EvoLink Seedance 2.0 Text-to-Video (Official)
ByteDance's video workhorse, up to 4K
- video
- task_id
- status_info
- result_urls
- response_json
Seedance is ByteDance's video model, and ByteDance plays a specific game: the flagship generation products stay API-only, forever, while the components underneath go open. You'll never download Seedance - the whole Seedream/Seedance line is a closed API product by design. So the moment this model appeared in ComfyUI as an API node, the community had two very loud reactions: excitement that it's reachable at all, and a hard "go kick rocks" from the local-only crowd over what a subscription costs. Both reactions are right. EvoLink Seedance 2.0 Text-to-Video (Official) is the door; the meter is the price of the door.
What it does
Prompt to video, 4 to 15 seconds, up to 4K, with synchronized audio generated for free - the generate_audio toggle defaults to on and the tooltip is explicit that it doesn't add to the bill. That's a real differentiator; most video APIs charge extra for sound or don't do it at all.
The inputs that matter:
- prompt (required) - the shot description, Chinese welcome (the tooltip suggests keeping it under ~500 characters for best results).
- duration - 4–15 seconds, default 5, billed by duration.
- quality - 480p / 720p / 1080p / 4k. 480p is the cheap lane; 4k is where the per-render price starts to really bite.
- aspect_ratio -
adaptiveby default (the model picks), or 16:9, 9:16, 1:1, 4:3, 3:4, 21:9. - generate_audio - on by default, costs nothing extra. Turn it off for pure silent drafts if you want.
Outputs: video (downloaded - connect a Save Video node or you'll see nothing), task_id, status_info (elapsed time and credits used), result_urls (24-hour expiry), response_json.
Install and key
Shared EvoLink install. ComfyUI Manager → search EvoLink → install → full restart (close the console window, not just refresh the browser). Manual:
cd ComfyUI/custom_nodes
git clone https://github.com/deeplearning-goethe/comfyui-evolink
Only requests as a dependency; no models, no GPU. Key from evolink.ai/dashboard/keys (sk-), pasted once into api_key - saved locally, blank after, never in a shared workflow JSON.
The honest economics
Seedance via API is the canonical example in this space of "it's extremely expensive" - the community's own math: a small number of videos per year on a $20–30/month subscription. The counterpoint is just as true: that's the real price of these renders; free/cheap was the loss-leader phase. So use this node the way you'd use a paid render farm: it's for the videos you need that you can't get locally, not for spraying drafts. The default 1800-second (30-minute) timeout tells you something about how long a real video job takes - it's a "start it, go make tea, come back" node, not a "waiting for it is the fun part" node. Unchanged params cache and don't re-bill, and failed or moderation-blocked tasks don't charge, which softens the ouch. But the ouch is real: every render is a line item.
Inputs (7)
| Name | Type | Default | Description |
|---|---|---|---|
| prompt | STRING | 镜头/画面描述,支持中文(建议 500 字内)。 | |
| durationopt | INT | 54–15 | 视频时长(秒),4~15 任意整数,按时长计费 |
| qualityopt | COMBO | 720p | 分辨率 |
| aspect_ratioopt | COMBO | adaptive | 画幅;adaptive 由模型自动选择 |
| generate_audioopt | BOOLEAN | true | 生成同步音频(不额外收费) |
| api_keyopt | STRING | EvoLink API Key(sk- 开头)。首次填写后自动保存到本机配置,之后可留空。分享工作流前请清空此框。获取:evolink.ai/dashboard/keys | |
| timeout_secondsopt | INT | 180060–3600 | 最长等待时间;生成失败或审核拦截的任务不扣费 |
Outputs (5)
| Name | Type | Description |
|---|---|---|
| video | VIDEO | 生成结果(已下载落地,可直连保存节点) |
| task_id | STRING | EvoLink 任务 ID,可到 evolink.ai/zh/dashboard/logs 查询 |
| status_info | STRING | 人读状态摘要(模型/用时/消耗 credits) |
| result_urls | STRING | 结果原始链接,每行一个(24 小时过期) |
| response_json | STRING | 平台 GET /v1/tasks/{id} 的完整 JSON 响应(status/usage/results 等),供下游节点解析 |