HogwartLora

Hogwart uniforms LoRA

<img src="https://huggingface.co/Aotsuyu/HogwartLora/resolve/main/images/0.png" width="800" height="512">

A LoRA for Hogwart uniforms, since Hogwarts Legacy renewed people's interest in the franchise.

What to get

I am including all epochs, but I've personally had the best results with the 2nd to 4th epochs, which I am renaming to hogsks-weak, hogsks-mid and hogsks-hard. Most models seem to have an idea as to how the uniform looks like so they only need a small push - that's why I suggest starting with hogsks-mid. Only go for higher epoch if you're sure that's what you need.

Invoking

I made the token hogsks. I also tried to tag each of the images in the dataset with the proper house, so you might have some results prompting for ravenclaw, gryffindor, slytherin and hufflepuff, but it's not super reliable.<br> For those using the native implementation of LoRA, remember to also activate it!<br> What I propose as a base prompting template:<br> hogsks, hogwarts school uniform, black robe, gray vest, slytherin, green tie<br> Color emblem and color scarf also seem to work reasonably well.

Adjust the house and colors for the desired house, obviously.

This image is made with a very basic prompt: <img src="https://huggingface.co/Aotsuyu/HogwartLora/resolve/main/images/1.png" width="512" height="768">

<details> <summary>Prompt</summary> <pre> best quality, 1girl, Hogsks, hogwarts school uniform, black cape, gray vest, slytherin, green tie, Negative prompt: (low quality, worst quality:1.4), (bad anatomy), by bad-artist, bad-hands-5, bad-image-v2-39000, extra digit, fewer digits, (extra arms:1.2), bad hands, artist name Steps: 25, Sampler: DPM++ 2M Karras, CFG scale: 7.5, Seed: 3963964880, Size: 512x762, Model: anything-v4.5-pruned, Denoising strength: 0.3, Clip skip: 2, ENSD: 31337, AddNet Enabled: True, AddNet Module 1: LoRA, AddNet Model 1: hogsksv2-000003(c945fe615333), AddNet Weight A 1: 0.85, AddNet Weight B 1: 0.85, Hires upscale: 2, Hires steps: 15, Hires upscaler: 4x-AnimeSharp</pre> </details> <br><br>

Previews

All the previews have prompts included, so read that!

The model I used for Hololive can be found here. It's a merge I did.

<img src="https://huggingface.co/Aotsuyu/HogwartLora/resolve/main/images/2.png" width="568" height="768">

<details> <summary>Prompt</summary> <pre> (best quality, 1girl, reimu hakurei, brown hair, red eyes, hogsks, hogwarts school uniform, slytherin, black robe, green scarf, perplexed, (gray vest:1.2), gray skirt, red ribbon, outside, snow, black-green robe Negative prompt: 2girls, (low quality, worst quality:1.4), (bad anatomy), by bad-artist, bad-hands-5, bad-image-v2-39000, extra digit, fewer digits, (extra arms:1.2), blue cloak, Steps: 25, Sampler: DPM++ 2M Karras, CFG scale: 11, Seed: 478121638, Size: 568x768, Model: anything-v4.5-pruned, Clip skip: 2, ENSD: 31337, AddNet Enabled: True, AddNet Module 1: LoRA, AddNet Model 1: hogsksv2-000002(2e60f62c128c), AddNet Weight A 1: 0.95, AddNet Weight B 1: 0.95 </pre> </details>

<img src="https://huggingface.co/Aotsuyu/HogwartLora/resolve/main/images/3.png" width="568" height="768">

<details> <summary>Prompt</summary> <pre> best quality, 1girl, flandre scarlet, blonde hair, vampire, fangs, red eyes, hogsks, hogwarts school uniform, hufflepuff, black robe, yellow scarf, (:3:0.5), (gray vest:1.2), gray skirt, outside, snow, black-yellow robe, crystal wings, side ponytail Negative prompt: 2girls, (low quality, worst quality:1.4), (bad anatomy), by bad-artist, bad-hands-5, bad-image-v2-39000, extra digit, fewer digits, (extra arms:1.2), blue cloak, Steps: 25, Sampler: DPM++ 2M Karras, CFG scale: 11, Seed: 1055056090, Size: 568x768, Model: anything-v4.5-pruned, Clip skip: 2, ENSD: 31337, AddNet Enabled: True, AddNet Module 1: LoRA, AddNet Model 1: hogsksv2-000002(2e60f62c128c), AddNet Weight A 1: 0.95, AddNet Weight B 1: 0.95 </pre></details>

<img src="https://huggingface.co/Aotsuyu/HogwartLora/resolve/main/images/4.png" width="568" height="768">

<details> <summary>Prompt</summary> <pre> best quality, 1girl, gawr gura, (loli:0.5), ravenclaw, hogsks, hogwarts school uniform, black robe, blue scarf, shark teeth, (:3:0.5), (gray vest:1.2), Negative prompt: (low quality, worst quality:1.4), (bad anatomy), by (bad-artist:1.0), bad-hands-5, (bad-image-v2-39000:1.0), extra digit, fewer digits, (extra arms:1.2), Steps: 25, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 3622139475, Size: 568x768, Model: Qcha-hllv1, Denoising strength: 0.3, Clip skip: 2, ENSD: 31337, AddNet Enabled: True, AddNet Module 1: LoRA, AddNet Model 1: hogsksv2-000002(2e60f62c128c), AddNet Weight A 1: 0.95, AddNet Weight B 1: 0.95, Hires upscale: 2, Hires steps: 15, Hires upscaler: 4x-AnimeSharp </pre></details>

<img src="https://huggingface.co/Aotsuyu/HogwartLora/resolve/main/images/5.png" width="568" height="768">

<details> <summary>Prompt</summary> <pre> best quality, 1girl, black hair, glasses, gryffindor, hogsks, hogwarts school uniform, black robe, red scarf, (scared), (gray vest:1.2), looking at viewer, evening, night, dark Negative prompt: (low quality, worst quality:1.4), (bad anatomy), by (bad-artist:1.0), bad-hands-5, (bad-image-v2-39000:1.0), extra digit, fewer digits, (extra arms:1.2), Steps: 20, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 2895674484, Size: 568x768, Model: pastelmix-better-vae-fp32, Denoising strength: 0.74, Clip skip: 2, ENSD: 31337, AddNet Enabled: True, AddNet Module 1: LoRA, AddNet Model 1: hogsksv2-000002(2e60f62c128c), AddNet Weight A 1: 0.9, AddNet Weight B 1: 0.9, Hires upscale: 1.8, Hires steps: 20, Hires upscaler: Latent (nearest-exact) </pre></details> <br><br>

Model comparison

This is trained on base NAI so any models off of that should do fine. <img src="https://huggingface.co/Aotsuyu/HogwartLora/resolve/main/images/grid.png" width="840" height="964"> <br>

Contact

If you have any questions, you can DM me on twitter.

My pixiv if you're up for lewds: Pixiv