Vidu 1.5

Vidu 1.5 is Shengshu AI's closed video model, and in this pack it's the node you reach for when you want a couple of production-y extras that local video models usually leave to post-processing: a native background-music toggle and an explicit movement-amplitude control. Those two knobs are what separate it from the generic I2V nodes sitting next to it in the menu.

The whole thing runs on Runware's cloud, so there's no checkpoint and no VRAM math - the model ID (vidu:1@5) never even appears in your workflow. That's the trade baked into this entire pack: you get a frontier closed model without the hardware, and you pay per generation.

What you set

No required inputs at all - you can literally queue a text-only generation. The ones that matter:

• positivePrompt - what you want on screen.
• referenceImages and frameImages - both IMAGE sockets. referenceImages is your style/subject anchor; frameImages is for first-frame conditioning (the "animate this exact image" case).
• width / height - defaults 1024×1024. Unlike some video nodes with locked sizes, Vidu lets you pick.
• duration - 4 or 8 seconds.
• providerSettings.vidu.movementAmplitude - auto, small, medium, large. This is the dial people actually fiddle with: big amplitude = dynamic but riskier motion, small = stable but could feel static. auto is a fine default until a clip feels wrong.
• providerSettings.vidu.bgm - flip this on and the model generates background music to go with the clip. It's a gimmick at first, then you realize it saves you a whole audio pass.
• providerSettings.vidu.style - general or anime.
• seed - exposed, so takes are reproducible if you keep the seed and settings fixed.

Output is video (VIDEO socket) → video preview/save node. outputFormat MP4/WEBM/MOV and outputQuality handle encoding.

Installing

This node is part of the Runware/ComfyUI-Runware pack; install the pack and it shows up under Runware/Video/vidu.

ComfyUI Manager: search Runware, install, restart.

Manual:

cd ComfyUI/custom_nodes
git clone https://github.com/Runware/ComfyUI-Runware
pip install -r ComfyUI-Runware/requirements.txt

No weights to fetch - deps are runware-sdk, pillow, soundfile. You need a Runware API key from runware.ai/api-keys (Settings panel or RUNWARE_API_KEY env var).

Where people get burned

Vidu's reputation in the community is decent but it's not the default the way Wan once was, and 1.5 is a paid API model - so cost discipline matters. Each numberResults is a separate generation, and 8-second clips cost more than 4-second ones. Practical tip: lock seed, set movementAmplitude to small for your first pass, and only flip on bgm when you've confirmed the visuals - otherwise you're paying for music on clips you'll regenerate anyway.