_total_duration on quote).
1:1, 2:3, 3:2, 3:4, 4:3, 4:5, 5:4, 9:16, 9:21, 16:9, 21:9, adaptive, auto "16:9"
The resolution of the video. Available options vary by model. Some models do not support resolution. Use upscale_factor for upscale models.
256p, 360p, 480p, 540p, 580p, 720p, 1080p, 1440p, 2160p, 4k, 2x, 4x, 2K, 768P, true_1080p "720p"
For upscale models only. 1 = quality enhancement, 2 = double resolution (default), 4 = quadruple.
1, 2, 4 2
For models which support audio generation and configuration. Defaults to true.
true
For image-to-video models, the reference image. Must be a URL (http/https) or a data URL (data:image/...).
"data:image/png;base64,iVBORw0K..."
For models that support end images or transitions, the end frame image. Must be a URL or data URL.
"data:image/png;base64,iVBORw0K..."
For models that support audio input, background music. Must be a URL or data URL. Supported: WAV, MP3. Max: 30s, 15MB.
"data:audio/mpeg;base64,SUQzBAA..."
For models that support video input (video-to-video, upscale). Must be a URL or data URL. Supported: MP4, MOV, WebM.
"data:video/mp4;base64,AAAAFGZ0eXA..."
For models with reference image support, up to 30 images for character/style consistency. Each must be a URL or data URL.
30For models with reference video support (e.g. Seedance 2.0 R2V), up to 10 reference video URLs (role: "reference_video") used to inherit subject motion, camera movement, and overall style. Per-clip 2–15 s, .mp4 or .mov, ≤50 MB; aggregate duration ≤15 s. Each must be a URL or data URL.
10For models with reference audio support (e.g. Seedance 2.0 R2V), up to 10 reference audio URLs (role: "reference_audio") used as donors for vocal timbre, narration, or sound effects. Per-clip 2–15 s, .wav or .mp3; aggregate duration ≤15 s. Must be paired with at least one reference image or reference video — audio-only Reference workflows are rejected at validation. Each must be a URL or data URL.
10For models with advanced element support (e.g., Kling O3 R2V). Up to 4 elements defining characters/objects. Reference in prompt as @Element1, @Element2, etc.
4For models with advanced element support. Up to 4 scene reference images. Reference in prompt as @Image1, @Image2, etc.
4For keyframe-driven models. Up to 10 keyframe images pinned to frame positions in the generated 24 fps video. Each frame_index must be unique and no greater than duration × 24.
10响应
Video generation request queued successfully
The ID of the model used for video generation.
"video-model-123"
The ID of the video generation request.
"123e4567-e89b-12d3-a456-426614174000"
Pre-signed URL to download the completed video. Only present for VPS-backed models. When provided, the retrieve endpoint returns JSON status only (no video stream). Fetch this URL after status is COMPLETED to get the video/mp4 file. Valid for 24 hours.