Feature Request: Support for SDXL DreamBooth Models and FLUX LoRAs for Face, Character, and Style Training

Current Functionality:

Users can only create SD 1.5 DreamBooth models, limiting the range of model customization and use cases.

Requested Feature Enhancement:

Add support for creating SDXL DreamBooth models and/or FLUX LoRAs. This request includes the ability to train distinct model types (DreamBooth and LoRA) for different use cases:

SDXL Dreambooth Models

Enable users to create DreamBooth models using the SDXL architecture, allowing for enhanced custom fine-tuning of the model for face, character, and style training. SDXL Dreambooth would offer users the ability to create personalized models with higher fidelity and richer detail compared to SD 1.5.

FLUX LoRAs

Allow users to create FLUX LoRAs specifically for face, character, and style customizations. LoRAs offer a more lightweight and flexible training option than DreamBooth, enabling users to apply style or character traits as modular add-ons to their existing models without needing full model re-training. This makes FLUX LoRAs suitable for users seeking quicker, less resource-intensive customization options.

Rationale:

Adding support for both SDXL Dreambooth and FLUX LoRAs would provide users with a more versatile model training toolkit, allowing them to choose the best approach (DreamBooth for comprehensive, high-detail training and LoRA for efficient modular training) based on their specific needs.