comfyui_vidu_api/特色预设

ViduFeaturedPreset is the "official template" node of this pack - Vidu's curated, purpose-built generation recipes that do one thing well without you having to assemble the pieces by hand. The README's example is 穿搭展示 ("outfit show"), the fashion-catalog shot where a garment is displayed to its best advantage.

Here's the honest part: at the time of writing, the 预设模板 dropdown has exactly one option, outfit_show. That's it. The pack's own future-plans list says "support more featured presets" like it's a roadmap item, not a done deal. So don't come here expecting a catalog of templates - you get one, and it does its one thing competently: feed it two images of the outfit (图像_1, 图像_2) and a prompt, and Vidu produces a video that shows the garment off.

How it works. Same shared machinery as the whole pack - key from api.json, image upload, then a call to /ent/v2/template2video with the template name plus whatever the template needs. For outfit_show that means exactly two images and a prompt - the node raises if you give it fewer than two. The interesting bits are the two widgets you won't find on the other nodes:

• 背景音乐 - a boolean that toggles background music on the generated clip. Default on.
• 额外JSON参数 - an escape hatch. It's a JSON string that gets parsed and merged into the request payload, so when Vidu adds a parameter this pack hasn't built a widget for, you can still push it through without waiting for an update. This is genuinely the most useful thing about the node: the pack is frozen (one commit, July 2025), but this field keeps the node current with the API.

There's also 提示词, 随机种子, plus the standard 输出路径 / 文件名前缀 for where the mp4 lands.

Outputs: video, 封面链接 (cover URL string), 任务ID - the pack-standard trio, with the same "error text lands on the string outputs" behavior when a run fails.

Install - identical to every node in this pack:

cd ComfyUI/custom_nodes
git clone https://github.com/josephxie1/comfyui_vidu_api_node
# restart, then create api.json beside the nodes:
#   { "api_key": "vda_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxx" }

Key from platform.vidu.cn; ComfyUI Manager search comfyui_vidu_api_node also works. The only dependency is requests - no models, no downloads, no VRAM.

Where people get burned. The two-image requirement is the first trip: you wire one image, hit run, and get an error asking for two. Wire both and it works. Otherwise the pack-wide caveats apply - this is a pay-per-call cloud node that blocks your queue while it renders, needs a ComfyUI build new enough for comfy_api, and sends your images to Vidu's servers where their content moderation applies. If you want the fashion-template workflow specifically, this is the node; if you just want a video, the plain ViduImage2Video is probably the better default. ViduFeaturedPreset is the "one trick" node of a small pack - know what the trick is before you build around it.