This LoRA has been trained on images of installations and sculptures by conceptual artist Yayoi Kusama, whose work is characterised by polkadot patterns, bright colours, and curved shapes - particularly her trademark stylised pumpkins.

This model can turn your prompts into avant-garde artworks - covered in polkadots, surrounded by mirrors and lights.

Two versions of this model are available: one trained against the base SD1.5 model, and one trained against a specialised cartoon-style checkpoint. The former will work well with base SD1.5 and photorealistic checkpoints, the latter will work better for Western animation and anime style checkpoints.