IMPORTANT: READ BELOW

there are 2 characters and 4 outfits please read below! thank you. To trigger both must use "a01_kur3" prompt!

Trigger words: kur3, long hair, black hair, red eyes,
Clothing 1: red circlet, white leotard, red sailor collar, red bowtie, red brooch, elbow gloves, red gloves, red skirt, white thighhighs,
Clothing 2: white serafuku, pleated skirt, neck ribbon, puffy long sleeves, black thighhighs,Trigger words: a0i, short hair, blue hair, blue eyes, large breasts,
Clothing 1: white leotard, blue sailor collar, elbow gloves, blue gloves, cyan bowtie, blue brooch, blue skirt, white thighhighs,Clothing 2: white serafuku, black pleated skirt, neck ribbon, puffy long sleeves, black pantyhose,

Both Trigger: a0i_kur3
Sample
Positive prompt ✅:

score_9, score_8_up, score_7_up, masterpiece, source_anime BREAK 2girls, a0i_kur3, a01, short hair, blue hair, blue eyes, large breasts, white leotard, blue sailor collar, elbow gloves, blue gloves, cyan bowtie, blue brooch, blue skirt, white thighhighs BREAK kur3, long hair, black hair, red eyes, red circlet, white leotard, red sailor collar, red bowtie, red brooch, elbow gloves, red gloves, red skirt, white thighhighs, looking at viewer, indoors, wooden wall, closed mouth, smile,

Negative prompt ❌:

3d, sketch, score_6, score_5, score_4, worst quality, low quality, floating hair, muscular female, outline, holding, book, paper, holding pencil, bag, school bag, badge, name tag, v, sunset,

Recommended resolution, aspect ratio for sdxl, pony 📷

768 x 1344: Vertical (9:16)

832 x 1216: Portrait (4:5) **

1024 x 1024: Square (1:1)

1182 x 886: Photo (4:3)

1254 x 836: Landscape (3:2)

1365 x 768: Widescreen (16:9)

Work flow 🔧

1. Collect dataset (booru, pixiv, anime screencaps, search,)

2. Clean up dataset using upscalers, inpainting, adetailer, photoshop

3. Tag dataset using booru github

4. Train LoRa in kohya ss locally github

5. Test epochs using xyz grid in controlnet to ensure LoRa is not overfitted/underfitted.

6. Submit model to CivitAI profile

Lora Model trained on Pony Diffusion V6 XL
Example images generated using AutismMix SDXL

Check lora metadata for training parameters

FAQ:

Q: How do I use this?
A: This is a Low-Rank Adaptation (LoRa), it is trained on the ponydiffusion v6 model, and is only compatible with pony checkpoints. You can use this LoRa model in civitai by clicking the "create" button next to the paintbrush icon. If you are generating locally, download the model and place the file in your LoRa folder.

Q: How do I make models like you?
A: I listed my workflow in the description of all my lora models, use it as reference. If you want you can contact me if you need help with your creating your own lora.

Q: Can I request lora model for free?
A: Yes you can message me, but I will only do characters that I like or I feel requires a LoRa. I am not obligated to respond to these requests or complete them.

Q: Are all your models free?
A: Yes all my models are free to use and to download, you can find them here and other websites.

Q: why not use character name as trigger word?
A: I use unique token to prevent other things from bleeding in when using the triggerword.

Q: do you have other account in other websites?
A: yes, tensor, pixai

Q: what external generator do you use?
A: a111 mostly, comfyui for img2vid

Q: What checkpoint you use to generate example images
A: I use autismmix checkpoint.