Spaces:
Running
on
Zero
Running
on
Zero
Update README.md
Browse files
README.md
CHANGED
@@ -8,4 +8,198 @@ sdk_version: 5.35.0
|
|
8 |
app_file: app.py
|
9 |
pinned: false
|
10 |
---
|
11 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
8 |
app_file: app.py
|
9 |
pinned: false
|
10 |
---
|
11 |
+
## ๐ FLUX ControlNet Image Upscaler: AI-Powered Super Resolution
|
12 |
+
|
13 |
+
### Transform Low-Resolution Images into Crystal-Clear High-Resolution Masterpieces
|
14 |
+
|
15 |
+
Welcome to **FLUX ControlNet Image Upscaler**, an advanced AI-powered image enhancement tool that leverages the FLUX.1-dev model with ControlNet technology to upscale images up to 4x their original resolution while preserving and enhancing fine details. Perfect for photographers, designers, and anyone needing high-quality image enlargement.
|
16 |
+
|
17 |
+
### What is FLUX ControlNet Image Upscaler?
|
18 |
+
|
19 |
+
FLUX ControlNet Image Upscaler is a **state-of-the-art super-resolution tool** that uses advanced diffusion models and ControlNet conditioning to intelligently upscale images. Unlike traditional interpolation methods, it generates new pixel information based on AI understanding of image content, resulting in sharp, detailed enlargements that look naturally rendered at higher resolutions.
|
20 |
+
|
21 |
+
### Key Features for Professional Image Enhancement
|
22 |
+
|
23 |
+
- **๐ 4X Upscaling**: Enlarge images up to 4 times their original dimensions
|
24 |
+
- **๐ฏ ControlNet Precision**: Advanced spatial control for detail preservation
|
25 |
+
- **๐ FLUX.1-dev Model**: Cutting-edge diffusion technology for quality
|
26 |
+
- **๐ผ๏ธ Smart Resizing**: Automatic aspect ratio preservation
|
27 |
+
- **โก GPU Acceleration**: Fast processing with CUDA support
|
28 |
+
- **๐จ Detail Enhancement**: AI generates missing pixel information
|
29 |
+
- **๐ Adjustable Parameters**: Fine-tune quality vs speed trade-offs
|
30 |
+
- **โ๏ธ Interactive UI**: Beautiful interface with dynamic snow effects
|
31 |
+
|
32 |
+
### How It Works
|
33 |
+
|
34 |
+
#### **Simple Upscaling Process**
|
35 |
+
1. **Upload Image**: Select any image you want to upscale
|
36 |
+
2. **Choose Factor**: Select 1x to 4x enlargement
|
37 |
+
3. **Click Run**: AI processes and enhances your image
|
38 |
+
4. **Compare Results**: Use slider to see before/after
|
39 |
+
|
40 |
+
#### **Advanced Technology**
|
41 |
+
- **ControlNet Conditioning**: Maintains structural integrity during upscaling
|
42 |
+
- **Diffusion Process**: Generates realistic details in enlarged areas
|
43 |
+
- **Smart Budget Management**: Automatically adjusts for memory constraints
|
44 |
+
- **Multi-step Inference**: Progressive refinement for optimal quality
|
45 |
+
|
46 |
+
### Perfect Use Cases
|
47 |
+
|
48 |
+
- **Photography Enhancement**: Enlarge photos without quality loss
|
49 |
+
- **Digital Art**: Upscale artwork for print or display
|
