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

用戶LoRA

SD

Hyper Pecs, hybrid style LoRA V2.0

#獸人
AI 工具
  • AI 動漫生成器
  • 動漫手辦生成器
  • 角色設定圖生成器
  • 雜誌封面生成器
  • 玩偶生成器
  • 巨型雕像生成器

探索

  • 熱門 Tags
  • 排名
  • 模型市集
  • 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

  • 下載 PixAI App
  • App Store
  • Google Play
© 2026 PixAI
  • 服務條款
  • 隱私政策
  • 版權政策
  • 第三方內容聲明
  • 特定商業交易法
Hyper Pecs, hybrid style LoRA V2.0 - AI Model封面圖片
上傳時間
2023年8月10日 下午4:07
使用數
39.1k
評論
非常好  (10)
觸發詞

hyper pecs, gigantic pecs, huge pecs, big pecs

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

カロリン Caroline (Courageous Sister) - AI Model封面圖片

用戶LoRA

カロリン Caroline (Courageous Sister)
3
Knotted - AI Model封面圖片

用戶LoRA

Knotted
1
Lara Croft - Tomb Raider 2013 - bigASP V2 - AI Model封面圖片

用戶LoRA

Lara Croft - Tomb Raider 2013 - bigASP V2
4
立ち絵 - AI Model封面圖片

用戶LoRA

立ち絵
108

描述

https://civitai.com/models/93126?modelVersionId=137956
Author's notes:
ABOUT
Hyper Pecs aim to give muscle male character with bigger pecs or make the pecs muscle look more heavy and strong and perhaps soft too.
The Pecs size can be control on how strong the lora and the prompt you put into it. The stronger the strength you add to the prompt, the bigger it will get like <lora:Hyper Pecs v2:0.5> (gigantic pecs:1.2) . The lora strength at 0.5 should be strong enough, you just need to add the strength on the prompt.
Version 2.0
This version 2 I decide to make it as hybrid. I notice there are like 3 category when comes to style, realistic, anime and furry and each style tags differently. Regardless, Hyper Pecs initially only works for Furry model but now it should be possible to be used for other style model. If it still not able to generate with this lora, that model doesn't have the model I use to train this lora.
Problems
I don't know how to prompt for realistic so I always end up with weird result. I notice these realistic human model, when the pecs become too big, it will generate skin stretch marks, other than that would be weird nipple. Anime is not big deal, seems to work just fine which same goes with furry model.
=Training=
The Lora was trained with mix of Boyfusion, Unstable Homerotic Diffusion and fluffyrock-lion-low-lr-e27. All of them are Trained model and I believe they're been uses a lot in other model merge. By combine these model, I don't have to make separate training base on style. Sure it more effective that way but this is more efficient just need to only have 1 lora instead of 3.
Lora train pool is 1500 and I provided 20 images database for each style (Realistic, Anime and Furry) which after calculation training step result is 75. (1500 / 20 = 75) . The database for training can be improve in future but this will do for now.

留言