This LoRA brings Atar from What God Does in a World Without Gods (Kamikatsu) into your generations. I wanted to capture her distinct, adorable appearance—including her signature side ponytail, dark blue dress, and striking yellow eyes—while keeping the model flexible enough for a wide variety of styles and concepts.
📢 Note on Proportions: If you notice the head-to-body ratio looks a bit too large or top-heavy in certain generations, putting chibi and loli into your negative prompt will help normalize her proportions and give you a more standard anime body scale.
⚙️ Technical Details
Base Model: Trained on waiIllustriousSDXL and designed primarily for Illustrious-based checkpoints. Results may vary on other standard SDXL models.
Dataset: Trained on a curated dataset of high-quality images to accurately capture her authentic anime design and distinct features.
Recommended Weight: 0.8–1.0 (Start at 1.0 and adjust down slightly depending on your checkpoint and prompting style.)
🎨 How to Use Core Trigger Words:
At@r, 1girl, white_hair, side_ponytail, yellow_eyes
Recommended Outfit Tags:
dark_blue_dress
Example Prompt:
Anime art, 1girl, At@r, white_hair, side_ponytail, yellow_eyes, dark_blue_dress, looking at viewer, standing, front view, masterpiece, high quality Negative Prompt (If head is too big): chibi, loli
For best results, keep the core character and facial tags in your prompt, especially when experimenting with different outfits, poses, or environments.
📸 Sample Prompt Ideas
Close-up portraits focusing on her vibrant yellow eyes and signature side ponytail.
Full-body character shots showing off her iconic dark blue dress.
Playful action and fantasy scenes inspired by the chaotic world of Kamikatsu.
Alternative outfits, casual modern clothes, or cute streetwear mixes.
Dramatic studio lighting to emphasize her unique look.
The LoRA was trained to capture Atar's exact look while remaining flexible enough for different poses, backgrounds, and creative variations.
📬 Requests & Feedback Feel free to post your generations in the review section! I always enjoy seeing what people create, and it helps me evaluate how the LoRA performs across different checkpoints and prompt styles.
If there is a character you'd like to see trained next, leave a comment and I'll consider it for a future release.
If you find this LoRA useful, please leave a rating and share your results. It helps others discover the model and supports future uploads.
❤️ Support My Work If you enjoy my releases and want to support future models, you can check out my Patreon:
Even a follow, rating, or shared generation helps a lot and keeps future projects going.
Thank you for downloading, and have fun creating with Atar!