A research team from Tsinghua University made a major breakthrough and successfully developed a model called Convolutional Reconstruction Model (CRM), which can generate a high-fidelity textured 3D model in just 10 seconds by uploading only a picture. This technology does not require a complex optimization process and is implemented by generating orthogonal view images and convolutional U-Net processing. This research result brings new possibilities to the field of generating three-dimensional models from a single image, and its high efficiency and high accuracy are expected to be widely used in various fields.
The Tsinghua University research team developed the Convolutional Reconstruction Model (CRM) model to achieve a breakthrough in generating 3D models from uploaded images. By generating orthogonal view images and convolutional U-Net processing, CRM can generate high-fidelity textured meshes in 10 seconds without optimization. The research team demonstrated the performance of CRM on a variety of objects, opening up new possibilities for single image to 3D texture mesh generation.
The emergence of the CRM model marks a new milestone in the technology of generating 3D models from a single image. Its high efficiency and convenience bring unlimited possibilities to the fields of virtual reality, augmented reality and three-dimensional modeling in the future. It is worth looking forward to its follow-up application and development.