| Property | Value |
|---|---|
| Base Model | black-forest-labs/FLUX.1-dev |
| License | CreativeML OpenRAIL-M |
| Training Images | 36 |
| Network Dimensions | 64 |
| Optimal Resolution | 768x1024 |
What is Flux-Midjourney-Mix2-LoRA?
Flux-Midjourney-Mix2-LoRA is a specialized LoRA model trained to emulate MidJourney v6's distinctive aesthetic while leveraging the FLUX.1-dev base model. It excels in generating realistic images, particularly in portrait photography, modeling, and detailed close-up shots.
Implementation Details
The model utilizes an AdamW8 optimizer with constant learning rate scheduling and incorporates advanced noise processing with a 0.03 offset and 0.1 multires noise discount. Training was conducted over 28 epochs with 3660 steps, using florence2-en for natural language processing in English.
- Network Alpha: 32
- Training Steps: 3660
- Noise Iterations: 10
- Optimal Resolution: 768x1024
Core Capabilities
- High-quality portrait photography
- Realistic close-up shots
- Detailed modeling and demonstrations
- Versatile style adaptation from street photography to fashion
- Enhanced detail preservation in various lighting conditions
Frequently Asked Questions
Q: What makes this model unique?
This model uniquely combines FLUX.1-dev's capabilities with MidJourney v6's aesthetic, optimized through careful parameter tuning and specialized training on 36 high-quality images. The model excels particularly in realistic portrait generation and close-up photography.
Q: What are the recommended use cases?
The model is best suited for generating realistic portraits, fashion photography, detailed close-ups, and demonstration images. It performs optimally at 768x1024 resolution and requires the trigger phrase "MJ v6" in prompts for best results.