Fire magic
使用したAI
Stable Diffusion XL
「ワンセンテンス生成」を試してみましたが、思ったような結果を得ることができませんでした。おそらく使用しているモデルやLoraには品質向上のためのプロンプトが必要です。そのため、少し調整しました。キャラクターの説明プロンプトは保持しつつ、キャラクターの動作のみを単文で生成し、衣装や背景はランダムに設定しました。ただし、以前掲載した画像の品質に合わせるため、画面上での不自然な部分については後処理で修正を行いました。
I tried "one-sentence generation," but couldn't achieve the desired results, possibly because the model and Lora I'm using need some quality prompts. So, I made some adjustments. While keeping the character description prompts, I only used a single sentence to generate the character's action, leaving the outfit and background as random. However, to match the quality of previously published images, I did some post-editing to correct any logical inconsistencies in the visuals.
See more R18+ images and versions without watermarks:
https://www.patreon.com/c/daniuchen
I tried "one-sentence generation," but couldn't achieve the desired results, possibly because the model and Lora I'm using need some quality prompts. So, I made some adjustments. While keeping the character description prompts, I only used a single sentence to generate the character's action, leaving the outfit and background as random. However, to match the quality of previously published images, I did some post-editing to correct any logical inconsistencies in the visuals.
See more R18+ images and versions without watermarks:
https://www.patreon.com/c/daniuchen
呪文
入力なし
