☁️BizyAir Nodes
[a/BizyAir](https://github.com/siliconflow/BizyAir) Comfy Nodes that can run in any environment.
Nodes (198)
BizyAir Apply InstantID — ComfyUI Node Guide
BizyAir Apply InstantID Advanced — ComfyUI Node Guide
BizyAir Apply PuLID Flux — ComfyUI Node Guide
BizyAir BAE Normal Map — ComfyUI Node Guide
BizyAir BasicGuider — ComfyUI Node Guide
BizyAir BasicScheduler — ComfyUI Node Guide
BizyAir Binary Lines — ComfyUI Node Guide
BizyAir Canny Edge — ComfyUI Node Guide
Cloud CFG guidance for ComfyUI's custom sampler pipeline
The entry point to every cloud BizyAir workflow
Picking a text encoder for cloud text-to-image chains
CLIP skip, running on BizyAir's cloud CLIP
BizyAir CLIP Text Encode (Prompt) — ComfyUI Node Guide
BizyAir CLIPTextEncodeFlux — ComfyUI Node Guide
The one BizyAir node that plays nice with a normal graph
The vision encoder IPAdapter and Kolors nodes need upstream
Prompt straight to image with Zhipu's open text-to-image model
Turning an image into a T2I-Adapter color map
Blend two prompts by weight, not by words
Run two prompts in parallel, not blended
Stitch two encoded prompts into one longer sequence
Pin a prompt to a rectangle of the canvas
Resolution-independent regional prompting
Regional prompting with an actual shape, not a rectangle
Make a prompt active for only part of the sample
An empty conditioning for when the graph demands one anyway
The legacy single-conditioning apply node
The one to actually use, with start/end-step control
The DiT-era ControlNet apply node
Mask-aware ControlNet inpainting for Flux/SD3-class models
Pulling a ControlNet from the cloud catalog
The fixed-list ControlNet loader, and what's actually on it
The depth ControlNet preprocessor nothing has displaced
Sneak more detail into any sampler for free
Detail Daemon for SamplerCustomAdvanced graphs
Matting-quality edges for a rough mask
Turn a grayscale mask into a per-pixel denoise dial
Tell an advanced sampler not to add fresh noise
BizyAir Download File — ComfyUI Node Guide
Load the two text encoders SDXL, SD3 and Flux actually need
The pose preprocessor that actually gets hands right
HED edges thinned down into scribble-style conditioning
Reverse a sigma schedule for unsampling and noise tricks
BFL's newest Flux tier, run on SiliconFlow's GPUs
The number that replaces CFG on Flux
BFL's instruction editor, no Dev weights or license fight required
A fast, cloud-hosted distilled generator
OpenAI's image model without an OpenAI account
Hand GPT Image 1 a photo and an instruction
The loader behind text-prompted object detection
Describe the object, get a mask
MiniMax's video model, closed-source and paid
BizyAir Hailuo2.3 Text To Video — ComfyUI Node Guide
The forgiving ControlNet edge detector
Turn one image into a 3D mesh with Hunyuan3D 2.1
BizyAir Hy3DExportMesh — ComfyUI Node Guide
BizyAir Load Trellis (BizyAir_IF_TrellisCheckpointLoader) — ComfyUI Node
BizyAir Trellis Predict (BizyAir_IF_TrellisImageTo3D) — ComfyUI Node
BizyAir Image Encode (BizyAir_Image_Encode) — ComfyUI Node
BizyAir Upscale Image (using Model) — ComfyUI Node Guide
BizyAir InpaintModelConditioning — ComfyUI Node
BizyAir InpaintModelConditioning_v2 — ComfyUI Node
The plumbing InstantID's IdentityNet needs, cloud-side
BizyAir Load InstantID Model (BizyAir_InstantIDModelLoader) — ComfyUI Node
BizyAir InstructPixToPixConditioning — ComfyUI Node
Cubiq's IP-Adapter node, running cloud-side
BizyAir IPAdapterModelLoader — ComfyUI Node
BizyAir IPAdapter — ComfyUI Node Guide
