ByteDance Seedance 2.0 Reference To Video (Official Edition)
Images, clips and audio as refs
- images
- videos
- audios
- api_config
- video
- request_id
- urls
The reference-to-video version of full Seedance 2.0 - the model that animates new video while respecting images, existing clips, even audio you feed it, running in the cloud through BizyAirPlus's Official Edition route. This is the highest-rung Seedance node in the pack, and it's aimed at one job: making video that stays faithful to what you already have.
Where T2V is a lottery and first-last-frame pins just two points, R2V gives the model a whole brief. Feed it a character sheet and it keeps the character. Feed it a product photo and it keeps the product. Feed it a reference clip and it echoes the motion. If you've been fighting consistency battles with local video models, this is the relief valve - and because it's the full model rather than Mini, the results hold up at 1080p.
How it works
Output node in the BizyAirPlus cloud layer. Flip BizyAirPlus ON in the action bar and the node forwards your references plus prompt to ByteDance's Seedance service, which renders the clip and returns it to your workflow. The reference sockets - images, videos, and audios - are all COMFY_AUTOGROW_V3 ports that accept lists, so you can feed several items of each type. This official node has the full reference set, audio included, which not every R2V node in the pack does.
The inputs that matter
- images / videos / audios - the references. Images carry subject and style, videos carry motion and pacing, audio can drive sync or mood. More isn't always better; a few coherent refs beat a pile of conflicting ones.
- prompt - the action. The references set the look; the prompt sets what happens.
- duration - COMBO, default 10 seconds.
- resolution -
1080pdefault. - aspect_ratio -
16:9default. - seed - 0 to 2147483647.
- generate_audio - on by default.
There's no watermark or web-search toggle here - the official full model keeps the option surface tight, which is honestly fine.
The outputs
video for preview/save, urls for the direct link, request_id for support. Standard trio.
Installing it
Part of BizyAirPlus. Manager (search "BizyAirPlus") or:
cd ComfyUI/custom_nodes
git clone https://github.com/siliconflow/BizyAirPlus.git
cd BizyAirPlus
python -m pip install -r requirements.txt
Restart, turn BizyAirPlus ON, key from bizyair.ai. Stored at ~/.BizyAirPlus/apikey.ini or BIZYAIR_API_KEY. The real machinery is the bizytrd and bizyair-cloudberry pip packages; nothing heavy, no model files.
Gotchas
Key missing or button OFF covers most first-run failures - read the ComfyUI console for the actual error. And the standing advice: keep your references consistent, or the model spends its quality budget negotiating between them. On billing, full R2V with multiple references is the premium end of the Seedance line, so iterate on the Mini nodes and save this one for shots that matter. If the Official route is down or pricier on your account, the Channel Edition node calls the same model.
Inputs (11)
| Name | Type | Default | Description |
|---|---|---|---|
| prompt | STRING | Text length limit: 1-20480 video generation prompts | |
| resolution | COMBO | 1080p | Video resolution |
| duration | COMBO | 10 | Video length in seconds |
| imagesopt | COMFY_AUTOGROW_V3 | — | |
| videosopt | COMFY_AUTOGROW_V3 | — | |
| audiosopt | COMFY_AUTOGROW_V3 | — | |
| aspect_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 |
| 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 | — |