ArtificialGuyBR

Home / Blog / Training LoRAs

Doodle Style LoRA for FLUX.2 Klein 9B

LoRA for FLUX.2-klein-9B that generates doodle-style illustrations with high detail and consistency.

2 sources cited Training LoRAs

On this page

  1. What the Model Does
  2. How It Was Trained
  3. Trigger Words
  4. How to Use It
  5. Recommended Settings
  6. License
  7. Common Errors and Fixes
  8. FAQ
  9. Sources

Doodle Style LoRA for FLUX.2 Klein 9B

TL;DR:

What the Model Does

This LoRA fine-tunes the FLUX.2-klein-9B base model to produce doodle style imagery. It was trained on a dataset of hand-drawn doodle artwork, enabling the diffusion model to output illustrated, sketch-like results while maintaining the underlying quality of the base model.

How It Was Trained

The LoRA was trained with GPU time donated by Redmond.AI. It uses low-rank adaptation weights distributed in safetensors format, keeping the file size minimal while preserving stylistic fidelity across diverse prompt themes.

Trigger Words

The model's author specifies two trigger tokens:

Including either token in your prompt activates the doodle style rendering.

How to Use It

Load the .safetensors file into a LoRA node and connect it to your FLUX.2-klein-9B model pipeline. Apply the LoRA at a strength between 0.6 and 0.8 for balanced doodle style.

A1111 / WebUI

Place the safetensors file in the models/Lora/ directory and select it in the LoRA panel. Use the same trigger words in your prompt.

Diffusers (HuggingFace)

Use the diffusers library with the PEFT adapter loaded on top of the FLUX.2-klein-9B pipeline.

Parameter | Value

|---|---

Base Model | black-forest-labs/FLUX.2-klein-9B LoRA Strength | 0.6 - 0.8 File Format | Safetensors Recommended UI | ComfyUI

License

This model is licensed under the Apache License 2.0, permitting personal and commercial use.

Common Errors and Fixes

Error | Cause | Fix | Source

|---|---|---|---

Doodle style not appearing | Trigger words missing or LoRA strength too low | Add "Doodle" or "DoodleRedm" to prompt, increase strength to 0.7 | HF Card Model not loading | Wrong base model or corrupted weights | Use FLUX.2-klein-9B as base, re-download safetensors file | HF Card

FAQ

Q: Does this work with SDXL or other base models? A: No. This LoRA was specifically trained for FLUX.2-klein-9B.

Q: Can I use it commercially? :A: Yes, under the Apache 2.0 license.

Q: Do I need both trigger words? A: Either "Doodle" or "DoodleRedm" will work.

Sources