ComfyUI workflows and other tools

ComfyUI Workflows with Flux1 and Flux2, dev, klein and other useful tools.

Flux2 workflows

In flux2 workflows, do not forget to change the QwenVL

1. flux-2-klein-9b-fp8 uses qwen_3_8b_fp8
2. flux-2-klein-4b uses qwen 4b
3. flux2 dev uses mistral 3 small.

LTX 2.3 image/text workflow

LTX 2.3 image/text to vid with 16GB VRAM.

Runs on NVIDIA 4080. I start comfyui with:

python3.13 -m venv comfy-env && source comfy-env/bin/activate && comfy launch -- --disable-cuda-malloc --use-sage-attention --reserve-vram 4 --cache-none

You find the models for 16 GB from https://ltxworkflow.com/models and https://huggingface.co/Kijai/LTX2.3_comfy

Other tools.

LoRA's metadata

read_lora_metadata.py is not a Workflow or Node!, it's Python script for reading LoRa metadata.

""" Requirements: pip install safetensors

---

Reads and displays metadata from a LoRA .safetensors file. Extracts:

• base model/architecture,
• trigger words,
• training info,
• tensor shapes,
• and any other embedded metadata.

Usage: python read_lora_metadata.py path/to/your/lora.safetensors

"""