灵动
LINGDONG
Local AI Image Generation, Privacy First
Powered by Z-Image-Turbo (fast) with an optional Base model for higher quality. All image generation happens entirely on your device — no internet required, your creative privacy protected.
Features
Text to Image
Describe what you want in English or Chinese, AI generates stunning images instantly
Fully Local
No internet required. Your data never leaves your device — maximum privacy protection
ControlNet Guide Image
Use reference images to guide generation — control pose, edges, depth and more
LoRA Style Presets
Built-in artistic styles, import local LoRA files for custom styles
Native Acceleration
Built on MLX framework, fully utilizing Apple Silicon GPU performance
Gallery Management
Auto-save generation history, easily browse, search, and export your creations
System Requirements
macOS 14.0+
Sonoma or later
Apple Silicon
M1 / M2 / M3 / M4 series chips
16GB+ RAM
24GB recommended for best experience
15GB Disk Space
For model file storage (~12GB for 8-bit model)
Download Now
The AI model (~12GB) will be downloaded automatically on first launch
What's New in v1.0.10
- Base model support - Higher-quality generation option (recommended: 50 steps, guidance 4)
FAQ
Where are models stored?
LingDong downloads AI models from HuggingFace and stores them in the cache directory:
~/.cache/huggingface/hub/models/
The models are stored in subdirectories:
mzbac/Z-Image-Turbo-8bit/(~12 GB, Turbo 8-bit)Tongyi-MAI/Z-Image-Turbo/(~31 GB, Turbo Full)mzbac/Z-Image-8bit/(~12 GB, Base 8-bit)Tongyi-MAI/Z-Image/(~31 GB, Base Full)mzbac/Z-Image-Turbo-Fun-Controlnet-Union-8bit/(ControlNet, if used)
How do I free up disk space?
Option 1: Using the app
- Open LingDong
- Go to Settings → Storage
- Click Delete Downloaded Models
Option 2: Manual deletion
Delete only LingDong models:
rm -rf ~/.cache/huggingface/hub/models/mzbac/Z-Image*
rm -rf ~/.cache/huggingface/hub/models/Tongyi-MAI/Z-Image*
rm -rf ~/.cache/huggingface/hub/models/alibaba-pai/Z-Image*
Will I need to re-download after deleting?
Yes. The next time you generate an image, the model will be downloaded again automatically. This is a one-time download that typically takes a few minutes depending on your internet speed.
How do the Inpainting mask tools work?
When using Inpainting, you paint a mask to define which areas to regenerate. The mask appears as a red overlay on your image for easy visibility.
- Red/painted areas = Will be regenerated (replaced with AI-generated content)
- Unpainted areas = Will be preserved (keep the original image)
Brush tool: Paint over areas you want to regenerate. These appear as red overlay.
Eraser tool: Remove painted areas from regeneration. Use it to correct mistakes or refine your selection.
Invert button: Swaps painted and unpainted areas across the entire mask. Useful when it's easier to paint what you want to keep rather than what you want to change.
Clear button: Resets the entire mask to unpainted (preserves everything).
Additional tools: Adjustable brush size (5-200px), Undo/Redo (Cmd+Z / Cmd+Shift+Z), and a magnifying loupe for precise painting.