On the Subject of Fashion Designer
• Built an interactive image generation tool using the LLaMA 3.1 model with Ollama and the DreamShaper model, integrated with the Huggingface Diffusers library.
• Implemented a two-step process where user prompts are refined by an LLM before being passed to the image generator, enhancing output quality.
• Added an intelligent feedback system that provides users with prompt improvement tips, tricks, and common mistake alerts to guide future use.
• Integrated fashion-designer-specific LoRA modules and designed the system to support dynamic combination of multiple LoRAs in the future.
• Deployed the tool with a responsive and dynamic Streamlit interface, offering a smooth, real-time experience for both text processing and image generation.
Project Genre
Technologies Used