ByteDance Seedance 2.0 First Last Frame (Official Edition)
First/last-frame video at full Seedance 2.0 quality, in the cloud
- first_frame_url
- last_frame_url
- api_config
- video
- request_id
- urls
First-last-frame generation at full Seedance 2.0 quality, run in the cloud through BizyAirPlus's Official Edition route. You supply a first image and an optional last image; ByteDance's Seedance 2.0 animates the space between them and returns a video. No local model, no GPU.
This is the combination that makes the technique actually useful: FLF gives you control over the composition, and full Seedance 2.0 gives you the coherence and 1080p output that the Mini tier skips. If you've already got an image you love and want it in motion, this node is the cleanest path in ComfyUI. It's the node you reach for at the end of a project, when the shots matter - not the one for the cheap draft iterations.
How it works
An output node in BizyAirPlus's cloud layer: with BizyAirPlus ON in the action bar, it uploads your first frame (and optional last frame) plus prompt and settings to Seedance's service, and the finished clip returns to your workflow. Official Edition means the vendor's direct API route rather than BizyAir's channel - the model is identical, and which route you want comes down to your account's billing and availability.
The inputs that matter
- first_frame_url - required, typed
IMAGE. The shot you're animating from. - last_frame_url - optional, up to one image (about 30 MB). Provide it to end the video on a specific frame.
- prompt - the motion and action between the frames.
- resolution -
1080pdefault; the full model earns it. - duration - COMBO, default 10 seconds.
- aspect_ratio -
16:9default. - seed - 0 to 2147483647 for reproducibility.
- generate_audio - on by default.
Note this node's frame inputs are single images, not lists - one first frame, one optional last frame. That's a deliberate difference from the reference-to-video node, which accepts whole batches.
The outputs
video for preview/saving, urls for the direct link, request_id for support traceability - the standard BizyTRD trio.
Installing it
In BizyAirPlus. ComfyUI Manager → "BizyAirPlus", or:
cd ComfyUI/custom_nodes
git clone https://github.com/siliconflow/BizyAirPlus.git
cd BizyAirPlus
python -m pip install -r requirements.txt
Restart, toggle ON, enter the key from bizyair.ai. It lives at ~/.BizyAirPlus/apikey.ini, or set BIZYAIR_API_KEY. Actual dependencies are the bizyair-cloudberry and bizytrd pip packages; no model downloads, no VRAM pressure - your machine is just a remote control.
Gotchas
First-run errors are the button or the key, and the ComfyUI console shows the real detail - check it before re-queueing. On the reputation front: if you've seen "is BizyAir malware?" threads, that was an old Manager auto-install mixup, fixed; SiliconFlow is a legit vendor. For the node itself, the frame-selection rule matters more than any setting: your two bookend images should match in subject, framing and lighting, or the model spends its quality budget reconciling them instead of making the motion good.
Inputs (10)
| Name | Type | Default | Description |
|---|---|---|---|
| first_frame_url | IMAGE | First image | |
| prompt | STRING | Prompt words for video generation | |
| resolution | COMBO | 1080p | Video resolution |
| duration | COMBO | 10 | Video length in seconds |
| last_frame_urlopt | IMAGE | Supports up to 1 image, each 30 MB | |
| 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 | — |