(日本語説明は後半にあります)The popular "Hotaru's LoRA series (for SD1)" is now also available for SDXL! More will be released gradually on CivitAI and<a target="_blank" rel="ugc" href="https://huggingface.co/JujoHotaru/lora/blob/main/sdxl/README.md"> HuggingFace</a>.This LoRA is designed to improve the depiction of characters with animal ears (such as cat ears, dog ears, etc.) by suppressing the generation of human ears. When using this LoRA, human ears are effectively removed or not rendered, making the character's appearance more natural and consistent with the presence of animal ears.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.This LoRA 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-63spb8h5y">Usage</h3>It works by simply enabling LoRA. No extra prompts are necessary. The effect will be a little strong by default weight(1.0). Decreasing LoRA weight (approx. 0.7 - 1.0) can reduce bad effect on the painting or composition.<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対応や新作も加えて再登場します。<a target="_blank" rel="ugc" href="https://huggingface.co/JujoHotaru/lora/blob/main/sdxl/README.md">HuggingFace</a>およびCivitAIにて順次公開していきます。このLoRAを使用すると、猫耳・犬耳・狐耳など動物耳キャラクターを生成する際、「人間の耳」の描写を防止することができます。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。このLoRA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="-jnxi5df66">使い方</h3>LoRAを有効にするだけで発動します。階層調整済みですので、基本的には階層指定せず強度指定のみで大丈夫です。デフォルトで少し強めに効きますので、人耳が出てこない範囲で強度を弱めることで、絵柄や構図への影響を減少させることができます。<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。

Human ear eraser (for animal ear) / 人耳消去（ケモ耳用）

erase_human_ear_v10_pony.safetensors

v1.0 for Pony

v1.0 for Illustrious

This LoRA is designed to improve the depiction of characters with animal ears (such as cat ears, dog ears, etc.) by suppressing the generation of human ears. When using this LoRA, human ears are effectively removed or not rendered, making the character's appearance more natural and consistent with the presence of animal ears.Both Illustrious version and Pony version are available.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.This LoRA 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>.UsageIt works by simply enabling LoRA. No extra prompts are necessary. The effect is a little strong by default weight(1.0). Decreasing LoRA weight (0.5 - 1.0) can reduce bad effect on the painting or composition.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.

Files with SHA256: 11055603df51b2386670f94838868d632a160b3e3fe267fa7bc7ad214b24acb8