In general, weight 1.0 is best (In some cases 0.95 or 0.9 works better)You can add or remove prompt if you don't wantChangeable clothes (General usage)<pre><code>tsukasadef, long hair, pink hair, hair ornament, bangs, red eyes, hair between eyes</code></pre>Daily modern clothes<pre><code>tsukasasuit, long hair, hair ornament, long sleeves, pink hair, frills, white blouse, black skirt</code></pre>Pajamas hair bun<pre><code>tsukasabun, bangs, red eyes, hair between eyes, collarbone, pink hair, hair bun, double bun, pink shirt, pajamas</code></pre>Daily modern clothes needs some generation to produce good imagePajamas hair bun is easier to generate 前陣子很熱衷這部動漫，發現女角lora的檔案都很大，而且也缺少一些造型，決定自己做一個。 這是我做的第一個算滿意的lora模型 (我給它90分但我懶得再試了🤣)，因為是第一次研究，加上我的顯卡不好跑很慢，前後花了好幾天才搞懂一些訓練的技巧，特別感謝gesen2gee的幫忙。

Yuzaki Tsukasa 由崎司 (Tsukuyomi Tsukasa)

tsukasa10.safetensors

Train_data.zip

v1.0

In general, weight 1.0 is best (In some cases 0.95 or 0.9 works better)You can add or remove prompt if you don't wantChangeable clothes (General usage)tsukasadef, long hair, pink hair, hair ornament, bangs, red eyes, hair between eyesDaily modern clothestsukasasuit, long hair, hair ornament, long sleeves, pink hair, frills, white blouse, black skirtPajamas hair buntsukasabun, bangs, red eyes, hair between eyes, collarbone, pink hair, hair bun, double bun, pink shirt, pajamas<ul><li>Daily modern clothes needs some generation to produce good image</li><li>Pajamas hair bun is easier to generate</li></ul>

Files with SHA256: b7f4a32563b82069e2cb9523e448d0378985e26be5d5e56064d535f7a0e7189d