☁️BizyAir Vidu Q1 Image To Video

Vidu is a Chinese video-generation service, and in the same conversation where people list off which video labs went fully closed-source, Vidu gets named right alongside Kling and Hailuo - none of them ship weights, all of them are API-only. This node is how you reach Vidu's Q1 model from inside ComfyUI: feed it a starting image and a prompt, BizyAir proxies the call through your account, and a clip comes back. No local model, no VRAM spent, no download.

What makes this one worth knowing about specifically, versus the Veo nodes sitting next to it in the same pack, is the automatic soundtrack option - more on that below.

What it does

Standard image-to-video: you supply a first frame and a text prompt, Vidu animates from there. It's built on the same premise as every API node in this pack - the actual model runs on someone else's GPUs, you're paying per generation through BizyAir credits, and none of your local hardware matters for output quality.

Inputs and outputs that matter

• prompt - text description of the motion, multiline.
• first_frame_image - the image the clip animates from. Required.
• seed - set a specific value to reproduce a result, or leave it and expect variation between runs.
• duration and resolution - both currently locked to a single option (5 seconds, 1080p). They're exposed as dropdowns rather than fixed values, which suggests more choices are coming, but as of this pack version there's nothing to actually pick.
• movement_amplitude - auto, small, medium, or large. Per the node's own tooltip, this controls how much the subjects in frame actually move. If you're getting a video that's basically a still with subtle drift, bump this up; if motion is too chaotic for what you wanted, dial it down.
• bgm - a boolean, and the genuinely distinctive feature here. Set it true and Vidu automatically composes and adds a fitting background track, no duration limit, auto-adapted to the clip length. Nothing else in this pack does this - the Veo nodes hand you back silent video (or whatever audio Google's model bakes in on its own terms), and none of the local Wan pipeline nodes touch audio at all.
• last_frame_image (optional) - bookend the clip with a second image, same idea as Veo's first/last-frame pairing. Handy if you want to control both ends of the motion instead of just extrapolating forward.

Output is video, plus bizyair_model_name, which reports back a fixed internal label - informational, not something to branch logic on.

How to install it

Same pack, same steps as every other BizyAir node:

• ComfyUI Manager - search "BizyAir", install, restart.
• Manual - cd ComfyUI/custom_nodes && git clone https://github.com/siliconflow/BizyAir.git, restart.
• Comfy CLI - comfy node install bizyair.

Set your API key first - the in-app login prompt or bizyair.siliconflow.cn. This node does nothing without one, and every run spends credits.

Common issues & troubleshooting

You expected a resolution or duration choice and only got one option. That's not you doing something wrong - both fields currently ship with exactly one selectable value. There's nothing to configure there yet.

The BGM sounds generic or mismatched. It's auto-composed from the visual content, not from your prompt text - there's no way to steer style or genre through this node's inputs. If you need specific music, it's simpler to mute this and score the clip separately.

BizyAir turned up in your custom nodes and you didn't put it there. A ComfyUI Manager bug used to mis-resolve missing-node references onto BizyAir, which is why there are online threads flagging it as suspicious. It's a legitimate paid pack, not malware, and safe to uninstall through the Manager if you didn't want it.