@Lam_Kin_Hang

Day 11
This is how it looks right now using the fine-tuned LoRA. However, it has to sacrifice quality and resolution to achieve a higher FPS. I’m thinking of using a 4B model from Hugging Face or adding a new function that allows the user to scan their room first, then batch the generated images so it doesn’t have to create a new one every time.

Day 10
I have fine-tuned my own LoRA, which is specially designed for generating medieval-style images. I used kohya_ss to fine-tune the SDXL model and used ComfyUI to build the workflow shown below.