Seedance 2.5 Standard 视频生成（6 合 1）

Seedance is the video model this whole pack is named after - ByteDance's flagship closed video generator, the one the local crowd measures open models against and can't run. This node front-ends Seedance 2.5 Standard's six variants: domestic and global T2V, I2V (first/last frame), and Multi - and Multi is where the numbers get silly: up to 30 images, 10 videos and 10 audio clips as reference material, all in one generation, referenced from the prompt as @Image 1, @Video 1, @Audio 1. That's a feature-reel's worth of source material for one render.

Standard T2V/I2V are the everyday modes; Multi is the one that justifies the node's existence.

How it works

The node's front end expands image/video/audio slots only as you connect them, so a T2V job stays compact and a Multi job grows a wall of inputs. For Multi, reference media all gets written into metadata.content; I2V uses image1 as the first frame and image2 as the optional last frame; the prompt can reference materials by number. seconds=-1 lets the model choose duration. Everything runs through the pack's proven pipeline - upload (rate-limited uploads are retried), submit, poll, stream the MP4 down with header verification.

The inputs that matter

• model - seedance-2.5-standard-{t2v,i2v,multi} plus the -global- twins. (ByteDance ships Seedance on both a domestic and an international line; global is the overseas route.)
• prompt - required for T2V and Multi (with @Image 1 / @Video 1 / @Audio 1 references), optional for I2V.
• seconds - 4 to 30, or -1 for the model to choose.
• resolution - 480p, 720p, 1080p, 2k, 4k, or native1080p (the documented native preset).
• ratio - adaptive or one of six fixed aspects.
• image1..image30, video1..video10, audio1..audio10 - the Multi pool (each video/audio clip must be 2–30s; combined reference media can't exceed 30 minutes total - the tooltip enforces this).
• generate_audio - add voice-over/sound effects to the render.

Outputs: video, video_url, task_id, response.

Installing it

ComfyUI Manager → search "ComfyUI Seedance", or:

cd ComfyUI/custom_nodes
git clone https://github.com/T8mars/ComfyUI_Seedance.git
cd ../..
python -m pip install -r custom_nodes/ComfyUI_Seedance/requirements.txt

Restart, key from api.seedance.nz/console, Seedance API Config into api_config.

Common issues

• Multi reference not landing - check the totals: 30 images, 10 videos, 10 audios max, and the 2–30s per-clip / 30-minute combined rule. Over either and the API pushes back.
• native1080p rejected - it's Standard-only and 2.5 doesn't take native4k; the pack validates this before submitting. Stick to the documented list.
• Long Multi uploads feel stuck - 50 references means 50 uploads against a per-token rate limit; the node retries automatically, so the "uploading" phase can genuinely take a while. That's expected, not a hang.
• Cost shock - Seedance is one of the closed models where per-call video pricing is real and adds up; the community's reaction to Seedance-in-ComfyUI was largely about exactly that. Validate on 4s/480p before spending on 30s/4k.

And the standing caveat for the whole pack: your prompts and references leave the machine for a gateway you don't control, behind ByteDance's content policy. For a model you can't download, that's the price of admission - and the reason people still pay it.