Reconstructing textured meshes from colored point clouds is an important but challenging task in 3D graphics and vision. Most existing methods predict colors as implicit functions in 3D or UV space, suffering from blurry textures or the lack of generalization capability. Addressing this, we propose PointDreamer, a novel framework for textured mesh reconstruction from colored point cloud. It produces meshes with enhanced fidelity and clarity by 2D image inpainting, taking advantage of the mature techniques and massive data of 2D vision. Specifically, we first project the input point cloud into 2D space to generate sparse multi-view images, and then inpaint empty pixels utilizing a pre-trained 2D diffusion model. Next, we design a novel Non-Border-First strategy to unproject the colors of the inpainted dense images back to 3D space, thus obtaining the final textured mesh. In this way, our PointDreamer works in a zero-shot manner, requiring no extra training. Extensive qualitative and quantitative experiments on various synthetic and real-scanned datasets show the SoTA performance of PointDreamer, by significantly outperforming baseline methods with 30\% improvement in LPIPS score (from 0.118 to 0.068). Code at: https://github.com/YuQiao0303/PointDreamer....

PointDreamer: Zero-shot 3D Textured Mesh Reconstruction from Colored Point Cloud by 2D Inpainting