Genius Nodes
Smart and practical custom nodes for ComfyUI.
Genius Nodes for ComfyUI
Smart and practical custom nodes for ComfyUI.
Genius Nodes is a growing collection of compact tools designed to simplify common ComfyUI workflow tasks.
The goal is simple:
Fewer nodes. Less manual calculation. Cleaner workflows.
Included Nodes
Genius Scale & Crop
Genius Scale & Crop prepares an image for a target output resolution in two connected stages.
- Scale the image proportionally so its longest side matches the desired maximum dimension.
- Crop the excess on the shorter side to reach the exact target resolution.
The important part is that the image is not stretched or distorted.
The node automatically calculates the proportional intermediate resolution, so there is no need to manually calculate aspect ratios or intermediate pixel dimensions.
How It Works
The basic idea is simple:
The longest side of the desired output resolution is used as the max_dimension.
This means one dimension of the final output is already established during the first stage.
The other dimension is calculated automatically through proportional scaling.
After that, only the excess on the shorter dimension needs to be cropped.
Portrait Example
Suppose the desired final output is:
480 × 864
The longest side of the desired output is:
864
Set:
max_dimension = 864
The input image is then scaled proportionally until its longest side is exactly 864 pixels.
For example, an input image of:
1024 × 1536
becomes:
576 × 864
The height is now already correct.
The proportional width of 576 pixels was calculated automatically.
There is no need to calculate it manually.
For the final stage, set:
target_width = 480
target_height = 864
crop = center
The 864-pixel height remains the intended final dimension.
Only the excess width needs to be removed:
576 × 864
↓
480 × 864
With a centered crop, the excess width is removed equally from both sides.
The result matches the requested output resolution without stretching the image.
Landscape Example
The same principle works in the opposite direction.
Suppose the desired final output is:
864 × 480
The longest side is again:
864
Set:
max_dimension = 864
After proportional scaling, the intermediate image could for example be:
864 × 576
The width is now already correct.
For the final stage, set:
target_width = 864
target_height = 480
crop = center
The 864-pixel width remains the intended final dimension.
Only the excess height needs to be removed:
864 × 576
↓
864 × 480
The image reaches the final target resolution without distortion.
The Core Idea
The workflow follows this principle:
Desired Output Resolution
↓
Take the Longest Side
↓
Use it as Max Dimension
↓
Proportional Scaling
↓
Automatic Intermediate Resolution
↓
Keep the Long Dimension
↓
Crop the Excess on the Short Dimension
↓
Exact Final Resolution
This avoids having to manually calculate the proportional intermediate size.
For example, if the final video resolution is:
480 × 864
you already know that:
max_dimension = 864
The node calculates the other dimension automatically.
If the result after proportional scaling is:
576 × 864
you immediately know how much image area will be removed when cropping to:
480 × 864
That intermediate result can also be previewed before the final crop.
Why This Is Useful
Without this workflow, preparing an image for another aspect ratio often requires multiple separate nodes or manual calculations.
You may need to:
- determine the correct proportional intermediate resolution
- calculate the aspect ratio manually
- resize the image without distortion
- inspect the result
- crop the remaining excess
- connect multiple scaling and cropping nodes
Genius Scale & Crop combines that process into one compact node.
This helps reduce:
- unnecessary node chains
- manual aspect-ratio calculations
- intermediate resolution calculations
- trial and error
- accidental image distortion
Inputs
image
The input image.
max_dimension
Defines the length of the longest side after the first scaling stage.
For the intended workflow, this should normally be the longest side of the desired final output resolution.
Example:
For:
480 × 864
use:
max_dimension = 864
For:
864 × 480
use:
max_dimension = 864
max_dim_method
Defines the scaling algorithm used during the proportional Max Dimension stage.
Available methods include:
area
nearest-exact
bilinear
bicubic
lanczos
target_width
Defines the final requested image width.
target_height
Defines the final requested image height.
For the intended Scale & Crop workflow, one of the two target dimensions normally remains equal to max_dimension.
The other dimension defines the final cropped size.
For example:
max_dimension = 864
target_width = 480
target_height = 864
or:
max_dimension = 864
target_width = 864
target_height = 480
This allows the same node to work naturally with both portrait and landscape formats.
scale_method
Defines the scaling method used by the final ComfyUI scaling/cropping operation.
Available methods include:
nearest-exact
bilinear
area
bicubic
lanczos
crop
Current options:
center
disabled
center
The image is cropped around its center to reach the requested final resolution.
disabled
Cropping is disabled.
Outputs
max_dim_image
The image after proportional scaling to max_dimension.
This output makes it possible to visually inspect the intermediate result before the final crop.
For example:
Original:
1024 × 1536
Max Dimension Result:
576 × 864
This helps show how much image information is available before cropping.
final_image
The finished image at the requested target resolution.
For example:
480 × 864
This is normally the output connected to the next stage of the ComfyUI workflow.
Why Two Outputs?
The two outputs make the transformation easy to inspect visually.
You can preview:
Original Image
↓
Max Dimension Result
↓
Final Cropped Result
This allows you to check two important things:
- The image quality and available resolution after proportional scaling.
- The final composition after cropping.
This is especially useful when preparing images for resolution-sensitive workflows such as image-to-video generation.
Installation
ComfyUI Manager / Registry
Genius Nodes is available through the Comfy Registry and can be installed through ComfyUI Manager.
Manual Installation
Clone this repository into the ComfyUI custom_nodes directory:
git clone https://github.com/GeniusGQ/ComfyUI-Genius-Nodes.git
Then restart ComfyUI.
The nodes are available under the Genius category.
Philosophy
Genius Nodes are designed around small workflow problems that should have simple solutions.
Each node should aim to:
- solve a real workflow problem
- reduce unnecessary node chains
- reduce manual calculations
- remain understandable
- stay lightweight
- integrate naturally with ComfyUI
- make repetitive workflow tasks easier
More Genius Nodes are planned.
Planned Improvements
Future versions of Genius Scale & Crop may add additional crop positioning controls.
Possible future options include:
center
top
bottom
left
right
This would allow users to control which part of an image is removed when the final crop is applied.
Version 1.0 focuses on a simple and reliable centered Scale & Crop workflow.
License
Genius Nodes are free to use in personal and commercial creative workflows.
Images, videos, workflows, and other outputs created with Genius Nodes may be used commercially.
The Genius Nodes software and source code itself may not be copied, redistributed, republished, rebranded, sold, incorporated into another distributed node pack, or commercially distributed without prior written permission.
ComfyUI workflow files that reference Genius Nodes may be freely shared.
Users receiving those workflows should obtain Genius Nodes through an official distribution channel.
See LICENSE.txt for the complete license terms.
Author
Genius
Comfy Registry Publisher:
@genius
GitHub:
GeniusGQ
Copyright © 2026 Genius / GeniusGQ
All rights reserved.