ComfyUI Extension: ComfyUI_StarBetaNodes
Welcome to Star Beta Nodes - a collection of experimental custom nodes for ComfyUI designed for beta testing and feedback. These nodes provide enhanced functionality for image processing, video handling, and workflow automation.
Custom Nodes (0)
README
β Star Beta Nodes for ComfyUI
Welcome to Star Beta Nodes β a staging ground for experimental custom nodes for ComfyUI. Most beta nodes have now been released to the main ComfyUI_StarNodes
repository. This beta repo currently contains the nodes listed below.
π Available Nodes
βStarNodes/Image And Latent
-
β Star Qwen Image Ratio (
StarQwenImageRatio
)- Dropdown-driven aspect ratio selector that outputs:
- An empty SD3 latent sized for the selected ratio (dict with
samples
tensor) - Width and height integers
- An empty SD3 latent sized for the selected ratio (dict with
- Supported ratios:
1:1
β 1328 Γ 132816:9
β 1664 Γ 9289:16
β 928 Γ 16644:3
β 1472 Γ 11043:4
β 1104 Γ 14723:2
β 1584 Γ 10562:3
β 1056 Γ 15845:7
β 1120 Γ 15687:5
β 1568 Γ 1120
- Inputs:
batch_size
(INT, default 1): latent batch dimensioncustom_width
,custom_height
(INT): used when selecting "Free Ratio (custom)", both enforced to be divisible by 16
- Dropdown-driven aspect ratio selector that outputs:
-
β Star Qwen / WAN Ratio (
StarQwenWanRatio
)- Select preset sizes for Qwen, WAN HD, or WAN Full HD, or auto-pick the closest ratio from an input image
- Outputs an empty latent plus width/height
- Optional:
image
to enable "use_nearest_image_ratio"
-
β Star Qwen Image Edit Inputs (
StarQwenImageEditInputs
)- Stitch up to 4 input images and prepare Qwen-sized latent
- Stitch rules:
- 1 image: keep size, downscale longest side to 1328 if larger
- 2 images: 1Γ2 concat, cap height to 1328
- 3/4 images: 2Γ2 grid, letterboxed cells, final 1328Γ1328 (white background)
- Resolution dropdown:
- "Use Best Ratio from Image 1" or choose a Qwen preset / Free custom
- Outputs: stitched
IMAGE
,LATENT
,width
,height
-
β Star Apply Overlay (Depth) (
StarApplyOverlayDepth
)- Blends a filtered image over a source image using either a provided mask or a depth/greyscale image converted to a mask
- Options include strength, invert mask, preview, and pixel-based Gaussian blur of the mask
- Outputs: composited
IMAGE
- Inputs (required):
source_image
,filtered_image
,strength
- Inputs (optional):
depth_image
ormask
(one required),invert_mask
,show_preview
,blur_mask_px
π¦ Installation
Method 1: Git Clone (Recommended)
-
Navigate to your ComfyUI custom nodes directory:
cd ComfyUI/custom_nodes
-
Clone this repository:
git clone https://github.com/Starnodes2024/ComfyUI_StarBetaNodes
-
Restart ComfyUI to load the new nodes.
Method 2: Manual Installation
- Download the ZIP file from the repository
- Extract to
ComfyUI/custom_nodes/ComfyUI_StarBetaNodes/
- Restart ComfyUI
π― Usage Examples
Star Qwen Image Ratio
- Select a ratio from the dropdown
- Connect the
latent
output to SD3.5/SD3 workflows expecting a LATENT dict ({"samples": ...}
) - Use the
width
andheight
outputs to configure downstream nodes - Increase
batch_size
to generate batched latents in one step
Star Apply Overlay (Depth)
- Provide
source_image
and afiltered_image
to overlay - Connect either a
mask
or adepth_image
(greyscale or RGB will be converted) - Optionally toggle
invert_mask
, enableshow_preview
to save a small mask preview, and adjustblur_mask_px
for smoother transitions
π Node Categories
All nodes are organized under:
- βStarNodes/Image And Latent β Image/latent utilities
Note: Previous beta nodes (image loaders, video frame tools, panorama saver, etc.) have been migrated to the main repository: Starnodes2024/ComfyUI_StarNodes
.
π Beta Testing & Feedback
As this is a beta testing repository, your feedback is crucial! Please:
- Report any bugs or issues
- Suggest new features or improvements
- Share your use cases and workflow examples
π Requirements
- ComfyUI (latest version recommended)
- Python 3.8+
- PyTorch (bundled with ComfyUI env)
- PIL/Pillow (for preview saving in overlay node)
- Standard ComfyUI dependencies
π€ Contributing
Contributions are welcome! Feel free to:
- Submit bug reports
- Propose new node ideas
- Submit pull requests for improvements
π License
This project is open source. Please check the repository for specific licensing information.
β Happy beta testing! Your feedback helps make these nodes better for everyone.