50 |
+
- **Web Graphics**: Convert low-res images for high-DPI screens
|
51 |
+
- **Archive Restoration**: Enhance old or small digital images
|
52 |
+
- **E-commerce**: Improve product image quality
|
53 |
+
- **Social Media**: Create high-res versions for platforms
|
54 |
+
- **Print Preparation**: Prepare images for large format printing
|
55 |
+
- **Game Assets**: Upscale textures and sprites
|
56 |
+
|
57 |
+
### Advanced Controls
|
58 |
+
|
59 |
+
- **Inference Steps**: Quality refinement (8-50 steps)
|
60 |
+
- **Upscale Factor**: Enlargement multiplier (1-4x)
|
61 |
+
- **ControlNet Scale**: Conditioning strength (0.1-1.5)
|
62 |
+
- **Seed Control**: Reproducible results
|
63 |
+
- **Random Seed**: Generate variations
|
64 |
+
- **Auto Resize**: Smart dimension handling
|
65 |
+
|
66 |
+
### Technical Specifications
|
67 |
+
|
68 |
+
- **Base Model**: black-forest-labs/FLUX.1-dev
|
69 |
+
- **ControlNet**: jasperai/Flux.1-dev-Controlnet-Upscaler
|
70 |
+
- **Framework**: Diffusers, Gradio, PyTorch
|
71 |
+
- **Precision**: bfloat16 for optimal performance
|
72 |
+
- **Max Resolution**: 1024x1024 pixel budget
|
73 |
+
- **GPU**: CUDA-enabled device recommended
|
74 |
+
- **Output Format**: High-quality JPEG
|
75 |
+
|
76 |
+
### Why Choose FLUX ControlNet Upscaler?
|
77 |
+
|
78 |
+
1. **Superior Quality**: AI-generated details surpass traditional methods
|
79 |
+
2. **Intelligent Enhancement**: Understands image content for better results
|
80 |
+
3. **Fast Processing**: GPU acceleration for quick turnaround
|
81 |
+
4. **User-Friendly**: Simple interface with powerful controls
|
82 |
+
5. **Flexible Options**: Customize quality/speed balance
|
83 |
+
6. **Professional Results**: Gallery-quality enlargements
|
84 |
+
|
85 |
+
### Best Practices for Optimal Results
|
86 |
+
|
87 |
+
- **Start with Clean Images**: Better input yields better output
|
88 |
+
- **Choose Appropriate Factor**: Higher isn't always better
|
89 |
+
- **Adjust Inference Steps**: More steps for complex images
|
90 |
+
- **Fine-tune ControlNet Scale**: Balance detail vs smoothness
|
91 |
+
- **Consider Content Type**: Adjust settings based on image type
|
92 |
+
- **Check Memory Limits**: Large upscales may need resizing
|
93 |
+
|
94 |
+
### Technical Innovation
|
95 |
+
|
96 |
+
This upscaler represents a breakthrough in image enhancement technology:
|
97 |
+
- **Context-Aware Generation**: AI understands what details to add
|
98 |
+
- **Structure Preservation**: Maintains original image integrity
|
99 |
+
- **Adaptive Processing**: Automatically handles various image types
|
100 |
+
- **Memory Efficient**: Smart budget management for large images
|
101 |
+
|
102 |
+
### Start Upscaling Today
|
103 |
+
|
104 |
+
Transform your low-resolution images into high-quality masterpieces. Whether for professional work or personal projects, FLUX ControlNet Upscaler delivers exceptional results with ease.
