Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,185 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: apache-2.0
|
| 3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
language:
|
| 4 |
+
- en
|
| 5 |
+
- zh
|
| 6 |
+
base_model:
|
| 7 |
+
- Tongyi-MAI/Z-Image-Turbo
|
| 8 |
+
pipeline_tag: text-to-image
|
| 9 |
+
library_name: diffusers
|
| 10 |
+
tags:
|
| 11 |
+
- text-to-image
|
| 12 |
+
- image-generation
|
| 13 |
+
- diffusion
|
| 14 |
+
- comfyui
|
| 15 |
+
- photorealistic
|
| 16 |
+
- bilingual
|
| 17 |
+
- chinese
|
| 18 |
+
- english
|
| 19 |
+
- 8-step
|
| 20 |
+
- fast-generation
|
| 21 |
+
---
|
| 22 |
+
|
| 23 |
+
# π Z-Image-Turbo-AIO | 8-Step Photorealistic Generation
|
| 24 |
+
|
| 25 |
+
<div align="center">
|
| 26 |
+
|
| 27 |
+
**Ultra-Fast β’ Bilingual Text Rendering β’ All-in-One β’ FP8 & BF16**
|
| 28 |
+
|
| 29 |
+
[](https://opensource.org/licenses/Apache-2.0)
|
| 30 |
+
[](https://github.com/comfyanonymous/ComfyUI)
|
| 31 |
+
|
| 32 |
+
</div>
|
| 33 |
+
|
| 34 |
+
## β¨ What is Z-Image-Turbo-AIO?
|
| 35 |
+
|
| 36 |
+
Z-Image-Turbo-AIO is an **All-in-One repackage** of Alibaba Tongyi Lab's 6B parameter photorealistic image generator, optimized for lightning-fast 8-step generation. This version includes **integrated VAE and Text Encoder** for maximum convenience - just download and generate!
|
| 37 |
+
|
| 38 |
+
### Available Versions
|
| 39 |
+
|
| 40 |
+
| Version | Size | Best For |
|
| 41 |
+
|---------|------|----------|
|
| 42 |
+
| π‘ **FP8-AIO** | ~10GB | Most users, testing, everyday use |
|
| 43 |
+
| π **BF16-AIO** | ~20GB | Maximum quality, professional work |
|
| 44 |
+
|
| 45 |
+
## π― Key Features
|
| 46 |
+
|
| 47 |
+
- β‘ **8-step generation** - 10-40 seconds per image
|
| 48 |
+
- π¦ **All-in-One** - No separate VAE/Text Encoder downloads needed
|
| 49 |
+
- πΈ **Photorealistic** - Professional quality output
|
| 50 |
+
- π **Bilingual** - English & Chinese text rendering
|
| 51 |
+
- π― **8GB VRAM** - Works on RTX 4060 and similar
|
| 52 |
+
- π **Apache 2.0** - Open license for any use
|
| 53 |
+
|
| 54 |
+
## π Which Version Should I Choose?
|
| 55 |
+
|
| 56 |
+
### π‘ FP8-AIO (Recommended for most users)
|
| 57 |
+
- β
Half the file size
|
| 58 |
+
- β
Faster downloads
|
| 59 |
+
- β
Excellent quality
|
| 60 |
+
- β
Perfect for 8GB VRAM
|
| 61 |
+
- β
Great for testing & everyday use
|
| 62 |
+
|
| 63 |
+
### π BF16-AIO (Maximum precision)
|
| 64 |
+
- β
BFloat16 full precision
|
| 65 |
+
- β
Absolute best quality
|
| 66 |
+
- β
Professional/commercial grade
|
| 67 |
+
- β
Still works on 8GB VRAM
|
| 68 |
+
|
| 69 |
+
## π₯ Quick Start (ComfyUI)
|
| 70 |
+
|
| 71 |
+
### Installation
|
| 72 |
+
|
| 73 |
+
1. Download your preferred version (FP8 or BF16)
|
| 74 |
+
2. Place in `ComfyUI/models/checkpoints/`
|
| 75 |
+
3. Load with "Load Checkpoint" node
|
| 76 |
+
4. Generate!
|
| 77 |
+
|
| 78 |
+
### Recommended Settings
|
| 79 |
+
|
| 80 |
+
| Parameter | Value |
|
| 81 |
+
|-----------|-------|
|
| 82 |
+
| Steps | 8 |
|
| 83 |
+
| CFG | 1.0 |
|
| 84 |
+
| Sampler | res_multistep |
|
| 85 |
+
| Scheduler | simple |
|
| 86 |
+
| Resolution | 1920Γ1088 |
|
| 87 |
+
|
| 88 |
+
**That's it! No separate VAE or Text Encoder needed!**
|
| 89 |
+
|
| 90 |
+
## π Performance
|
| 91 |
+
|
| 92 |
+
All tests on **RTX 4060 (8GB VRAM)** β’ FP8 β’ 1920Γ1088 β’ 8 steps
|
| 93 |
+
|
| 94 |
+
| Test | Generation Time |
|
| 95 |
+
|------|-----------------|
|
| 96 |
+
| Urban Interior | ~32s |
|
| 97 |
+
| Architecture | ~32-34s |
|
| 98 |
+
| Food Photography | ~32s |
|
| 99 |
+
| Bilingual Signage | ~32s |
|
| 100 |
+
|
| 101 |
+
## π‘ Prompting Guide
|
| 102 |
+
|
| 103 |
+
### β
Natural Language Works Best!
