Limitations.DTC has limitations. Despite our efforts to inject fine-grained details, the model may still overlook fine-grained details, particularly in text engravings, a challenge inherent to Stable Diffusion (see Fig.7). Additionally, the model might alter human poses since it doesn’t consider...
The primary requirement is a machine that can run Stable Diffusion and LLaMA. I personally use my local Windows machine that carries a 4090 GPU, but you can tweak the instructions easily for a cloud machine or a weaker GPU (generation will just take longer). ...