BizyAir IPAdapterStyleComposition — ComfyUI Node
The one-click setup for IPAdapterAdvanced
BizyAir JanusImageGeneration — ComfyUI Node
BizyAir JanusImageUnderstanding — ComfyUI Node
BizyAir JanusModelLoader — ComfyUI Node
The plain, no-frills version of the LoRA community's favorite captioner
The community's favorite captioner, with the full preset menu
BizyAir Joy Caption3 — ComfyUI Node Guide
BizyAir Kling 2.1 Image To Video — ComfyUI Node Guide
BizyAir Kling 2.1 Text To Video (BizyAir_Kling_2_1_T2V_API) — ComfyUI Node
BizyAir Kling 2.5 Image To Video — ComfyUI Node Guide
BizyAir Kling 2.6 Image To Video — ComfyUI Node Guide
BizyAir Kling 2.6 Text To Video (BizyAir_Kling_2_6_T2V_API) — ComfyUI Node
BizyAir Kling 2 Image To Image (BizyAir_Kling_2_I2I_API) — ComfyUI Node
BizyAir Kling 2 Text To Image (BizyAir_Kling_2_T2I_API) — ComfyUI Node
BizyAir Kling O1 Image To Video (BizyAir_Kling_O1_I2V_API) — ComfyUI Node
BizyAir Kling O1 Text To Video (BizyAir_Kling_O1_T2V_API) — ComfyUI Node
BizyAir Kling O1 Reference Video/Images To Video (BizyAir_Kling_O1_VI2V_REF_API) — ComfyUI Node
BizyAir Kling O1 Video Edit (BizyAir_Kling_O1_VIDEO_EDIT_API) — ComfyUI Node
The standard sampler, run in SiliconFlow's cloud
Multi-pass sampling without the local GPU
Pick a sampler as an object, not a dropdown baked into a mega-node
Load SAM2 Model (Advance) — ComfyUI Node Guide
Box-guided segmentation with real matting, no local model
A solid depth ControlNet preprocessor, minus the local install
Clean line extraction for ControlNet, run in the cloud
Pull a reference image in from the web, no upload dialog
Apply a LoRA to a cloud-hosted model, including your own
The older LoRA node, picking from BizyAir's built-in catalog
Tell the model it has less noise to remove than it actually does
A depth preprocessor that actually knows real-world scale
Surface-orientation maps from a metric-scale depth model
The original ControlNet depth preprocessor, cloud-hosted
The classic normal-map preprocessor, no local model needed
The text encoder Kolors actually needs
Straight-edge detection for architecture and interiors
Dial in Flux's noise schedule for your actual resolution
One shift value, big effect on rectified-flow models
Scale part of the noise schedule directly
One ControlNet loader, more than just Kolors
BizyAir MinusZone - KolorsUNETLoaderV2 — ComfyUI Node Guide
Google's Gemini image model, minus the hunt for an API console
BizyAir NanoBananaPro — ComfyUI Node Guide
Gemini 3 Pro Image in ComfyUI, no Google API key needed
BizyAir Nunchaku FLUX DiT Loader — ComfyUI Node Guide
Stacking a LoRA onto a quantized Flux — and why the name is misleading
Pasting a face's identity onto a quantized Flux model
The one-input node that starts a PuLID chain
Flux's two text encoders, hosted and swappable
BizyAir OneFormer ADE20K Segmentor — ComfyUI Node Guide
Turning a photo into a layout map, not a likeness
Skeletons for ControlNet, run in the cloud
Turning a chatty vision model's answer into a mask SAM2 can use
A node whose own schema won't tell you what it does
The other softedge preprocessor
BizyAir Load Eva Clip (PuLID Flux) — ComfyUI Node Guide
The face-detection half of PuLID
Grabbing ByteDance's identity adapter without the HuggingFace trip
Where your seed actually becomes noise
The whole face-swap stack in one node
The restore-model config for a face swap, not a swapper itself
One settings bundle, reused across every swap in your graph
The same swap node, with settings pulled out
