<pre><code>I don't care about guys! Not one bit...</code></pre>Main rival craftknight from legendary yuri game Summon Night: Swordcraft Story, being the first introduced and last fought before credits. She’s a tsundere big sister type with an inferiority complex compared to her own older sister. Base: Sanary_SCS, green eyes, red hair, high ponytail, ponytail, neck ribbon Normal: pink jacket, juliet sleeves, belt, sheath, weapon on back, belt, black skirt, white boots, fold-over boots, knee boots, “Ribbon ‘n Lace Sanary”: hair bow, neck ribbon, puffy sleeves, maid, black dress, frills, back bow (note: "back bow" really, really wants to turn the subject around, even with strong negatives and there's not a lot of pics of this outfit in training data anyways)Made with 22 pics. 3 official art, a pixel portrait, and 18 pieces of fanart (11 of which were monochrome). Training data wasn’t nearly as good as Razzy unfortunately. This LoRA was the first where I attempted to minimize file size by decreasing network dim and alpha to 16.

Sanary - Summon Night: Swordcraft Story / サナレ - サモンナイトクラフトソ-ド物語

Sanary_-_Summon_Night_Swordcraft_Story__サナレ_-_サモンナイトクラフトソ-ド物語.safetensors

926694_training_data.zip

<pre><code>I don't care about guys! Not one bit...</code></pre>Main rival craftknight from legendary yuri game Summon Night: Swordcraft Story, being the first introduced and last fought before credits. She’s a tsundere big sister type with an inferiority complex compared to her own older sister. Base: Sanary_SCS, green eyes, red hair, high ponytail, ponytail, neck ribbon Normal: pink jacket, juliet sleeves, belt, sheath, weapon on back, belt, black skirt, white boots, fold-over boots, knee boots, “Ribbon ‘n Lace Sanary”: hair bow, neck ribbon, puffy sleeves, maid, black dress, frills, back bow (note: "back bow" really, really wants to turn the subject around, even with strong negatives and there's not a lot of pics of this outfit in training data anyways)Made with 22 pics. 3 official art, a pixel portrait, and 18 pieces of fanart (11 of which were monochrome). Training data wasn’t nearly as good as Razzy unfortunately. This LoRA was the first where I attempted to minimize file size by decreasing network dim and alpha to 16.

Files with SHA256: 355f796cb0a0cbe7337d00f1bc4bdac6ab23e4adda5b192110c9e031ed32bffc