ByteDance Seedance 2.0 Fast Reference To Video (Channel Edition)
Iterate fast, render once
- images
- videos
- audios
- api_config
- video
- request_id
- urls
Seedance 2.0's reference-to-video mode is the flexible one - images, videos, and audio as steering input, plus a text prompt, and the model assembles a clip that honors all of it. The Fast tier exists for people who want to do that a lot, cheaply and quickly, because reference-driven video is inherently iterative: you swap a reference, tweak a prompt, and want to see what broke before you commit. This Channel Edition node is that draft machine. The standard Seedance R2V node is the finish line you move to once the concept is locked.
How it works
Cloud execution through BizyAir, same as the rest of the pack: reference media uploads to api.bizyair.ai, the request hits seedance-2-0-fast-base/reference-to-video, and the clip comes back as a native ComfyUI VIDEO. No local model, no VRAM, metered billing on BizyAir credits.
Inputs that matter
- prompt - the text steer, 1–20480 characters. Describe motion and intent, not a freeze-frame.
- images / videos / audios - the autogrowing reference slots. For a fast tier, lean on fewer references: one strong image (look/character) plus a text prompt drafts faster and with fewer conflicting signals than a five-reference pile.
- resolution - defaults to
2k; native 480p/720p/Native1080p, upscaled 1080p/2K/4K on top of a 720p base. - duration - seconds, default
5, range 4–15 - the shortest default of the Seedance R2V nodes, very much "quick draft" energy. - ratio - aspect ratio, default
16:9. - seed - integer, default
0. - generate_audio - default on; off for silent drafts.
- return_last_frame - default off; on to chain the final frame forward.
- api_config / skip_error - per-request key/base override and placeholder-on-error.
Outputs: video (into preview/save), request_id, urls.
Install and first run
ComfyUI Manager → "BizyAirPlus", or:
cd /path/to/ComfyUI/custom_nodes
git clone https://github.com/siliconflow/BizyAirPlus.git
cd BizyAirPlus
python -m pip install -r requirements.txt
Restart, register at bizyair.ai, set the key in Settings > BizyAirPlus > API Key (or BIZYAIR_API_KEY), and flip the BizyAirPlus action-bar toggle ON.
Gotchas
The eternal setup trio: missing key, toggle off, wrong Python (python -m pip show bizyair-cloudberry). For this tier, the discipline tip is to treat drafts as throwaway: at the Fast price and speed, it's cheaper to try ten wild prompt variations here than two careful ones on the full model - but match reference aspect ratios and lighting so your comparisons are actually apples-to-apples. When a draft finally clicks, re-run that exact input on the standard Seedance R2V node for the quality pass.
Inputs (12)
| Name | Type | Default | Description |
|---|---|---|---|
| prompt | STRING | Text length limit: 1-20480 video generation prompts | |
| resolution | COMBO | 2k | Video resolution. The resolution is divided into the resolution of the model native output (480p, 720p, Native1080p) and the resolution of super-resolution amplification based on 720p native generation (1080p, 2K, 4K) . |
| duration | INT | 54–15 | Video length in seconds |
| imagesopt | COMFY_AUTOGROW_V3 | — | |
| videosopt | COMFY_AUTOGROW_V3 | — | |
| audiosopt | COMFY_AUTOGROW_V3 | — | |
| ratioopt | COMBO | 16:9 | Aspect ratio of video |
| seedopt | INT | 0-1–2147483647 | Input range value: 0-2147483647 seed integer to control randomness of generated content. |
| generate_audioopt | BOOLEAN | true | Whether to generate video and audio |
| return_last_frameopt | BOOLEAN | false | Whether to return the last frame of the video |
| api_configopt | BIZYAIR_OPENAPI_CONFIG | 单次请求覆盖base_url和api_key | |
| skip_erroropt | BOOLEAN | false | 开启后遇到错误不中断工作流,输出对应类型的错误占位符 |
Outputs (3)
| Name | Type | Description |
|---|---|---|
| video | VIDEO | — |
| request_id | STRING | — |
| urls | STRING | — |