With this LoRA model, we can transform anything into the iconic South Park style. Imagine inserting popular characters and personalities right into the bustling streets of South Park city. The foreground detail is usually spot-on, though the backgrounds could use some enhancement. Important! To achieve even better results, I propose using embeddings in addition to LoRA weights. This could further boost the quality of generation. I cannot append embeddings file to this publication, so i put here a link to them: <a target="_blank" rel="ugc" href="https://huggingface.co/alexionby/south_park_lora">lora+embeddings</a> The supplemental images were produced using the Diffusers library, so I won't be providing image metadata. However, you can find additional prompt information via the link above. Additionally, I am providing training <a target="_blank" rel="ugc" href="https://huggingface.co/datasets/alexionby/south_park_lora">dataset with captions</a>, so you could beat my results! You can support future updates <a target="_blank" rel="ugc" href="https://ko-fi.com/alexionby">https://ko-fi.com/alexionby</a>! If we collect 10 USD, I will add Randy Marsh to this LoRA model!

South Park (SDXL)

The dataset has been expanded to 32 images with detailed captions. Here is new link to <a rel="ugc" href="https://huggingface.co/alexionby/south_park_lora_v1-6">lora+embeddings</a>

Files with SHA256: 3d8b82a2ba3f5543c57575b4c19dcbd45f39b3da3ead399437117c09e3f216ed