Seed Volcengine T2V Config

Seedance is ByteDance's video model, and like Seedream it's API-only - there's no local checkpoint to download, no ComfyUI-native wrapper to babysit. SeedVolcengineT2VConfig is the node that turns a text prompt into a Seedance request through the Volcengine ARK endpoint. It's also the one video config in the pack that defaults to generating audio, which immediately sets it apart from the silent clips everywhere else.

Config node rules apply: output is a MEDIAHUB_VIDEO_REQUEST that flows into MediaHubSubmitVideo → MediaHubResolveVideo → SaveVideo. The ARK key is the same one the Seedream image nodes use.

Inputs that matter

• model - five Seedance versions, again date-stamped: doubao-seedance-2-0-260128 and its -fast-260128 sibling, then the older 1-5-pro-251215, 1-0-pro-250528, and 1-0-pro-fast-251015. The -fast variants are your draft loop; the 2.0 is the quality pick.
• resolution - 480p, 720p, or 1080p. This is the only video provider here offering a 480p tier, which is genuinely handy for cheap iteration.
• ratio - seven choices including 16:9, 9:16, 1:1, the 4:3/3:4 pair, 21:9, and a adaptive option that lets the provider decide based on your prompt's implied framing.
• duration - seconds, default 5, range up to 15 (-1 means "provider's choice"). Most providers cap at 10; Seedance lets you push a bit further.
• generate_audio - on by default. The model tries to produce matching sound. If you want silent footage (or you're going to add your own audio later), switch it off - otherwise you're paying for audio synthesis you'll throw away.
• watermark - off by default.
• seed - -1 = random; non-negative for reproducibility.

Output: request → MediaHubSubmitVideo.

Why you'd reach for it

Audio-plus-video in one API call is the headline feature, and the 480p tier makes it the cheapest way in this pack to test whether a video idea works before spending on a full-res render. If you're building quick storyboard clips or mood-reel drafts, this is the node.

Installing it

cd ComfyUI/custom_nodes
git clone https://github.com/vantang/ComfyUI-MediaHub.git
/path/to/ComfyUI/python -m pip install -r ComfyUI-MediaHub/requirements.txt

Restart ComfyUI; node under MediaHub/Provider/Seed Volcengine/Video. Only requests/Pillow/numpy, and an ARK API key from Volcengine.

Common gotchas

Forget to check generate_audio and you'll wonder why every clip has noise you didn't ask for - it defaults to on. And remember the async pipeline costs you queue time: ResolveVideo polls up to ten minutes, and a 15-second 1080p Seedance job with audio can genuinely sit in the provider queue that long. A timeout isn't automatically a failure - check the ARK console for the task before re-submitting, or you'll pay twice for the same clip.