About this document
LangScene-X: 3D Scene Reconstruction by khanakmittal92 is a document available to read on EtoBox.
LangScene-X is a novel generative framework designed to reconstruct generalizable 3D language-embedded scenes from sparse views using TriMap video diffusion and a Language Quantized Compressor (LQC). It enables the generation of consistent RGB images, normal maps, and semantic maps from as few as two input images, facilitating open-ended language queries in 3D environments. Extensive experiments demonstrate its superiority over existing methods in terms of quality and generalizability, making it a significa
- Author
- khanakmittal92
- Language
- EN