Flux's official style-transfer adapter, with the knobs exposed
Telling an edit model which image to hold onto
A cutout node with the model choice already made
BizyAir Restore Face 🌌 ReActor — ComfyUI Node Guide
BizyAir Load SAM Model — ComfyUI Node Guide
BizyAir SamplerCustomAdvanced — ComfyUI Node Guide
Segment Anything repurposed as a ControlNet input
ComfyUI Node Guide
Animate a photo with ByteDance's video API
BizyAir Seedance 1.0 Pro Text To Video — ComfyUI Node Guide
Animate a photo, now with synced audio
ByteDance's closed video model, on-demand
ComfyUI Node Guide
ByteDance's earlier Seedream release, still live in this pack
ByteDance's newest image model, run from ComfyUI
ComfyUI Node Guide
BizyAir Text Guided SAM — ComfyUI Node Guide
SegmentAnythingUltra V2 (BizyAir_SegmentAnythingUltraV2): ComfyUI Node Guide
Telling a union ControlNet which condition it's looking at
BizyAir Shared Load ControlNet Model — ComfyUI Node Guide
ComfyUI Node Guide
Turn a rough idea into a real prompt, via an LLM call
BizyAir SiliconCloud VLM API — ComfyUI Node Guide
BizyAir Sora2 Image To Video — ComfyUI Node Guide
OpenAI's video model, without an OpenAI account
ComfyUI Node Guide
ComfyUI Node Guide
ComfyUI Node Guide
ComfyUI Node Guide
ComfyUI Node Guide
ComfyUI Node Guide
BizyAir Switch Server Endpoint — ComfyUI Node Guide
ComfyUI Node Guide
ComfyUI Node Guide
BizyAir Third-Party VLM API — ComfyUI Node Guide
ComfyUI Node Guide
ComfyUI Node Guide
BizyAir UltimateSDUpscale — ComfyUI Node Guide
BizyAir Load Diffusion Model — ComfyUI Node Guide
ComfyUI Node Guide
ComfyUI Node Guide
ComfyUI Node Guide
ComfyUI Node Guide
The mask-growing step that keeps seams from showing
ComfyUI Node Guide
Google's video model without a Google API key
Animate from a cast, not a frame
Prompt in, clip out, no source image needed
The one that scores its own background music
Seven reference slots, more than anything else in the pack
The only node in the pack with an anime switch
The loader for real alpha matting, not segmentation
Turn a trimap into a real alpha matte
A scheduler you almost certainly don't need
Wan's native i2v node, running in the cloud
Wan 2.1 14B i2v in one node, no sampler chain needed
Your own Wan LoRA, running in BizyAir's cloud
The small, fast Wan tier, not the flagship
Watch for the URL output, it's not a VIDEO
The Wan you can't run locally, because it isn't local
BizyAir Wan2.5 Text To Video — ComfyUI Node Guide
Closed, longer clips, a shot-type control 2.5 doesn't have
BizyAir Wan2.6 Text To Video — ComfyUI Node Guide
BizyAir Zoe Depth Map — ComfyUI Node Guide
One node, six control types, no local weights
Not a creative node, a transport diagnostic
An SSIM regression gate, for people building pipelines, not art
BizyAir
- [2025/07/02] 🌩️ BizyAir releases BizyBot which helps ComfyUI to edit image by text.
- [2025/06/27] 🌩️ BizyAir supports FLUX Kontext dev model.
- [2025/06/24] 🌩️ BizyAir supports nunchaku FLUX fill model.
- [2025/06/23] 🌩️ BizyAir supports Hunyuan3D 2.1 shape mesh.
- [2025/06/10] 🌩️ BizyAir supports nunchaku FLUX PulID.
- [2025/06/05] 🌩️ BizyAir supports Joycaption beta one (Joycaption3).
- [2025/05/30] 🌩️ BizyAir supports Wan2.1 I2V and Wan LoRAs.
- [2025/04/24] 🌩️ BizyAir introduces new billing functionality.
- [2025/03/12] 🌩️ BizyAir adds BizyBot
- [2025/03/06] 🌩️ BizyAir supports cogview4 6b.
- [2025/02/27] 🌩️ BizyAir supports wan_t2v_1.3b.
- [2025/02/23] 🌩️ BizyAir supports Joycaption.
- [2025/01/28] 🌩️ BizyAir supports Janus Pro.
- [2025/01/24] 🌩️ BizyAir add trellis nodes.
