ComfyUI-lhyNodes
Some useful nodes for ComfyUI.
Nodes (49)
A Python eval in your graph — powerful, and the pack will warn you about it
Pixelate a mask into chunky blocks in one node
One dropdown controls every checkpoint loader in your graph
Pick your text encoder once, use it everywhere
A clean dropdown for your GGUF text encoders only
The text encoder that stops re-doing work it already did
Pick the vision model behind your control and identity nodes
Prompts with comments, the way code should have them
One dropdown to choose which ControlNet your whole graph uses
Roll a random prompt fragment from a CSV, seeded so you can reproduce it
Same random pick, but now the count rolls too
Set CUDA_VISIBLE_DEVICES mid-run, without restarting ComfyUI
See your mask. Actually see it, in color, on the image
Redraw WanAnimate pose skeletons without burning your VRAM
One JSON config spawns a whole control panel of native widgets
GrowMask, but it doesn't make you wait
Unpack an image batch into an actual list of images
Staple a folder of images into one fixed-size batch tensor
Shave a fixed number of pixels off every side
Composite one image onto another with real blend modes
Scale to a target megapixel count, rounded to a size the model likes
Watch your video render — without the full VAE decode
A Floating TV for Whatever ComfyUI Is Drawing Right Now
Load one image at a time from a folder — without reconnecting wires
Load Image from Zip
LoraName Does Exactly One Thing
Turn a Painted Mask Into SAM2 Click Points, Automatically
Mask to Coordinates V2
MiniMax H3 image-to-video in two passes
Morse code, in seven languages, without leaving the graph
When 'None' Is the Right Answer
Your Workflow's Control Panel, Unpacked Into Real Sockets
Fix the Black Frames Killing Your Wan Animate Motion Transfer
See the text that's actually flowing without breaking the wire
Make ComfyUI Wait Before It Runs the Next Thing
RemoteCLIP Client
Your Second Machine, Lending Out Its Text Encoder
Make an HTTP Call Mid-Workflow and Keep Your Data Moving
Batching Your Outputs Into a Downloadable Archive
When Your Output Is a Thousand Text Files, Zip Them First
String Format, the Quiet Workhorse
String Formatting With a Kill Switch Per Slot
A Clean Dropdown for Diffusion Models, and a String Out
The Name Dropdown for Your Quantized UNets
Name the Upscaler, Wire the Name Into Your Filenames
One Less Typed String in Your Workflow
Wan Animate too long to fit? Let the math pick your chunk size
Wan Animate masking is fiddly — this node is the cleanup step
Wan Animate dropped a frame's face? This node holds the last good one
Installation
You can:
⭐️ Search for lhyNodes in ComfyUI-Manager and install it.
or
-
Install it manually in command line:
cd ComfyUI/custom_nodes git clone https://github.com/lihaoyun6/ComfyUI-lhyNodes.git python -m pip install -r ComfyUI-lhyNodes/requirements.txt
Nodes
Load Image Batch
Upload multiple images directly through the browser and generate a image batch or image list.
Image Batch to Images
Convert the
image_batchobject into a batch of images.
Image Batch to Image List
Convert the
image_batchobject into a list of images.
Load Image from ZIP
Read images from uploaded zip file and generate
image_batchobject.
Save Image as ZIP
Save the image (with optional additional text) as a ZIP file and provide it for download.
String Format
Format and output the text containing placeholders
{}according to the variables.
String Format (Advanced)
Format and output the text containing placeholders
{}according to the variables.
And you can enable or disable any variable.
CSV Random Picker
Randomly select elements from a CSV string by seed.
CSV Random Picker (Advanced)
Randomly select elements from a CSV string by seed with more options.
Queue Handler
Use triggers to control when any node executes.
None
Do nothing, just output
None
Set CUDA Device
Modify the value of the environment variable
CUDA_VISIBLE_DEVICESduring runtime.
Image Overlay
Overlay an image onto another image.
Grow Mask
Growing input masks extremely fast.
Blockify Mask / Draw Mask On Image
Just like their names.
WanAnimate Mask Preprocessor
All-in-one Wan Animate mask preprocessor.
WanAnimate Face Reformer
Remove faceless frames from the facial frame sequence and restore its coherence.
WanAnimate Pose Reformer
Use it to remove all-black frames from any frame sequence and restore its coherence.
WanAnimate Best Frame Window
Calculate the optimal frame window size based on the total frames count.
Mask to Coordinates
Generate SAM conditions by drawing points in ComfyUI's Mask Editor.
Mask to Coordinates V2
Generate positive and negative SAM conditions using mask brush and colored brush.
Credits
- ComfyUI @comfyanonymous