MUMBAI, India, Jan. 2 -- Intellectual Property India has published a patent application (202541123296 A) filed by Vellore Institute Of Technology, Vellore, Tamil Nadu, on Dec. 7, 2025, for 'ai-driven diffusion and lora models for customizable 3d room visualization.'
Inventor(s) include Anand Prabu Arun; Mukund Niranjan Senthilkumar; Sarveshwara Ram Manikandan; and Nehal Sreekand Menon.
The application for the patent was published on Jan. 2, under issue no. 01/2026.
According to the abstract released by the Intellectual Property India: "The present disclosure provides a system for AI-driven customizable 3D room visualization. The system includes an input interface configured to receive multimodal user inputs, a processing module configured to convert the multimodal user inputs into structured data, a diffusion model configured to generate an initial 3D room layout based on the structured data, a LoRA customization module configured to apply iterative refinements to the initial 3D room layout based on user feedback, and a rendering engine configured to generate a visualization of the refined 3D room layout. The multimodal user inputs include text descriptions, sketches, and mood boards. The processing module includes a natural language processor, a computer vision processor, and an image analysis processor. The diffusion model operates through iterative denoising that progressively refines random noise into spatially coherent 3D room structures, while the LoRA customization module applies lightweight neural fine-tuning techniques."
Disclaimer: Curated by HT Syndication.