- [2025/01/09] 🌩️ BizyAir supports Shuttle 3.1 Aesthetic. Shuttle 3.1 Aesthetic
- [2024/12/17] 🌩️ BizyAir supports Flux Upscale Model. FLUX Upscale
- [2024/11/27] 🌩️ BizyAir supports Stable Diffusion 3.5 Large ControlNet Canny, Depth, and Blur. ControlNet Canny ControlNet Depth ControlNet Blur
- [2024/11/22] 🌩️ BizyAir supports FLUX Fill, ControlNet and Redux modes. canny depth fill redux
- [2024/11/06] 🌩️ BizyAir PixelWave Flux.1-dev Text to Image node is released. PixelWave Flux.1-dev Text to Image
- [2024/10/30] 🌩️ BizyAir Segment-Anything nodes has been released. text_guided_segment-anything and box_point_guided_segment-anything
- [2024/10/24] 🌩️ BizyAir supports stable Diffusion 3.5. Stable Diffusion 3.5 Text to Image
- [2024/10/21] 🌩️ BizyAir supports VLModels. We add a tutorial for utilizing the VLM node. AI Assistant
- [2024/10/11] 🌩️ BizyAir supports PuLID for Flux. FLUX PuLID workflow
- [2024/09/29] 🌩️ BizyAir has support to share your custom LoRAs. Share Your LoRAs
- [2024/09/19] 🌩️ BizyAir has updated the model upload feature, allowing the upload process to display a progress bar, and the size of the models uploaded is no longer restricted by ComfyUI. (https://siliconflow.github.io/BizyAir/model-host/introduce.html)
- [2024/09/06] 🌩️ BizyAir supports InstantID for SDXL now. SDXL InstantID workflow
- [2024/09/05] 🌩️ BizyAir supports users in running custom LoRA models, including SDXL and Flux LoRA. How to upload and run custom model
- [2024/08/23] 🌩️ BizyAir now support ultimateSDupscale nodes upscale workflow
- [2024/08/14] 🌩️ BizyAir JoyCaption node has been released. Try the example to recreate a image by JoyCaption and Flux, thanks to fancyfeast/joy-caption-pre-alpha
- [2024/08/05] 🌩️ FLUX.1-dev has been supported. FLUX.1-dev Text to Image, FLUX.1-dev Image to Image
- [2024/08/02] 🌩️ FLUX.1-schnell has been supported. FLUX.1-schnell Text to Image
- [2024/08/01] 🌩️ BizyAir MinusZone Kolors nodes have been released(thanks to MinusZoneAI/ComfyUI-Kolors-MZ), and BizyAir now supports over 10 new base models. They all support ControlNet, LoRA, and IPAdapter. BizyAir KSampler
- [2024/07/31] 🌩️ The super-resolution node has been released, capable of enlarging images four times their original size. BizyAir Photorealistic Image Super Resolution
- [2024/07/25] 🌩️ Users can load BizyAir workflow examples directly by clicking the "☁️BizyAir Workflow Examples" button. Get started
- [2024/07/23] 🌩️ BizyAir ChatGLM3 Text Encode node is released.
- [2024/07/16] 🌩️ BizyAir Controlnet Union SDXL 1.0 node is released.
- [2024/07/14] 🌩️ We release Image Caption nodes and add a tutorial for utilizing the LLM node. AI Assistant
- [2024/07/11] 🌩️ We release new nodes that can perform ControlNet preprocessing. BizyAir Controlnet Auxiliary Preprocessors
- [2024/07/11] 🌩️ We release new nodes, featuring removing image background, photo-quality image generation, and animation super-resolution capabilities.
- [2024/07/10] 🌩️ We release nodes that support Kolors and LLM(large language models).
BizyAir is a collection of ComfyUI nodes that help you overcome environmental and hardware limitations, allowing you to more easily generate high-quality content with ComfyUI.
BizyAir now supports many models and nodes. Product introduction
Table of Contents
Features
Run ComfyUI anywhere, anytime, without worrying about the environment or hardware requirements.
Usage
Please see the Install page to set up BizyAir.
Make sure to set your API KEY for the first time use! You can click on the "click to login" link for a quick setup.Register an account
Installation
You can install BizyAir through several methods: using the ComfyUI Manager, the Comfy CLI, downloading the standalone package for Windows, or by cloning the BizyAir repository into the custom_nodes subdirectory of ComfyUI.
Method 1: Install via ComfyUI Manager
Assuming your ComfyUI already has the ComfyUI Manager installed, search for BizyAir. Click "Install" to complete the installation.
Method 2: Install via git clone
You can install BizyAir by downloading the BizyAir repository to the custom_nodes subdirectory of ComfyUI by using git clone.
cd /path/to/ComfyUI/custom_nodes && \
git clone https://github.com/siliconflow/BizyAir.git
Then, restart ComfyUI.
Method 3: Install via Comfy CLI
- Prerequisites
- Ensure
pip install comfy-cliis installed. - Installing ComfyUI
comfy install
- Ensure
To install the BizyAir, use the following command:
comfy node install bizyair
Method 4: Download windows portable ComfyUI
For NA/EU users:
https://github.com/siliconflow/ComfyUI/releases/tag/latest
For CN users:
https://bizy-air.oss-cn-beijing.aliyuncs.com/new_ComfyUI_windows_portable_nvidia_none_or_cpu.7z
Examples
There are some workflow examples in the examples directory.