True Boundary Realism XL_Red_X

0
0
0
0
RealisticStyle & FilterGirlStyle BoostPlant
最近更新:首次發布:
Realistic,Style & Filter,Girl,Checkpoint,SDXLImage info
Realistic,Style & Filter,Girl,Checkpoint,SDXLImage info
Realistic,Style & Filter,Girl,Checkpoint,SDXLImage info
Realistic,Style & Filter,Girl,Checkpoint,SDXLImage info
Realistic,Style & Filter,Girl,Checkpoint,SDXLImage info
Realistic,Style & Filter,Girl,Checkpoint,SDXLImage info
Realistic,Style & Filter,Girl,Checkpoint,SDXLImage info
Realistic,Style & Filter,Girl,Checkpoint,SDXLImage info
Realistic,Style & Filter,Girl,Checkpoint,SDXLImage info
Realistic,Style & Filter,Girl,Checkpoint,SDXLImage info
Realistic,Style & Filter,Girl,Checkpoint,SDXLImage info
Realistic,Style & Filter,Girl,Checkpoint,SDXLImage info
Realistic,Style & Filter,Girl,Checkpoint,SDXLImage info
Realistic,Style & Filter,Girl,Checkpoint,SDXLImage info
Realistic,Style & Filter,Girl,Checkpoint,SDXLImage info
Realistic,Style & Filter,Girl,Checkpoint,SDXLImage info
Realistic,Style & Filter,Girl,Checkpoint,SDXLImage info
Realistic,Style & Filter,Girl,Checkpoint,SDXLImage info
Realistic,Style & Filter,Girl,Checkpoint,SDXLImage info
Realistic,Style & Filter,Girl,Checkpoint,SDXLImage info

Model description:

"True Reality Photography XL_Red_X": An independent version in the XL realistic model, took one week to complete with over 200 hours of testing;

It has gone through various experiences and challenges;

This version benefits from recent advancements in XL portrait models and understanding, leaning towards a more realistic photography style in terms of artistic control, for example, character keywords like "k-pop girl" can influence the character's portrayal;

RED version direction: will emphasize the independent and professional direction in the subsequent production of models; all RED versions will be labeled with the keyword RED;

Modeling method:

Large model fusion + LORA + fine-tuning;

In this production, the large model used the old white as the base film, added the realistic Asian rabbit and monkey model helloword4.0, as well as dominator v6, v9, and whitev10;

After fusion, it becomes the foundation of the model;

①Using a total of 16 LORA for model direction and detail fine-tuning;

②Fine-tuning was applied to the large model in at least 3 versions, finally forming a stable version v3.3;

Usage:

CFG: 5-7;

Sampling: DPM++ 2M SDE Exponential; DPM++2M Karras; Euler

Aspect ratio: 832x1280 and other ratios you prefer



Recommended related models:

1.5 Portrait series:   "Photography Player v1", "Photography Player v2", "mix80", "mix90"

SDXL series:  "True Reality Photography v1", "True Reality Photography v2 Items",  "True Reality Photography v3"

Architectural models: "Tower Top Structure", "ARC Interior Building Model 0.4", "ARC Interior Building Model 0.45"

SDXL_Red model: "True Reality Photography v3.3"


Example image: Pure prompt, without cushioning images, without CN, and without other interfering factors.

In theory, with a locked seed, the output images are the same




討論

最熱門
|
最新
發送
即將推出
下載
(0.00KB)
詳細資訊
類型
線上生成次數
0
下載次數
0
推薦參數
採樣器
CFG
5.5
VAE

圖庫

最熱門
|
最新