Nodes/☁️BizyAir Nodes/☁️BizyAir Kling 2.6 Image To Video
ComfyUI Node

☁️BizyAir Kling 2.6 Image To Video

BizyAir Kling 2.6 Image To Video — ComfyUI Node Guide

By siliconflow·Created 2 years ago·Updated 11 months ago· 855
☁️BizyAir Kling 2.6 Image To Video
  • first_frame_image
  • video
  • bizyair_model_name
prompt
soundfalse
model_namekling-v2-6
duration5

What it is

Kling is one of the video models the open-source crowd measures itself against and can never actually run - it's closed, API-only, and the community's own summary of the field is blunt about it: "Vidu, Kling, Hailuo, all closed source." This node takes a still image and turns it into a short video clip through Kling's paid API, routed via BizyAir. You're not running a model here, you're renting one - competitively good output, zero local control.

Reach for it when you need Kling's specific animation quality on a source image and you don't mind paying per generation. The KB notes Kling has historically ranked well against open alternatives on raw quality - one comparison put Wan 2.2 "on par with kling 2.0 master" at full quality - so it's a real benchmark, not just marketing.

How it works

Feed it a starting frame and a prompt describing the motion or change you want, and Kling's servers generate the video from there. It's image-to-video, not text-to-video: the model needs that first frame to anchor composition, subject, and style, and the prompt steers what happens next.

Inputs and outputs that matter

first_frame_image is required and does the heavy lifting - this is your starting point, not a style reference. prompt describes the motion/action. sound is a boolean toggle for whether the output should include generated audio - off by default, so flip it on deliberately if you want sound baked into the clip. model_name is currently locked to kling-v2-6 (the node's own enum only offers one choice). duration is 5 or 10 seconds - no other options - so plan your storyboard around those two lengths.

Output is video for the generated clip, plus bizyair_model_name, an internal tag identifying the underlying model used for billing/logging rather than something you'd normally wire further.

Installing it

Search "BizyAir" in ComfyUI Manager and install, or clone it manually: cd ComfyUI/custom_nodes && git clone https://github.com/siliconflow/BizyAir.git, then restart ComfyUI. Every node in this pack, this one included, needs a BizyAir API key before it'll do anything - register through the login prompt that appears on first use.

Common issues

Cost is the thing that catches people off guard with any of these video-API nodes - this is Kling's commercial rate passed through your BizyAir credits, not a free tier, and video generation is priced well above image generation across this whole category (comparable Kling tiers have been reported north of a dollar for a few seconds at high resolution elsewhere in the ecosystem, so don't assume image-node pricing applies here). If a generation fails outright, check your account balance first - that's a far more common cause than a workflow bug. And because duration only offers 5 or 10 seconds with nothing in between, if you need a longer sequence you're stitching multiple clips together outside ComfyUI, not extending this node's output.

Category☁️BizyAir/External APIs/Kling

Inputs (5)

NameTypeDefaultDescription
first_frame_imageIMAGE首帧图片
promptSTRING
soundBOOLEANfalse是否开启声音
model_nameCOMBOkling-v2-61 options: kling-v2-6
durationCOMBO52 options: 5, 10

Outputs (2)

NameTypeDescription
videoVIDEO
bizyair_model_name{"kling-v2-6": "kling-v2-6"}