| Property | Value |
|---|---|
| Author | strangerzonehf |
| Base Model | FLUX.1-dev |
| Training Images | 49 |
| Optimal Resolution | 1280x832 |
| Network Dimensions | 64 |
| Repository | https://huggingface.co/strangerzonehf/Halftone-Recraft-Flux |
What is Halftone-Recraft-Flux?
Halftone-Recraft-Flux is a specialized LoRA model built on the FLUX.1-dev foundation, designed specifically for generating stylized automotive photography with halftone effects. The model has been trained on 49 carefully curated images using florence2-en labeling for natural language processing in English.
Implementation Details
The model utilizes an AdamW optimizer with a constant learning rate scheduler, featuring a network dimension of 64 and alpha of 32. Training parameters include a noise offset of 0.03 and multires noise iterations set to 10, with the model trained across 23 epochs.
- Optimized for 3:2 aspect ratio (1280x832)
- Recommended inference steps: 30-35
- Uses bfloat16 precision for efficiency
- Implements noise discounting at 0.1
Core Capabilities
- Specialized in automotive photography generation
- Halftone effect processing
- High-detail car exterior rendering
- Dynamic lighting and perspective handling
Frequently Asked Questions
Q: What makes this model unique?
This model specializes in creating halftone-style automotive photographs with precise control over perspective and lighting, particularly effective for detailed car exterior shots with specific viewing angles and lighting conditions.
Q: What are the recommended use cases?
The model excels at generating detailed automotive photography, particularly suited for creating professional-looking car photographs with stylized halftone effects. It's optimized for specific viewing angles and can handle complex lighting scenarios, making it ideal for automotive marketing and artistic applications.