|
105 |
+
|
106 |
+
**Community**: [Discord - Openfree AI](https://discord.gg/openfreeai) | **More AI Tools**: [OpenFree Best AI Services](https://huggingface.co/spaces/openfree/Best-AI)
|
107 |
+
|
108 |
+
---
|
109 |
+
|
110 |
+
## ๐ FLUX ControlNet ์ด๋ฏธ์ง ์
์ค์ผ์ผ๋ฌ: AI ๊ธฐ๋ฐ ์ด๊ณ ํด์๋ ๋ณํ
|
111 |
+
|
112 |
+
### ์ ํด์๋ ์ด๋ฏธ์ง๋ฅผ ์ ๋ช
ํ ๊ณ ํด์๋ ๊ฑธ์์ผ๋ก ๋ณํ
|
113 |
+
|
114 |
+
**FLUX ControlNet ์ด๋ฏธ์ง ์
์ค์ผ์ผ๋ฌ**์ ์ค์ ๊ฒ์ ํ์ํฉ๋๋ค. ControlNet ๊ธฐ์ ์ด ์ ์ฉ๋ FLUX.1-dev ๋ชจ๋ธ์ ํ์ฉํ์ฌ ์ธ๋ฐํ ๋ํ
์ผ์ ๋ณด์กดํ๊ณ ํฅ์์ํค๋ฉด์ ์ด๋ฏธ์ง๋ฅผ ์๋ณธ ํด์๋์ ์ต๋ 4๋ฐฐ๊น์ง ํ๋ํ๋ ๊ณ ๊ธ AI ๊ธฐ๋ฐ ์ด๋ฏธ์ง ํฅ์ ๋๊ตฌ์
๋๋ค. ์ฌ์ง์๊ฐ, ๋์์ด๋ ๋ฐ ๊ณ ํ์ง ์ด๋ฏธ์ง ํ๋๊ฐ ํ์ํ ๋ชจ๋ ์ฌ๋์๊ฒ ์๋ฒฝํฉ๋๋ค.
|
115 |
+
|
116 |
+
### FLUX ControlNet ์ด๋ฏธ์ง ์
์ค์ผ์ผ๋ฌ๋?
|
117 |
+
|
118 |
+
FLUX ControlNet ์ด๋ฏธ์ง ์
์ค์ผ์ผ๋ฌ๋ ๊ณ ๊ธ ํ์ฐ ๋ชจ๋ธ๊ณผ ControlNet ์กฐ๊ฑด๋ถ ์ ์ด๋ฅผ ์ฌ์ฉํ์ฌ ์ด๋ฏธ์ง๋ฅผ ์ง๋ฅ์ ์ผ๋ก ํ๋ํ๋ **์ต์ฒจ๋จ ์ด๊ณ ํด์๋ ๋๊ตฌ**์
๋๋ค. ๊ธฐ์กด์ ๋ณด๊ฐ ๋ฐฉ๋ฒ๊ณผ ๋ฌ๋ฆฌ ์ด๋ฏธ์ง ์ฝํ
์ธ ์ ๋ํ AI์ ์ดํด๋ฅผ ๋ฐํ์ผ๋ก ์๋ก์ด ํฝ์
์ ๋ณด๋ฅผ ์์ฑํ์ฌ ๊ณ ํด์๋์์ ์์ฐ์ค๋ฝ๊ฒ ๋ ๋๋ง๋ ๊ฒ์ฒ๋ผ ๋ณด์ด๋ ์ ๋ช
ํ๊ณ ์์ธํ ํ๋๋ฅผ ์ ๊ณตํฉ๋๋ค.
|
119 |
+
|
120 |
+
### ์ ๋ฌธ ์ด๋ฏธ์ง ํฅ์์ ์ํ ์ฃผ์ ๊ธฐ๋ฅ
|
121 |
+
|
122 |
+
- **๐ 4๋ฐฐ ์
์ค์ผ์ผ๋ง**: ์๋ณธ ํฌ๊ธฐ์ ์ต๋ 4๋ฐฐ๊น์ง ์ด๋ฏธ์ง ํ๋
|
123 |
+
- **๐ฏ ControlNet ์ ๋ฐ๋**: ๋ํ
