Vidu Q3 Pro Image To Video (Channel Edition)

Image-to-video is where Vidu Q3 Pro earns its reputation. People building AI-animated series keep reporting that Q3 nails subtle eye and mouth movements that open-weight models only approximate - and this Channel (Base) edition gives you that quality with the fewest knobs and typically the friendliest price.

What this node does

BizyTRD_Vidu_Q3_Pro_Base_I2V_API takes one image as the first frame and animates it forward according to your prompt. It's the node you'll reach for most: consistent characters, product shots, any clip that starts from a known still. The Channel edition skips seed control and the paid recommendation-prompt feature, so it's the lean route to the same model.

Cloud execution, as with every BizyTRD node: flip BizyAirPlus ON, and Vidu Q3 Pro runs on BizyAir's servers while ComfyUI uploads your image and streams back the result. Zero local inference.

The inputs that matter

• image - the first-frame still. This is the whole anchor; its aspect ratio and content drive the clip.
• prompt - 1–4000 characters of motion ("she glances at the camera and smiles").
• resolution - 720P by default; aspect_ratio defaults to 16:9 and inherits from the image when not set.
• duration - 1–16 seconds.
• style - general default.
• audio - on by default: synchronized dialogue and sound effects.
• movement_amplitude - auto default; nudge it for more or less motion.
• Bonus: first_frame_image / last_frame_image are optional here, and feeding both effectively upgrades this node into first-last-frame mode. Handy if you realize mid-shot that you want a defined ending - you don't need to swap to a different node.

Plus the shared api_config (endpoint/key override) and skip_error (placeholder instead of abort).

Outputs

• video - the generated clip, ready for preview or save.
• request_id - cloud task ID.
• urls - hosted URLs.

How to install

cd /path/to/ComfyUI/custom_nodes
git clone https://github.com/siliconflow/BizyAirPlus.git
cd BizyAirPlus
python -m pip install -r requirements.txt

Restart ComfyUI (or install "BizyAirPlus" via ComfyUI Manager). Deps: bizyair-cloudberry and bizytrd; no model files. First run: account at bizyair.ai → API key → enter when prompted (or Settings > BizyAirPlus > API Key; key at ~/.BizyAirPlus/apikey.ini).

Where people get burned

Two things. First, Q3 is a strong enough model that your input image quality becomes the bottleneck - a blurry or badly composed first frame gives a blurry video, and no prompt fixes that. Second, remember this is the Channel edition: you don't get the seed control of the Official node, so you can't reproduce a specific take deterministically. If you need seed-locked iteration, that's the Official I2V node's job.

And the pack note: BizyAir's "malware" reputation is a leftover from a 2025 ComfyUI Manager auto-install bug, not reality. Legit SiliconFlow pack; remove it with pip uninstall bizyair-cloudberry bizytrd if you ever want it gone.