Nodes/☁️BizyAir Nodes/☁️BizyAir Vidu Q1 Reference Images To Video
ComfyUI Node

☁️BizyAir Vidu Q1 Reference Images To Video

Seven reference slots, more than anything else in the pack

By siliconflow·Created 2 years ago·Updated 11 months ago· 855
☁️BizyAir Vidu Q1 Reference Images To Video
  • ref_image_1
  • ref_image_2
  • ref_image_3
  • ref_image_4
  • ref_image_5
  • ref_image_6
  • ref_image_7
  • video
  • bizyair_model_name
prompt
seed0
duration5
resolution1080p
movement_amplitudeauto
bgmfalse

Every other reference-conditioned video node in this pack tops out at three images. This one takes seven - one required, six optional. If you're juggling multiple characters in a scene, several angles of the same product, or a cast that needs to stay consistent across a shot, this is the node with actual headroom for it, and it's a genuine differentiator worth knowing about before you reach for the Veo equivalent out of habit.

What it does

Same Vidu Q1 model as the plain image-to-video node in this pack, called through your BizyAir account, but conditioned on a set of reference images instead of a literal first frame. Vidu is one of the fully closed Chinese video labs - no local weights exist, and this node is the only way to touch the model from ComfyUI.

Inputs and outputs that matter

  • prompt - text describing the scene and motion.
  • ref_image_1 - required, your primary reference.
  • ref_image_2 through ref_image_7 - all optional, all the same type. Add as many as your scene needs.
  • seed - set it for reproducibility, leave it for variety between runs.
  • duration and resolution - currently fixed to one option each (5 seconds, 1080p), same as the plain i2v sibling.
  • movement_amplitude - auto / small / medium / large, controlling how much the scene actually moves. Worth turning down if a multi-subject composition is coming out chaotic - more references generally means more for the model to reconcile, and less amplitude gives it less room to get it wrong.
  • bgm - auto-generates and mixes a background track, no duration cap, auto-adapted to length. The same feature the plain Vidu i2v node has and nothing else in the pack does.

Output is video plus bizyair_model_name, an internal label for logging rather than something you'd branch on.

How to install it

  • 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 before anything else runs - either the in-app login flow or a direct account at bizyair.siliconflow.cn. This is a metered API node: every request costs credits, whether it's one reference image or seven.

Common issues & troubleshooting

More references made results worse, not better. This is the honest limit of any reference-conditioned generator: it's usually good at capturing the gist of two or three inputs and starts losing fidelity to any individual one as you add more. If a seven-reference generation looks muddled, try trimming to the two or three references that actually matter and see if consistency improves.

One subject dominates and the others barely show up. There's no per-reference weighting exposed in this node's schema - you can't tell it "prioritize ref_image_3." The only lever you have is which images you include and in what order you feed them; reordering is worth trying before concluding the model just can't do what you're asking.

BizyAir shows up uninvited in your node list. An old ComfyUI Manager bug occasionally pulled BizyAir in while resolving a missing node from a workflow that never needed it, which is where the "is this safe" concerns online trace back to. It's a legitimate commercial pack - safe to keep, and just as safe to remove via the Manager if you don't want a cloud-billed dependency sitting around.

Nothing runs. Check the API key and your BizyAir credit balance first - there's no local fallback here.

Category☁️BizyAir/External APIs/Vidu

Inputs (13)

NameTypeDefaultDescription
promptSTRING
ref_image_1IMAGE
seedINT00–2147483647
durationCOMBO51 options: 5
resolutionCOMBO1080p1 options: 1080p
movement_amplitudeCOMBOauto画面中物体的运动幅度。
bgmBOOLEANfalse当设置为 true 时,系统将自动添加合适的 BGM。BGM 无时长限制,系统会自动适配。
ref_image_2optIMAGE
ref_image_3optIMAGE
ref_image_4optIMAGE
ref_image_5optIMAGE
ref_image_6optIMAGE
ref_image_7optIMAGE

Outputs (2)

NameTypeDescription
videoVIDEO
bizyair_model_name{"vidu-q1-reference2video": "vidu-q1"}