์ผ ๋ณด์กด์ ์ํ ๊ณ ๊ธ ๊ณต๊ฐ ์ ์ด
|
124 |
+
- **๐ FLUX.1-dev ๋ชจ๋ธ**: ํ์ง์ ์ํ ์ต์ฒจ๋จ ํ์ฐ ๊ธฐ์
|
125 |
+
- **๐ผ๏ธ ์ค๋งํธ ํฌ๊ธฐ ์กฐ์ **: ์๋ ์ข
ํก๋น ๋ณด์กด
|
126 |
+
- **โก GPU ๊ฐ์**: CUDA ์ง์์ผ๋ก ๋น ๋ฅธ ์ฒ๋ฆฌ
|
127 |
+
- **๐จ ๋ํ
์ผ ํฅ์**: AI๊ฐ ๋๋ฝ๋ ํฝ์
์ ๋ณด ์์ฑ
|
128 |
+
- **๐ ์กฐ์ ๊ฐ๋ฅํ ๋งค๊ฐ๋ณ์**: ํ์ง ๋ ์๋ ๊ท ํ ๋ฏธ์ธ ์กฐ์
|
129 |
+
- **โ๏ธ ์ธํฐ๋ํฐ๋ธ UI**: ๋์ ๋ ํจ๊ณผ๊ฐ ์๋ ์๋ฆ๋ค์ด ์ธํฐํ์ด์ค
|
130 |
+
|
131 |
+
### ์๋ ๋ฐฉ์
|
132 |
+
|
133 |
+
#### **๊ฐ๋จํ ์
์ค์ผ์ผ๋ง ํ๋ก์ธ์ค**
|
134 |
+
1. **์ด๋ฏธ์ง ์
๋ก๋**: ํ๋ํ๋ ค๋ ์ด๋ฏธ์ง ์ ํ
|
135 |
+
2. **๋ฐฐ์จ ์ ํ**: 1๋ฐฐ์์ 4๋ฐฐ ํ๋ ์ ํ
|
136 |
+
3. **์คํ ํด๋ฆญ**: AI๊ฐ ์ด๋ฏธ์ง๋ฅผ ์ฒ๋ฆฌํ๊ณ ํฅ์
|
137 |
+
4. **๊ฒฐ๊ณผ ๋น๊ต**: ์ฌ๋ผ์ด๋๋ก ์ ํ ๋น๊ต
|
138 |
+
|
139 |
+
#### **๊ณ ๊ธ ๊ธฐ์ **
|
140 |
+
- **ControlNet ์กฐ๊ฑด๋ถ ์ ์ด**: ์
์ค์ผ์ผ๋ง ์ค ๊ตฌ์กฐ์ ๋ฌด๊ฒฐ์ฑ ์ ์ง
|
141 |
+
- **ํ์ฐ ํ๋ก์ธ์ค**: ํ๋๋ ์์ญ์์ ์ฌ์ค์ ์ธ ๋ํ
์ผ ์์ฑ
|
142 |
+
- **์ค๋งํธ ์์ฐ ๊ด๋ฆฌ**: ๋ฉ๋ชจ๋ฆฌ ์ ์ฝ์ ์๋ ์กฐ์
|
143 |
+
- **๋ค๋จ๊ณ ์ถ๋ก **: ์ต์ ํ์ง์ ์ํ ์ ์ง์ ๊ฐ์
|
144 |
+
|
145 |
+
### ์๋ฒฝํ ์ฌ์ฉ ์ฌ๋ก
|
146 |
+
|
147 |
+
- **์ฌ์ง ํฅ์**: ํ์ง ์์ค ์์ด ์ฌ์ง ํ๋
|
148 |
+
- **๋์งํธ ์ํธ**: ์ธ์ ๋๋ ์ ์์ฉ ์ํธ์ํฌ ์
์ค์ผ์ผ
|
149 |
+
- **์น ๊ทธ๋ํฝ**: ๊ณ DPI ํ๋ฉด์ฉ ์ ํด์๋ ์ด๋ฏธ์ง ๋ณํ
|
150 |
+
- **์์นด์ด๋ธ ๋ณต์**: ์ค๋๋๊ฑฐ๋ ์์ ๋์งํธ ์ด๋ฏธ์ง ํฅ์
|
151 |
+