|
| 104 |
+
|
| 105 |
+
**Good Example:**
|
| 106 |
+
```
|
| 107 |
+
A cozy bookstore with floor-to-ceiling wooden shelves filled with
|
| 108 |
+
colorful books, comfortable reading nooks with cushions near large
|
| 109 |
+
windows, warm pendant lighting, peaceful afternoon atmosphere,
|
| 110 |
+
professional interior photography
|
| 111 |
+
```
|
| 112 |
+
|
| 113 |
+
**Bad Example:**
|
| 114 |
+
```
|
| 115 |
+
bookstore, books, chairs, window, cozy, warm light, interior
|
| 116 |
+
```
|
| 117 |
+
|
| 118 |
+
### π Bilingual Text Rendering
|
| 119 |
+
|
| 120 |
+
**English Text:**
|
| 121 |
+
```
|
| 122 |
+
Neon sign reading "OPEN 24/7" in bright blue letters above entrance.
|
| 123 |
+
Modern sans-serif font, glowing effect against brick wall.
|
| 124 |
+
```
|
| 125 |
+
|
| 126 |
+
**Chinese Text:**
|
| 127 |
+
```
|
| 128 |
+
Traditional tea house entrance with sign reading "ε€ι΅θΆε" in elegant
|
| 129 |
+
gold Chinese calligraphy on red wooden board with ornate carved border.
|
| 130 |
+
```
|
| 131 |
+
|
| 132 |
+
**Both Languages:**
|
| 133 |
+
```
|
| 134 |
+
Modern cafe exterior with bilingual sign. "Morning Brew Coffee" in
|
| 135 |
+
white elegant script above, "ζ¨ζ¦εε‘" in matching Chinese characters
|
| 136 |
+
below. Both glowing warmly at dusk.
|
| 137 |
+
```
|
| 138 |
+
|
| 139 |
+
### π Prompting Tips
|
| 140 |
+
|
| 141 |
+
| Do β
| Don't β |
|
| 142 |
+
|------|---------|
|
| 143 |
+
| Use natural language descriptions | Use tag-style prompts (tag1, tag2) |
|
| 144 |
+
| Be detailed (100-300 words optimal) | Write very short prompts (<50 words) |
|
| 145 |
+
| Include lighting and mood | Add negative prompts (not used) |
|
| 146 |
+
| Describe camera angle and style | Include conflicting instructions |
|
| 147 |
+
| Specify materials and colors | |
|
| 148 |
+
|
| 149 |
+
## π Credits & Acknowledgments
|
| 150 |
+
|
| 151 |
+
### Original Model
|
| 152 |
+
- **Developer:** Tongyi Lab (Alibaba Group)
|
| 153 |
+
- **Architecture:** Single-Stream Diffusion Transformer (6B parameters)
|
| 154 |
+
- **Algorithm:** Decoupled-DMD + DMDR
|
| 155 |
+
- **License:** Apache 2.0
|
| 156 |
+
|
| 157 |
+
### AIO Conversion
|
| 158 |
+
- **Created by:** [SeeSee21](https://huggingface.co/SeeSee21)
|
| 159 |
+
- **Format:** Integrated VAE + Text Encoder
|
| 160 |
+
- **Purpose:** Simplified single-file deployment
|
| 161 |
+
|
| 162 |
+
### Resources
|
| 163 |
+
- π€ [Original HuggingFace](https://huggingface.co/Tongyi-MAI/Z-Image-Turbo)
|
| 164 |
+
- π» [GitHub Repository](https://github.com/Tongyi-MAI/Z-Image)
|
| 165 |
+
- π¨ [ComfyUI Files](https://huggingface.co/Comfy-Org/z_image_turbo)
|
| 166 |
+
- πΌοΈ [CivitAI Page](https://civitai.com/models/2173571)
|
| 167 |
+
|
| 168 |
+
## π Version History
|
| 169 |
+
|
| 170 |
+
### v1.0 - Initial AIO Release
|
| 171 |
+
- FP8-AIO version (10GB)
|
| 172 |
+
- BF16-AIO version (20GB)
|
| 173 |
+
- Integrated VAE + Text Encoder
|
| 174 |
+
- Single-file deployment
|
| 175 |
+
- Based on Tongyi-MAI/Z-Image-Turbo
|
| 176 |
+
- Tested on RTX 4060 8GB
|
| 177 |
+
- Optimized for 1920Γ1088
|
| 178 |
+
|
| 179 |
+
---
|
| 180 |
+
|
| 181 |
+
<div align="center">
|
| 182 |
+
|
| 183 |
+
**Download, load with "Load Checkpoint", and generate professional photos in seconds! π**
|
| 184 |
+
|
| 185 |
+
</div>
|