Hi everyone! 👋

This is my very first LoRA, and to be completely honest, I’m still quite new to the whole AI training process. I initially created this model as a personal tool to help me guide my generations, but since I think it turned out surprisingly okay, I wanted to share it with the you! I would absolutely love to hear your thoughts and get your feedback or suggestions for improvement.

What is this LoRA good for?
It was specifically trained for Krea 2 (Base) to easily inject a volumetric, modern anime style into your generations without needing a huge, complicated prompt. It's heavily focused on enhancing the overall atmosphere and lighting to deliver more vibrant neon colors and richer, deeper blacks.

Note on flexibility: My goal was to keep this LoRA as flexible as possible. It is meant to enhance your creations, not to completely lock them down. Because this, it does not replace a good lighting prompt—you should still describe your desired lighting

Important Tips & Known Issues:

• Dark vs. Bright: Since it was trained on moody nighttime images, the anime style works beautifully in dark environments, cyberpunk alleys, or evening scenes. However, it struggles in bright daylight conditions, where the anime effect will fade significantly.

• Weight Settings (Matter of Taste): It is completely a matter of personal preference! If you want a slightly softer, more balanced look where the colors aren't too aggressive, try a LoRA weight of 0.7. Personally, I prefer running it at 0.7-0.9 !

• Trigger Word: ambianzx

• You can copy my exact prompts by clicking on the preview images!

• If you use a Krea 2 checkpoint that is heavily focused on realism, adding a simple phrase like anime illustration to your prompt will help guide the model.
  

Future Plans:

• I am currently gathering and curating a much larger and more diverse dataset for a potential V2. The goal for the next version will be to make the anime style more robust and flexible, so it carries over cleanly even onto the highly realistic Krea 2 checkpoints that currently struggle with it.

• How you can help: If you test this model, please post your generations. Seeing your images and the prompts you used helps me immensely to understand where the model shines and where it needs tweaking for the upcoming version.

Should you still be reading... pair this with an anime capable checkpoint and enjoy! :)