- **์ ์์๊ฑฐ๋**: ์ ํ ์ด๋ฏธ์ง ํ์ง ๊ฐ์
|
152 |
+
- **์์
๋ฏธ๋์ด**: ํ๋ซํผ์ฉ ๊ณ ํด์๋ ๋ฒ์ ์์ฑ
|
153 |
+
- **์ธ์ ์ค๋น**: ๋ํ ์ธ์์ฉ ์ด๋ฏธ์ง ์ค๋น
|
154 |
+
- **๊ฒ์ ์์
**: ํ
์ค์ฒ ๋ฐ ์คํ๋ผ์ดํธ ์
์ค์ผ์ผ
|
155 |
+
|
156 |
+
### ๊ณ ๊ธ ์ ์ด
|
157 |
+
|
158 |
+
- **์ถ๋ก ๋จ๊ณ**: ํ์ง ๊ฐ์ (8-50๋จ๊ณ)
|
159 |
+
- **์
์ค์ผ์ผ ๋ฐฐ์จ**: ํ๋ ๋ฐฐ์ (1-4๋ฐฐ)
|
160 |
+
- **ControlNet ์ค์ผ์ผ**: ์กฐ๊ฑด๋ถ ๊ฐ๋ (0.1-1.5)
|
161 |
+
- **์๋ ์ ์ด**: ์ฌํ ๊ฐ๋ฅํ ๊ฒฐ๊ณผ
|
162 |
+
- **๋ฌด์์ ์๋**: ๋ณํ ์์ฑ
|
163 |
+
- **์๋ ํฌ๊ธฐ ์กฐ์ **: ์ค๋งํธ ์น์ ์ฒ๋ฆฌ
|
164 |
+
|
165 |
+
### ๊ธฐ์ ์ฌ์
|
166 |
+
|
167 |
+
- **๊ธฐ๋ณธ ๋ชจ๋ธ**: black-forest-labs/FLUX.1-dev
|
168 |
+
- **ControlNet**: jasperai/Flux.1-dev-Controlnet-Upscaler
|
169 |
+
- **ํ๋ ์์ํฌ**: Diffusers, Gradio, PyTorch
|
170 |
+
- **์ ๋ฐ๋**: ์ต์ ์ฑ๋ฅ์ ์ํ bfloat16
|
171 |
+
- **์ต๋ ํด์๋**: 1024x1024 ํฝ์
์์ฐ
|
172 |
+
- **GPU**: CUDA ์ง์ ์ฅ์น ๊ถ์ฅ
|
173 |
+
- **์ถ๋ ฅ ํ์**: ๊ณ ํ์ง JPEG
|
174 |
+
|
175 |
+
### FLUX ControlNet ์
์ค์ผ์ผ๋ฌ๋ฅผ ์ ํํด์ผ ํ๋ ์ด์
|
176 |
+
|
177 |
+
1. **์ฐ์ํ ํ์ง**: AI ์์ฑ ๋ํ
์ผ์ด ๊ธฐ์กด ๋ฐฉ๋ฒ์ ๋ฅ๊ฐ
|
178 |
+
2. **์ง๋ฅ์ ํฅ์**: ๋ ๋์ ๊ฒฐ๊ณผ๋ฅผ ์ํ ์ด๋ฏธ์ง ์ฝํ
์ธ ์ดํด
|
179 |
+
3. **๋น ๋ฅธ ์ฒ๋ฆฌ**: ๋น ๋ฅธ ์ฒ๋ฆฌ๋ฅผ ์ํ GPU ๊ฐ์
|
180 |
+
4. **์ฌ์ฉ์ ์นํ์ **: ๊ฐ๋ ฅํ ์ ์ด ๊ธฐ๋ฅ์ด ์๋ ๊ฐ๋จํ ์ธํฐํ์ด์ค
|
181 |
+
5. **์ ์ฐํ ์ต์
