Nodes/comfyui-byokey/BYOKey Seedance 2.0 Text to Video
ComfyUI Node

BYOKey Seedance 2.0 Text to Video

Seedance 2.0 text-to-video on your own ByteDance key

By MeteorAndy·Created 4 months ago·Updated 3 months ago· 0
BYOKey Seedance 2.0 Text to Video
    • video
    api_key
    base_urlhttps://ark.ap-southeast.bytepluses.com
    modelSeedance 2.0
    prompt
    resolution720p
    ratio16:9
    duration7
    generate_audiotrue
    seed0
    watermarkfalse
    poll_interval9
    max_wait600

    This is the plain text-to-video half of the Seedance 2.0 pair. No image in, no bookend control - just a prompt, a resolution, a duration, and ByteDance's strong general-purpose video model doing the rest. If your image-to-video sibling is the fancy one, this is the workhorse: character-led scenes, motion, camera moves, all from a paragraph of text.

    It's also the node that sits at the center of the API-video pricing conversation in the KB. Seedance 2.0 arriving in ComfyUI as an API call drew the community's canonical cost objection - a "really small number of videos per year for a monthly subscription of 20 or even 30 bucks." That's not a bug in this node; it's the real price of the model surfacing. You pay ByteDance per clip either way; this node just skips Comfy's prepaid-credit storefront and hits ByteDance's BytePlus ark endpoint with your own key.

    How it works. The familiar submit-then-poll dance: POST a generation job to ark.ap-southeast.bytepluses.com (the default base_url, overridable for a relay or proxy) with your ark key as a Bearer token, poll every poll_interval seconds until the clip is ready, download the mp4, and wrap it in ComfyUI's native VIDEO type. It lands in the same preview/save nodes as any official API video node.

    The inputs that matter:

    • prompt - the whole creative input. Describe subject, motion, camera, and mood; Seedance takes direction well and rewards specific language.
    • model - Seedance 2.0 for full quality, or Seedance 2.0 Fast when you're iterating. Fast only goes up to 720p, which is fine for drafts.
    • resolution - 480p / 720p / 1080p. The cost lever as much as a quality one.
    • ratio - 16:9 default, through 4:3, 1:1, 3:4, 9:16, 21:9, or adaptive.
    • duration - 4 to 15 seconds, and this is where the per-clip bill climbs fastest.
    • generate_audio - on by default, so clips come with an audio track. Turn it off if you're doing sound later; it shaves the job down.
    • watermark - off by default; on adds ByteDance's watermark, usually at a lower rate.
    • seed - non-deterministic regardless of value, per the tooltip. It only gates re-runs.

    poll_interval and max_wait are the babysitters; the 10-minute cap is fine for typical jobs.

    Output: one video (VIDEO) socket.

    Install. Shared pack, one clone:

    cd ComfyUI/custom_nodes
    git clone https://github.com/MeteorAndy/comfyui-byokey.git
    # restart ComfyUI; it's under api/byokey/video
    

    No model downloads, no extra dependencies - ComfyUI's bundled torch, aiohttp, PIL, numpy and av cover it.

    Gotchas. Budget for the cost before you start iterating - this is the model where a long session adds up. And the pack is young with no community footprint yet, so give the source a skim before handing it a real key; that's the standing rule for any API-wrapper node.

    Categoryapi/byokey/video

    Inputs (12)

    NameTypeDefaultDescription
    api_keySTRINGBytePlus ark API key (Bearer token).
    base_urlSTRINGhttps://ark.ap-southeast.bytepluses.comBytePlus ark host. Override for a relay/proxy.
    modelCOMBOSeedance 2.0Seedance 2.0 for max quality; Seedance 2.0 Fast for speed.
    promptSTRINGText prompt for video generation.
    resolutionCOMBO720pOutput resolution. Seedance 2.0 Fast supports 480p/720p only.
    ratioCOMBO16:9Aspect ratio of the output video.
    durationINT74–15Duration of the output video in seconds (4-15).
    generate_audioBOOLEANtrueEnable audio generation for the output video.
    seedoptINT00–2147483647Seed (results are non-deterministic regardless of seed).
    watermarkoptBOOLEANfalseWhether to add a watermark to the video.
    poll_intervaloptINT92–60Seconds between status polls.
    max_waitoptINT60060–3600Max seconds to wait.

    Outputs (1)

    NameTypeDescription
    videoVIDEO