Zhenzhen Upscaler 视频超分
A one-click video super-resolution pass up to 4k, billed per clip
- input_video
- api_config
- video
- video_url
- task_id
- response
The last mile of AI video is resolution, and Zhenzhen_Upscaler_Video is the pack's answer to "my clip is 720p and I want it sharper." It's a hosted video upscaler (zhenzhen-upscaler via the T8mars/ComfyUI_Seedance reseller API): feed it a local VIDEO or a public MP4 URL, pick a target - 720p, 1080p, 2k, or 4k - and it returns an upscaled clip as a decoded ComfyUI VIDEO. Fewer knobs than anything else in the pack.
The honest context, from the KB's upscaling essay: the local community has genuinely strong free video upscalers now - SeedVR2 is ByteDance's Apache-2.0 one-step video restoration model and is the current crowd favorite, and the KB notes it beats FlashVSR on heavy upscales. So this node is a convenience play, not a capability you can't get elsewhere: if you're already paying for the pack and just need a clip finished, it's one node and no VRAM. If you upscale constantly or care about control, the local path is free and better documented. The one thing the hosted route does offer is zero local GPU cost - an upscale that might churn your card for a while happens server-side here.
How it works
Connect input_video (a local ComfyUI VIDEO, auto-uploaded) or fill in video_url with a public MP4 direct link - the tooltip is explicit that you leave video_url empty when input_video is connected. Then pick resolution and run. The node submits to the /v1/videos endpoint with the zhenzhen-upscaler model, polls, downloads, and validates the result as a real MP4 before handing it back. The pack's download machinery is genuinely defensive here - it writes to .part files, checks the MP4 header, and falls back across download paths - so an upscale that finishes on the server rarely dies on the download.
Inputs and outputs that matter
video_url- optional public MP4 URL; leave empty ifinput_videois connected.resolution-720p,1080p,2k, or4k(default1080p).input_video- optional local video to upload for upscaling.
Outputs: video, video_url, task_id, response.
Install
One node in T8mars/ComfyUI_Seedance:
# ComfyUI Manager: search "ComfyUI Seedance", or:
comfy node install seedance
Or clone https://github.com/T8mars/ComfyUI_Seedance.git into ComfyUI/custom_nodes, pip install -r requirements.txt, restart, and configure a key from api.seedance.nz/console via the Seedance API Config node or SEEDANCE_API_KEY.
Where people get burned
Two things. First, the source quality sets a floor: a badly compressed 480p clip upscaled to 4k is still a badly compressed clip, just bigger - the model can't invent detail that was never there, so match the target to what the source can support. Second, the cost: video upscaling is metered per clip and 4k is the top tier, and the download is capped by default at 8 GiB (SEEDANCE_VIDEO_MAX_MIB), so a long 4k output can legitimately run into that ceiling - shorten the clip or adjust the env var. If you're on a low-res generation from the pack's own video nodes, this node is a natural finishing step on the same bill; if your source is already decent, spend your money elsewhere and upscale locally.
Inputs (6)
| Name | Type | Default | Description |
|---|---|---|---|
| video_url | STRING | Optional public MP4 URL. Leave empty when connecting input_video. | 可选公网 MP4 直链;连接 input_video 时可留空。 | |
| resolution | COMBO | 1080p | Target resolution: 720p, 1080p, 2k, or 4k. | 目标分辨率:720p、1080p、2k 或 4k。 |
| input_videoopt | VIDEO | Optional local ComfyUI video to upload for upscaling. | 可选本地 ComfyUI 视频,节点会先上传再超分。 | |
| api_configopt | SEEDANCE_CONFIG | Connect Seedance API Config; otherwise SEEDANCE_API_KEY is used. | |
| skip_erroropt | BOOLEAN | false | On failure return a placeholder error video instead of stopping the workflow. | 失败时输出占位错误视频。 |
| seedopt | INT | 00–18446744073709550000 | ComfyUI cache seed. Fixed reuses the cached result while all other inputs stay unchanged; randomize/increment/decrement starts a new execution. This value is not sent to models without documented seed support. | ComfyUI 缓存种子;Fixed 在其他输入不变时复用缓存,随机、递增或递减会触发新任务。未声明支持 seed 的模型不会收到此参数。 |
Outputs (4)
| Name | Type | Description |
|---|---|---|
| video | VIDEO | — |
| video_url | STRING | — |
| task_id | STRING | — |
| response | STRING | — |