6月の季節物今回は日本伝統の花嫁衣装の白無垢角隠しと綿帽子の区別はしてなく､例によってv001は顔が潰れやすい<h2 id="v001-018xj5njz">v001 スクリプトのテスト</h2><a target="_blank" rel="ugc" href="https://github.com/NEXTAltair/Lorayougazouwoiroirosurusukuriputo">画像前処理用スクリプト</a>アップデートしてsqlデータベースから非推奨タグを推奨タグに置き換える処理を追加その後dataset tag edditerで重複タグの削除とトリガー｢shiromuku｣とだけ追加して学習画像から余計な情報のクリーンナップとかもやってない<h2 id="v002-ssrjtpgph">v002</h2>既存タグに対してクリーンナップしたタグを追加してたので修正したスクリプトを使用<h2 id="v003-8a6bbwgjc">v003</h2>タグが大文字･小文字混在なのを全て小文字に統一asian､ Chinese､ korean など文化圏を判別できてないタグの手動削除sd-script/dev の '--alpha_mask' を使用トリガーを 'shiromuku, 白無垢'へ､'japanese clothing'を'kimono'へ変更<h2 id="v004-7krk5ljct">v004</h2>手動でタグ整理､とても手間な割には出来が満足とも言えない角隠しを理解させようとその画像を増やしたが学ばない<h1 id="eng-mglbvnt1k">ENG</h1>Seasonal for JuneThis time, it's the traditional Japanese bridal attire, Shiromuku.There is no distinction between tsunokakushi and wataboshi, and as usual, v001 tends to distort the face.<h2 id="v001-script-test-6pcpaq7ni">v001 Script Test</h2>Image preprocessing scriptUpdated to include a process that replaces deprecated tags with recommended tags from the SQL database.After that, duplicate tags were removed using the dataset tag editor, and only the trigger "shiromuku" was added for training.No cleanup of unnecessary information from the images was done either.<h2 id="v002-r3r2a2s5q">v002</h2>Added cleaned-up tags to existing tags, using a corrected script.<h2 id="v003-19kbvcdy8">v003</h2>Unified tags to lowercase as they were mixed case.Manually removed tags like "asian", "Chinese", "korean" which could not distinguish cultural regions.Used <code>--alpha_mask</code> from <code>sd-script/dev</code>.Changed trigger from "shiromuku, 白無垢" to "shiromuku" and "japanese clothing" to "kimono".<h2 id="v004-8isz9ojff">v004</h2>Manually organized tags; despite the effort, the results are unsatisfactory.Increased images to make the model understand "角隠し" (horn concealing headdress), but it did not learn.

白無垢 shiromuku-XLPon

shiromuku-XLPony_v004.safetensors

v004

v003

v002

v001

creator: nextaltair6月の季節物今回は日本伝統の花嫁衣装の白無垢角隠しと綿帽子の区別はしてなく､例によってv001は顔が潰れやすいv001 スクリプトのテスト<a href="https://github.com/NEXTAltair/Lorayougazouwoiroirosurusukuriputo" rel="ugc noopener noreferrer nofollow" target="_blank">画像前処理用スクリプト</a>アップデートしてsqlデータベースから非推奨タグを推奨タグに置き換える処理を追加その後dataset tag edditerで重複タグの削除とトリガー｢shiromuku｣とだけ追加して学習画像から余計な情報のクリーンナップとかもやってないv002既存タグに対してクリーンナップしたタグを追加してたので修正したスクリプトを使用v003タグが大文字･小文字混在なのを全て小文字に統一asian､ Chinese､ korean など文化圏を判別できてないタグの手動削除sd-script/dev の '--alpha_mask' を使用トリガーを 'shiromuku, 白無垢'へ､'japanese clothing'を'kimono'へ変更v004手動でタグ整理､とても手間な割には出来が満足とも言えない角隠しを理解させようとその画像を増やしたが学ばないENGSeasonal for JuneThis time, it's the traditional Japanese bridal attire, Shiromuku.There is no distinction between tsunokakushi and wataboshi, and as usual, v001 tends to distort the face.v001 Script TestImage preprocessing scriptUpdated to include a process that replaces deprecated tags with recommended tags from the SQL database.After that, duplicate tags were removed using the dataset tag editor, and only the trigger "shiromuku" was added for training.No cleanup of unnecessary information from the images was done either.

Files with SHA256: 3aaf881cf0db8f752210377c71d82b889c32310d3510d5bd9a4cafa95d6f3b4c