☁️BizyAir Seedance 1.5 Pro Text To Video
ByteDance's closed video model, on-demand
- video
- bizyair_model_name
Seedance is ByteDance's video line, and like its image sibling Seedream, it's API-only by policy, not by accident - ByteDance keeps every finished product closed while giving away the components underneath it (Depth Anything, PuLID, and friends). This node is how you reach doubao-seedance-1-5-pro-251215 from inside a ComfyUI graph: type a prompt, get a video back, pay BizyAir credits per call.
What it does
Straightforward text-to-video: a prompt, plus a small pile of video-specific knobs. resolution (480p/720p/1080p) and ratio (16:9, 4:3, 1:1, 3:4, 9:16, 21:9, or adaptive) work together to set the actual output dimensions - the tooltip is explicit that the two jointly decide frame size, so check ByteDance's own docs if you need an exact pixel count for a specific combination. duration runs 2–12 seconds. fps shows up as a dropdown but only has one real option, 24 - it's effectively locked, not a setting you're choosing.
camerafixed asks the platform to try adding a "keep the camera still" instruction to your prompt behind the scenes - the tooltip is upfront that this isn't guaranteed to work, just attempted.
The one genuinely new capability over the 1.0 line is generate_audio. Seedance 1.5 Pro can generate synced dialogue, sound effects, and background music that match the visual content, driven straight from your prompt - the tooltip's own example is a man telling a woman "you can't point at the moon with your finger anymore," with the advice to wrap dialogue in quotes for cleaner audio results. It's flagged as affecting billing, so it's not a free toggle.
seed runs -1 (random) to a fixed max for reproducibility.
Output
A single video, plus the passthrough bizyair_model_name label.
Installing it
ComfyUI Manager: search BizyAir, install, restart. Or:
cd ComfyUI/custom_nodes && git clone https://github.com/siliconflow/BizyAir.git
Restart after. Nothing to download locally - the generation happens on ByteDance's infrastructure via BizyAir's proxy, not your GPU. You do need a BizyAir account and an API key configured before this node will do anything (the pack walks you through it on first use).
Where people get burned
Set expectations on quality before you burn credits. Community reports on Seedance 1.5's actual video output have been mixed - complaints about motion coherence on anything beyond simple actions show up in the wild, and this is a fast-moving line where ByteDance ships new versions (2.0, 2.5) within months. If a generation looks off, it may be the model, not your prompt.
fps isn't actually adjustable despite looking like a dropdown - don't spend time hunting for a way to change it beyond 24.
Audio costs extra. If your bill looks higher than expected, check whether generate_audio got left on from a previous run.
It's a paid API, not a free local render. BizyAir turned on billing in April 2025; queuing this node repeatedly while iterating on a prompt adds up fast compared to a local video model you already own the GPU time for.
Inputs (9)
| Name | Type | Default | Description |
|---|---|---|---|
| prompt | STRING | — | |
| model | COMBO | doubao-seedance-1-5-pro-251215 | 1 options: doubao-seedance-1-5-pro-251215 |
| resolution | COMBO | 1080p | 分辨率+比例共同决定视频尺寸,具体尺寸请参考官方文档说明 |
| ratio | COMBO | adaptive | 比例+分辨率共同决定视频尺寸,具体尺寸请参考官方文档说明 |
| duration | INT | 122–12 | — |
| fps | COMBO | 24 | 帧率固定24 |
| camerafixed | BOOLEAN | false | 平台可能在提示词中追加固定摄像机指令(效果不保证) |
| generate_audio | BOOLEAN | false | 影响计费。控制生成的视频是否包含与画面同步的声音,Seedance 1.5 pro 能够基于文本提示词与视觉内容,自动生成与之匹配的人声、音效及背景音乐。建议将对话部分置于双引号内,以优化音频生成效果。例如:男人叫住女人说:“你记住,以后不可以用手指指月亮。” |
| seed | INT | -1-1–2147483647 | — |
Outputs (2)
| Name | Type | Description |
|---|---|---|
| video | VIDEO | — |
| bizyair_model_name | {"doubao-seedance-1-5-pro-251215": "doubao-seedance-1-5-pro"} | — |