ControlNet Canny SDXL 1.0
| Property | Value |
|---|---|
| License | OpenRAIL++ |
| Base Model | Stable Diffusion XL 1.0 |
| Downloads | 30,407 |
| Community Rating | 485 likes |
What is controlnet-canny-sdxl-1.0?
controlnet-canny-sdxl-1.0 is a specialized ControlNet model trained on Stable Diffusion XL base 1.0, designed to provide precise control over image generation using Canny edge detection. This model enables users to generate highly detailed images while maintaining structural fidelity to input edge maps.
Implementation Details
The model was trained in two phases: initially for 20,000 steps on LAION 6a with 384px resolution, followed by another 20,000 steps on minimum 1024px images. Training utilized 8xA100 GPUs with a batch size of 64 and employed fp16 mixed precision.
- Constant learning rate of 1e-4 scaled by batch size
- Data parallel processing with 8 per GPU batch size
- Optimized for high-resolution image generation
Core Capabilities
- High-quality edge-guided image generation
- Support for high-resolution outputs
- Seamless integration with SDXL pipeline
- Efficient CPU offloading support
- Controllable generation through conditioning scale
Frequently Asked Questions
Q: What makes this model unique?
This model combines the power of SDXL with precise edge-based control, allowing for detailed structural guidance in image generation while maintaining the high-quality output characteristic of SDXL.
Q: What are the recommended use cases?
The model excels at tasks requiring structural precision such as architectural visualization, portrait manipulation, and detailed scene reconstruction where edge guidance is crucial for maintaining specific shapes and layouts.
