<h1 id="sword-art-online-game-cg-style-lora-for-anima-1.0">Sword Art Online Game CG Style LoRA for Anima 1.0</h1>Fan-made style LoRA trained for Anima 1.0 to reproduce Sword Art Online game CG / anime cutscene look.My first LoRA training on Anima 1.0. It may need some adjustments, feedback is welcome.Even without <code>anime coloring</code> or <code>anime screencap</code>, the output already looks anime-styled, and the SAO-style eyes are noticeable.<h2 id="trigger">Trigger</h2><pre><code>@saogc</code></pre><h2 id="training-info">Training info</h2><ul><li>Base model: Anima 1.0</li><li>Training tool: diffusion-pipe</li><li>Type: LoRA</li><li>Rank: 32</li><li>Epoch: 30</li><li>Trigger prefix used during training: <code>@saogc,</code></li><li>Captions: Booru-style captions</li></ul><h2 id="note">Note</h2><code>anime coloring</code> and <code>anime screencap</code> are not required trigger words. They were only used in the comparison examples to create a closer non-LoRA baseline.

Sword Art Online Game CG Style

Initial public release.This is the epoch 30 checkpoint, selected as v1 because it gives a recognizable SAO game-CG style while still keeping decent prompt flexibility.

Files with SHA256: 3fc108e24e2f13bad108935d0efab54cca3f755ea3ff1dbbc97d03298b086ddb