comfyui-lumi-tools
A ComfyUI custom node pack by Lumi.
Nodes (16)
The Gemini image settings that don't live in a sampler — aspect, resolution, temperature
The fast lane to Gemini imagen — direct Google API, no middleman
Generate images with Gemini imagen right inside ComfyUI
An LLM that rewrites your prompt before it ever touches the sampler
Load Image that remembers the order — the reference-image chain for Gemini edits
Got a NOISE object? Here's the integer hiding inside it
Gemini imagen through OpenRouter — one key, one bill, no Google account juggling
One provider node to point every LLM call at OpenRouter
Getting Rid of the Fourth Channel Without Breaking Your Graph
A Save Image that refuses to hand you a 30MB PNG
A seed node with the control widget actually built in
See what that text node actually output — a debug window that stays in the save
Shuffle your tag order so a fixed seed stops meaning fixed output
The multiline text box you were about to build yourself
Wildcards, the ComfyUI way — random picks from your own word banks
Slam a prefix and suffix onto any prompt without retyping it
Lumi Tools
A collection of utility nodes for ComfyUI.
Installation
- Clone or symlink this repository into your ComfyUI
custom_nodesfolder - Install dependencies:
uv sync - Restart ComfyUI
Development Setup
uv sync --dev
Nodes
All nodes appear under Lumi/ in the node menu.
Prompt & Text Nodes
Lumi Shuffle Prompt
Shuffles tokens in a prompt string. Useful for randomizing tag order.
- Strips newlines and commas
- Splits by spaces, shuffles, and rejoins
- Seeded for reproducibility
Lumi Wildcard Processor
Processes wildcard prompts using dynamicprompts.
Lumi Wrap Text
Wraps text with optional prefix and suffix strings.
Lumi Text Input
Provides an arbitrary multiline text input.
Lumi Show Text
Displays text output for debugging.
LLM Nodes
Lumi OpenRouter Provider
Provides OpenRouter API configuration for LLM inference. Requires OPENROUTER_API_KEY environment variable.
Lumi LLM Prompt Processor
Processes prompts using LLM inference via OpenRouter. Useful for prompt enhancement and rewriting.
Image Generation Nodes
Lumi Gemini Imagen Config
Configuration node for Gemini image generation models. Sets aspect ratio (default: 16:9), image size (default: 2K), temperature, and other generation parameters.
Lumi Google Imagen Provider
Direct Google AI Studio API provider for image generation. Uses GOOGLE_API_KEY environment variable. Much faster than OpenRouter (~4x).
Available models:
gemini-3-pro-image-preview(default, supports up to 4K)gemini-3.1-flash-image-preview(supports up to 4K)gemini-2.5-flash-image(1K only)
Lumi OpenRouter Imagen Provider
OpenRouter API provider for Gemini image generation. Uses OPENROUTER_API_KEY environment variable.
Lumi LLM Imagen Processor
Generates images using configured Gemini imagen providers. Connects to provider and config nodes.
Utility Nodes
Lumi Noise To Seed
Extracts an integer seed from a NOISE object for nodes that expect an INT seed.
Lumi Seed
Outputs a seed value with control_after_generate support (randomize, increment, decrement, fixed).
Lumi Save Image
Saves images as PNG with workflow metadata. If PNG exceeds 4MB, also saves a JPG version with configurable quality (default 100). Directory and filename are separate widgets, each supporting ComfyUI token replacements. Directory defaults to %year%-%month%-%day%.
Configuring Wildcard Paths
Configure wildcard paths in ComfyUI/extra_model_paths.yaml:
my_wildcards:
wildcards: D:/dev/wildcards
Alternatively:
- Set the
LUMI_WILDCARDS_PATHenvironment variable - Place wildcards in
{ComfyUI}/wildcards
License
GPL-3.0