Material Realism is a visual realism LoRA focused on physically believable surface materials. It enhances microtexture, roughness variation, natural imperfections, weathering, erosion, pores, grains, cracks, scratches, stains, worn edges, and realistic light interaction across stone, rock, concrete, brick, wood, metal, fabric, soil, and other surfaces.

The LoRA is designed to improve material realism without defining the overall composition or subject. It works especially well with landscape, environment, architectural, and object-focused LoRAs, adding convincing physical surface detail while allowing the primary LoRA to control the overall visual style.
The training dataset (999 picture 1024x1024) is primarily composed of stone, rock, concrete, brick, wood, metal, fabric, soil, and related physical materials. It is not specifically trained as a skin, fur, feather, or biological-surface LoRA.