A Texture LoRA Lab transferring the tactile qualities of Japanese traditional materials into AI expression

TextureLoRALab is a Texture LoRA research project focused on transferring the material qualities of Japanese traditional materials into AI expression.
Through the creation and study of LoRA models, TextureLoRALab explores how the physical materiality, sheen, grain, surface texture, and dimensional qualities of materials such as gold leaf, mother-of-pearl, mineral pigments, washi paper, and lacquer can be reflected in AI-generated art.

With a background spanning Japanese painting, art studies, and museum studies at a UK graduate school, TextureLoRALab works with tools such as Stable Diffusion, ComfyUI, kohya_ss, and Python to build LoRA training and generation environments. Rather than simply reproducing visual style, the project pursues an approach that teaches AI the tactile qualities and material textures of the original substances themselves.

One of its core initiatives is the Texture LoRA “SHIFUKU Series,” which aims to bring material expressions derived from real traditional materials into AI generation, including the fine wrinkles of gold leaf, the granular surface of mineral pigments, and the deep gloss of lacquer. While general style LoRAs tend to learn the appearance or visual style of a work, TextureLoRALab is distinctive in its focus on the physical qualities of materials themselves.

TextureLoRALab also expands the practical possibilities of AI-based material expression through a production structure designed for commercial delivery and internal implementation, as well as texture packs, a note membership called “Urazaishiki,” and public LoRA releases on CivitAI.

By connecting Japanese traditional materials with AI technology, TextureLoRALab deeply explores the field of “texture” in AI art, combining research depth with practical creative application.

Main Activities

• LoRA research focused on transferring the textures of Japanese traditional materials into AI
• Creation and release of the Texture LoRA “SHIFUKU Series”
• LoRA distribution and research publication on CivitAI
• Operation of the note membership “Urazaishiki”
• Creation and distribution of texture packs
• Development of generative AI production environments using Stable Diffusion, ComfyUI, kohya_ss, and Python
• Research and development of AI-based material expression for commercial delivery and internal production workflows