ComfyUI-ReiyaNodes

ComfyUI-ReiyaNodes

Custom ComfyUI nodes for image stitching and variable management.

Installation

1. Navigate to your ComfyUI custom_nodes directory
2. Clone or copy this repository:

   git clone https://github.com/FemReiyaDev/ComfyUI-ReiyaNodes.git
   

3. Restart ComfyUI

Features

• Rei-ImageStitch: Horizontally stitch multiple images together with dynamic input slots
• Rei-GetNode: Custom Get node with for improved workflow organization - Only variation right now being it renames the title to "Get-" instead of "Get_" when you add a variable

Nodes

Rei-ImageStitch

Stitches multiple images together horizontally (side by side).

Inputs

| Name | Type | Default | Description | |------|------|---------|-------------| | num_images | INT | 3 | Number of images to stitch (2-10) | | match_image_size | BOOLEAN | True | Resize images to match the height of the first image | | image_1 | IMAGE | - | First image (required) | | image_2 through image_10 | IMAGE | - | Additional images (optional, appears based on num_images) |

Outputs

| Name | Type | Description | |------|------|-------------| | IMAGE | IMAGE | The horizontally stitched image |

Behavior

• Dynamic Inputs: The node dynamically shows/hides image input slots based on the num_images parameter
• Batch Handling: Automatically matches batch sizes by repeating the last image from smaller batches
• Size Matching: When match_image_size is True, all images are resized to match the height of the first image while preserving aspect ratio
• Channel Matching: Automatically pads images with fewer channels to match the maximum number of channels (e.g., RGB to RGBA)
• Ordering: Images are arranged left-to-right: image_1 | image_2 | image_3 | ...

Technical Details

• Upscale Method: Uses Lanczos resampling for high-quality resizing
• Batch Processing: Supports batched images with different batch sizes
• Channel Support: Handles images with varying channel counts (1-4 channels)
• Device: Operates on the same device (CPU/GPU) as input images

Example Usage

1. Set num_images to the number of images you want to stitch (e.g., 3)
2. Connect your image outputs to image_1, image_2, and image_3
3. Adjust match_image_size to control whether images are resized to match heights
4. The node will output a single image with all input images stitched horizontally

Rei-GetNode

A modified Get node that improves workflow organization and visual clarity.

Features

• Auto-type Detection: Automatically detects and matches the type from corresponding SetNode
• Auto-coloring: Automatically colors the node based on data type (when KJNodes auto-coloring is enabled)
• Virtual Link Visualization: Option to show/hide virtual connections to the corresponding SetNode
• Navigation: Right-click menu option to jump to the corresponding SetNode

Usage

1. Add a Rei-GetNode to your graph
2. Select the variable name from the dropdown (must match a SetNode variable)
3. The node will automatically configure its output type to match the SetNode
4. Connect the right-side output to other nodes

Right-click Menu Options

• Go to setter: Centers the camera on the corresponding SetNode
• Show/Hide connections: Toggle visualization of virtual link to SetNode

Technical Details

• Node Type: Virtual node (does not impact the resulting prompt)
• Compatibility: Works with KJNodes SetNode
• Color Mapping: Supports all standard ComfyUI data type colors (MODEL, LATENT, IMAGE, etc.)

Technical Information

Dependencies

• PyTorch
• ComfyUI
• LiteGraph.js (for UI components)

File Structure

ComfyUI-ReiyaNodes/
├── __init__.py          # Node registration and mappings
├── nodes.py             # Python node implementations
├── pyproject.toml       # Project metadata
├── web/
│   └── js/
│       ├── reiImageStitch.js   # Dynamic input management for ReiImageStitch
│       └── reiGetNode.js       # ReiGetNode UI implementation

Python API

ReiImageStitch Class

class ReiImageStitch:
    MAX_IMAGES = 10
    RETURN_TYPES = ("IMAGE",)
    FUNCTION = "stitch"
    CATEGORY = "ReiyaNodes/image"

Methods:

• INPUT_TYPES(cls): Defines input parameters and dynamic image slots
• stitch(self, num_images, match_image_size, image_1, **kwargs): Main stitching function

Processing Pipeline

1. Image Collection: Gathers all provided images up to num_images
2. Batch Normalization: Matches all images to the largest batch size
3. Size Normalization: Optionally resizes images to match first image height
4. Channel Normalization: Pads images to match maximum channel count
5. Concatenation: Combines images horizontally along dimension 2

JavaScript Extensions

ReiImageStitch Extension

Manages dynamic input slot creation/removal based on num_images widget value.

Key Functions:

• updateInputs(): Adds or removes image input slots
• nodeCreated(): Initializes the node and sets up callbacks
• loadedGraphNode(): Restores correct input count when loading saved graphs

ReiGetNode Extension

Implements the custom Get node.

Key Features:

• Custom node class extending LGraphNode
• Virtual node that doesn't affect prompt generation
• Dynamic type matching with SetNode
• Visual connection rendering

License

MIT License

Credits

• ReiGetNode is based on KJNodes SetGet by diffus3