PixAI 會員
PixAI
PixAI
Generate
Sign in
主頁
生成
模型
Studio
NEW
比賽
用 Mio.2 創作
工具箱
NEW
指南

用戶LoRA

XLDiT.1DiT.2
細節增強

score_up color/pose/angle/detail/quality

創作

  • AI 圖像生成器
  • AI 動畫生成器
  • 工具箱
  • 藝術主題產生器
  • 訓練 LoRA
  • Mio.2 智慧助手
  • Studio

探索

  • 熱門 Tags
  • 排名
  • 模型市場
  • 比賽
  • 新聞

關於

  • 指南
  • PixAI 部落格
  • Tsubaki.2
  • 認識 Mio,我們的吉祥物
  • 內容規則

定價與說明

  • 會員計劃
  • 購買點數
  • 聯絡

移動端 APP

  • App Store
  • Google Play
© 2026 PixAI
  • 服務條款
  • 隱私政策
  • 版權政策
  • 特定商業交易法
score_up color/pose/angle/detail/quality - AI Model cover image
訓練者
Avatar image of aj

aj

訓練你自己的 LoRA
上傳時間
2026年4月5日 凌晨2:00
使用數
19.3k
返利
576,559
評論
非常好  (2)
觸發詞

score_9, score_8_up, score_7_up, masterpiece, best quality, clean outlines, cel-shading, digital line art, sharp lighting, detailed, intricate, beautiful colors, harmonious color palette, well-balanced body, natural proportions, (dynamic pose:1.3),

許可權
  • 允許圖片生成與分享
  • 允許使用者下載此模型
  • 允許生成圖片用於商業用途

推薦 LoRA

Dogeza with Full-body view and facial close-up view - AI Model cover image

用戶LoRA

Dogeza with Full-body view and facial close-up view
18
[IL v0.1] Mimi - Princess Connect | ミミ プリンセスコネクト!Re:Dive (3+1 costumes) - AI Model cover image

用戶LoRA

[IL v0.1] Mimi - Princess Connect | ミミ プリンセスコネクト!Re:Dive (3+1 costumes)
8
Color & Face enhance - AI Model cover image

用戶LoRA

Color & Face enhance
2
蟲モンスター - AI Model cover image

用戶LoRA

蟲モンスター
0

描述

Lora特性特化型 :Lora特性を引き上げながら、画質効果を上げる

(安定型で作成しています。⚠quality,detailは盛りすぎると破綻確率が上がります)

source_animeは抜いてます。 Lora使わなくてもプロンプトでキャラは出ます (⚠有名キャラ、数多くインストールされているキャラ、モデル相性があります。)

anime Illustrationにしたい際に自身は入れるくらいにしてます。

自身はLora特徴、特性を使いたいのでsource_animeはあまり使いません。

LoRA Characteristic Specialization Style

​Concept: Enhancing the core traits of the LoRA while simultaneously boosting overall image quality. ​Key Features ​ Target: Focuses on character traits, unique features, and specific artistic characteristics. ​Stability First: This prompt/setting is designed for high stability.

​Caution on Over-Processing: Avoid overusing "quality" or "detail" tags, as excessive weights can lead to image artifacts or "deep-fried" results. ​Usage Tips

​Source Tags: I intentionally omit source_anime. This allows the LoRA’s specific characteristics to take center stage without being over-shadowed by generalized dataset tags.

​Character Consistency: Most characters can be generated via prompts alone; however, this approach is designed to leverage the LoRA for nuance. (Note: Performance may vary depending on the character's popularity and model compatibility).

​Stylization: If you specifically want an Anime Illustration look, I recommend adding that tag manually. Otherwise, the focus remains on extracting the maximum "essence" from the LoRA itself.

Negative:

score_6, score_5, score_4, low score, mid score, poor score, average quality, mediocre quality,

unfinished, draft, rough, incomplete details, simplified details,

weak lighting, flat shading, muddy shading, washed-out colors,

generic anime style, plain composition, boring pose,

low effort, rushed drawing, amateurish, inconsistent quality

評論