**: ํ์ง/์๋ ๊ท ํ ์ฌ์ฉ์ ์ ์
|
182 |
+
6. **์ ๋ฌธ๊ฐ ๊ฒฐ๊ณผ**: ๊ฐค๋ฌ๋ฆฌ ํ์ง์ ํ๋
|
183 |
+
|
184 |
+
### ์ต์ ์ ๊ฒฐ๊ณผ๋ฅผ ์ํ ๋ชจ๋ฒ ์ฌ๋ก
|
185 |
+
|
186 |
+
- **๊นจ๋ํ ์ด๋ฏธ์ง๋ก ์์**: ๋ ๋์ ์
๋ ฅ์ด ๋ ๋์ ์ถ๋ ฅ ์์ฑ
|
187 |
+
- **์ ์ ํ ๋ฐฐ์จ ์ ํ**: ๋์์๋ก ํญ์ ์ข์ ๊ฒ์ ์๋
|
188 |
+
- **์ถ๋ก ๋จ๊ณ ์กฐ์ **: ๋ณต์กํ ์ด๋ฏธ์ง์๋ ๋ ๋ง์ ๋จ๊ณ
|
189 |
+
- **ControlNet ์ค์ผ์ผ ๋ฏธ์ธ ์กฐ์ **: ๋ํ
์ผ ๋ ๋ถ๋๋ฌ์ ๊ท ํ
|
190 |
+
- **์ฝํ
์ธ ์ ํ ๊ณ ๋ ค**: ์ด๋ฏธ์ง ์ ํ์ ๋ฐ๋ผ ์ค์ ์กฐ์
|
191 |
+
- **๋ฉ๋ชจ๋ฆฌ ์ ํ ํ์ธ**: ๋๊ท๋ชจ ์
์ค์ผ์ผ์ ํฌ๊ธฐ ์กฐ์ ํ์
|
192 |
+
|
193 |
+
### ๊ธฐ์ ํ์
|
194 |
+
|
195 |
+
์ด ์
์ค์ผ์ผ๋ฌ๋ ์ด๋ฏธ์ง ํฅ์ ๊ธฐ์ ์ ๋ํ๊ตฌ๋ฅผ ๋ํ๋
๋๋ค:
|
196 |
+
- **์ปจํ
์คํธ ์ธ์ ์์ฑ**: AI๊ฐ ์ถ๊ฐํ ๋ํ
์ผ ์ดํด
|
197 |
+
- **๊ตฌ์กฐ ๋ณด์กด**: ์๋ณธ ์ด๋ฏธ์ง ๋ฌด๊ฒฐ์ฑ ์ ์ง
|
198 |
+
- **์ ์ํ ์ฒ๋ฆฌ**: ๋ค์ํ ์ด๋ฏธ์ง ์ ํ ์๋ ์ฒ๋ฆฌ
|
199 |
+
- **๋ฉ๋ชจ๋ฆฌ ํจ์จ์ฑ**: ๋์ฉ๋ ์ด๋ฏธ์ง๋ฅผ ์ํ ์ค๋งํธ ์์ฐ ๊ด๋ฆฌ
|
200 |
+
|
201 |
+
### ์ค๋ ์
์ค์ผ์ผ๋ง ์์ํ๊ธฐ
|
202 |
+
|
203 |
+
์ ํด์๋ ์ด๋ฏธ์ง๋ฅผ ๊ณ ํ์ง ๊ฑธ์์ผ๋ก ๋ณํํ์ธ์. ์ ๋ฌธ ์๏ฟฝ๏ฟฝ๏ฟฝ์ด๋ ๊ฐ์ธ ํ๋ก์ ํธ๋ FLUX ControlNet ์
์ค์ผ์ผ๋ฌ๋ ์ฝ๊ฒ ํ์ํ ๊ฒฐ๊ณผ๋ฅผ ์ ๊ณตํฉ๋๋ค.
|
204 |
+
|
205 |
+
**์ปค๋ฎค๋ํฐ**: [Discord - Openfree AI](https://discord.gg/openfreeai) | **๋ ๋ง์ AI ๋๊ตฌ**: [OpenFree Best AI Services](https://huggingface.co/spaces/openfree/Best-AI)
|