<h1>Toudou Yurika (Aikatsu) LoRA</h1><h2>02-05-2023/V5 Update:</h2><ul><li>Reduce the model file size to 9MB.</li></ul><ul><li>Use a way lower network dim (8 instead of 160) and Alpha (1 instead of 160) to prevent overfitting</li><li>Add a new token for each canon outfit (using booru token was a bad idea): <code>GothMagicCoord</code><code>StarlightAcademyWinter</code><code>CasualDress</code></li></ul>This is a LoRA, this mean that you need to update Voldy/AUTOMATIC1111's webui to use it (use <code>git pull</code>) .That LoRA does not have a particular trigger (except the default <code>&lt;lora:LoRaName:1&gt;</code>).<h3>Weight:</h3>V3 : I mostly tested it at weight <code>1.0</code> and it work well between weight <code>0.7</code> and <code>1.0</code>. But the more you reduce the weight the less her canon outfit are accurate.<img src="https://imagecache.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/17460efc-b78a-4bbe-32a1-765734b2b300/width=525" /><img src="https://imagecache.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/a4177073-4dd5-4e12-8f39-0d2eaf1adf00/width=525" /><img src="https://imagecache.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/cb93f34d-4f14-4865-13ff-cab2ec0b6500/width=525" /><h3>Recommended prompt:</h3>That LoRA can reproduice some of her canon outfit.Goth Magic Coord (purple/blue gothic dress dress):<pre><code>purple_dress,drill hair,mini top hat,gothic lolita, idol,bat wings, white ascot,frilled dress,purple headwear, striped pantyhose, vertical-striped legwear, uneven legwear, black wings, wings</code></pre>V5: Add : <code>GothMagicCoord</code>Starlight Academy uniform:<pre><code>starlight academy uniform,white skirt,miniskirt</code></pre>V5: Add : <code>StarlightAcademyWinter</code>Casual dress:<pre><code>pinafore dress, green dress, white shirt,frilled sleeves, red ribbon, neck ribbon, glasses, black-framed eyewear</code></pre>V5: Add : <code>CasualDress</code><h3>Training settings:</h3><pre><code>101 images, 10 repeats,7 epoch,7070 steps
Learning rate: 1e-4
Text encoder learning rate: 5e-5 
Unet learning rate: 1e-4 
max bucket resolution: 512 
clip skip: 2 
dim/alpha: 160 
batch size=1 
Base model: AnythingV3, this mean that this LoRA work on any model derived from that particular model like AnythingV4/4.5, Pastel Mix, AOM 1/2,etc...</code></pre><h3>Model used for the example:</h3>V3:Pastel Mix, AnythingV4.5 and AbyssOrangeMix2 nsfw . V5:AnythingV4.5That character owned by Bandai Namco and Sunrise.

Toudou Yurika (Aikatsu) LoRA (9.1MB update)

<pre><code>109 images, 10 repeats,6 epoch,5450 steps
Text encoder learning rate: 1e-5 
Unet learning rate: 1e-3 
max bucket resolution: 512 
clip skip: 2 
dim/alpha: 8 
batch size=1 
Base model: NAI</code></pre>

Files with SHA256: 39c1b4856dfdd5c2d1af29dfce28b3c87f6782fae6f02ec4ec97dc964d79c9b8