Hello Friends🍨.

Today I am dropping another Ino model but in some new flavors 🫠. The goal was to give each of her looks over time the individual care they deserve instead of one giant model. This lora focuses on her genin look in the original Naruto series. I'm super excited to feature my first Qwen image model along with illustriousXL_v01 (Best anime model for anime based on sdxl).

Qwen image was Trained on RTX 6000 pro. kohya-ss / musubi-tuner with 60 images.

illustriousXL was trained on a single 3090ti bmaltais / kohya_ss gui with similar data set. The dataset was all drawn by humans and touched with photoshop, no AI.

Qwen tips:

Feel free to experiment, qwen seems pretty flexible and versatile if you prompt it right. My best attempt to balance speed vs quality used the setting below:

• City 96 q8-ggufs for all example images but it works fine at lower quants.

• lightx2v/Qwen-Image-Lightning 8 step lora.

• added film grain.

• added minor sharping.

• For anime looking images add, "anime screenshot" or "anime screencap".

• Seems compatible with other Qwen loras.

Her outfit can be customized pretty well by adding and removing from positive prompt Or give her a new look all together.

iillustriousXL Tips:

Pretty much compatible with most illustriousXL fine tunes. Should be able to use in existing workflows. I still like Prefect illustrious XL