File size: 1,245 Bytes
bded544
5f08b65
 
 
bded544
2f66141
a2b33b3
 
2f66141
3ab8372
cfa32aa
adf6aab
 
 
 
7634562
adf6aab
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
---
license: other
license_name: flux-1-dev-non-commercial-license
license_link: https://huggingface.co./black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md
---
# Verus Vision — Model for Realistic Image Generation

# Overview
**Verus Vision** is a fine-tuned model designed for generating high-quality, realistic images. Built on the **Flux Dev (de-distill)** model with 12 billion parameters.

<hr>

# How to Use Transformer Version

**For ComfyUI:**
* Place the **AE** in the following path: **ComfyUI_windows_portable\ComfyUI\models\vae**
* Place **Clip-L** and **T5XXL** in the following path: **ComfyUI_windows_portable\ComfyUI\models\clip**
* Place the **Transformer** in the following path: **ComfyUI_windows_portable\ComfyUI\models\unet** or **ComfyUI_windows_portable\ComfyUI\models\diffusion_models**

<hr>

# Optimal Generation Settings

* Sampling Method and Schedule Type: Euler Beta
* Sampling Steps: 25+
* CFG Scale: 3.5
 
> [!WARNING]
> **Warning:**  Do not use a negative prompt as it can lead to the generation of images with artifacts and distortions.

<hr>

# Limitations

* This version of the model does not have a variety of styles and is intended for realism and photorealism.
* There may be issues with anatomy and text.