The research team at Tsinghua University has developed the Convolutional Reconstruction Model (CRM) to achieve a breakthrough in generating 3D models from uploaded images. The CRM generates orthogonal view images and processes them through a convolutional U-Net, allowing for the creation of high-fidelity textured meshes within 10 seconds without the need for optimization. The team demonstrated the CRM's performance on a variety of objects, opening new possibilities for single-image to 3D textured mesh generation.