(日本語説明は後半にあります)The popular "Hotaru's LoRA series" (for SD1) is now available for SDXL! More will be released gradually on CivitAI (already <a target="_blank" rel="ugc" href="https://huggingface.co/JujoHotaru/lora/blob/main/sdxl/README.md">available on HuggingFace</a>).This LoRA reproduces <code>(&gt;_&lt;)</code> shaped eyes, which is familiar in Japanese manga or anime.Both Illustrious version and Pony version are available. Switch version tab above↑ to choose matching type for your checkpoint.This LoRA is suitable for 2D-2.5D illustrations only and not suitable for photorealistic checkpoints. My <a target="_blank" rel="ugc" href="https://civarchive.com/user/JujoHotaru/models?sort=Newest&amp;baseModels=Pony&amp;types=Checkpoint">"Breed" series checkpoints</a> will make good outputs. Sample images are generated with "Breed" series listed on "Suggested Resources" section.<a target="_blank" rel="ugc" href="https://civarchive.com/models/134236/comic-expression-eyes-1-1">SD1 version</a> is also available.This LoRA (v2.0) is developed by a new training method called "<a target="_blank" rel="ugc" href="https://note.com/hakomikan/n/n716397e39d56">ADDifT</a>" (Japanese written article), implemented in <a target="_blank" rel="ugc" href="https://github.com/hako-mikan/sd-webui-traintrain">TrainTrain extension</a>.<h3 id="usage-86mxs5fgm">Usage</h3>It works by simply enabling LoRA. No extra prompts are necessary, but do not use with <code>closed eyes</code> and eye color prompts.If the eye shape is not good, try increasing LoRA weight to 1.2-1.5.<hr />More various other SD1/SDXL LoRA's and checkpoints are available on my <a target="_blank" rel="ugc" href="https://huggingface.co/JujoHotaru">HuggingFace repository</a> or <a target="_blank" rel="ugc" href="https://civarchive.com/user/JujoHotaru/models">CivitAI models list</a>. Please check it out.<hr />SD1でご好評いただいていた十条蛍のLoRAシリーズが、SDXL対応になって再登場します。<a target="_blank" rel="ugc" href="https://huggingface.co/JujoHotaru/lora/blob/main/sdxl/README.md">HuggingFaceで先行公開中</a>ですが、CivitAIにも順次掲載していきます。デフォルメ表現でおなじみの、<code>(&gt;_&lt;)</code>の形の眼を再現します。Illustrious系・Pony系の2バージョンを公開しています。ページ上部のバージョンタブから、お使いのモデルタイプに合わせたものを選んでダウンロードしてください。2D・2.5Dイラスト系のデータモデル用で、実写系には不向きです。拙作データモデル「<a target="_blank" rel="ugc" href="https://civarchive.com/user/JujoHotaru/models?sort=Newest&amp;baseModels=Pony&amp;types=Checkpoint">Breedシリーズ</a>」と組み合わせると、高い効果を発揮します。サンプル画像もBreedシリーズ（"Suggested Resources"セクションに掲載）で生成されています。<a target="_blank" rel="ugc" href="https://civarchive.com/models/134236/comic-expression-eyes-1-1">SD1バージョン</a>も公開しています。このLoRA（v2.0）は、2025年3月発表の新しいLoRA学習方式である「<a target="_blank" rel="ugc" href="https://note.com/hakomikan/n/n716397e39d56">ADDifT</a>」（<a target="_blank" rel="ugc" href="https://github.com/hako-mikan/sd-webui-traintrain">TrainTrain拡張機能</a>で利用可能）を用いて開発されています。<h3 id="-fhnkcevy8">使い方</h3>LoRAを有効にするだけで発動しますが、<code>closed eyes</code>および目の色を示すプロンプトは記述しないようにしてください。効きが悪い場合は、LoRA適用強度を1.2～1.5くらいまで上げてみてください。<hr />他にもさまざまなSD1/SDXL用LoRAおよびデータモデルを公開していますので、<a target="_blank" rel="ugc" href="https://huggingface.co/JujoHotaru">HuggingFaceリポジトリ</a>または<a target="_blank" rel="ugc" href="https://civarchive.com/user/JujoHotaru/models">CivitAIの配布モデルリスト</a>をご覧ください。

[SDXL] Comic expression eyes 01 / デフォル眼01

hotarueye_xl_comic01_v20_pony.safetensors

v2.0 for Pony

v2.0 for Illustrious

v1.0

The popular "Hotaru's LoRA series" (for SD1) is now available for SDXL! More will be released gradually on CivitAI (already <a target="_blank" rel="ugc" href="https://huggingface.co/JujoHotaru/lora/blob/main/sdxl/README.md">available on HuggingFace</a>).This LoRA reproduces (&gt;_&lt;) shaped eyes, which is familiar in Japanese manga or anime.This LoRA is compatible with Pony / Animagine / Illustrious based checkpoints, but Pony based checkpoints will make better results. Some checkpoints may have less compatibility.This LoRA is suitable for 2D-2.5D illustrations only and not suitable for photorealistic checkpoints. My <a target="_blank" rel="ugc" href="https://civitai.com/user/JujoHotaru/models?sort=Newest&amp;baseModels=Pony&amp;types=Checkpoint">"Breed" series checkpoints</a> will make good outputs. Sample images are generated with "Breed" series listed on "Suggested Resources" section.<a target="_blank" rel="ugc" href="https://civitai.com/models/134236/comic-expression-eyes-1-1">SD1 version</a> is also available.UsageIt works by simply enabling LoRA. No extra prompts are necessary, but do not use with closed eyes and eye color prompts.If the eye shape is not good (especially using with Illustrious or Animagine based checkpoints), try increasing LoRA weight to 1.5-2.0.More various other SD1/SDXL LoRA's and checkpoints are available on my <a target="_blank" rel="ugc" href="https://huggingface.co/JujoHotaru">HuggingFace repository</a> or <a target="_blank" rel="ugc" href="https://civitai.com/user/JujoHotaru/models">CivitAI models list</a>. Please check it out.CAUTION : Some sample images' prompt is using <a target="_blank" rel="ugc" href="https://github.com/hako-mikan/sd-webui-negpip">NegPiP extension</a> for minus weight.

Files with SHA256: c6d497b6fe633655a62be6bed9078bd595e79bc22a3d78f55a1bd08b22af42eb