Training methods, 3D reconstruction methods, devices, storage media, and program products for 3D Gaussian point rendering models.

Abstract

This disclosure relates to a training method, a 3D reconstruction method, an apparatus, a storage medium, and a program product for a 3D Gaussian point rendering model. The training method for the 3D Gaussian point rendering model includes: obtaining a pair of reference images of a target scene; inputting first camera parameters corresponding to a first reference image into the 3D Gaussian point rendering model corresponding to the target scene to generate a first depth map corresponding to the first camera parameters; inputting second camera parameters corresponding to a second reference image into the 3D Gaussian point rendering model to generate a second depth map corresponding to the second camera parameters; determining a depth consistency loss value for the 3D Gaussian point rendering model based on the difference information between the first and second depth maps; and updating the parameters of the 3D Gaussian point rendering model according to the depth consistency loss value. This disclosure can effectively eliminate floating points during the training process of the 3D Gaussian point rendering model, achieving more stable and efficient 3D reconstruction and new perspective generation effects.

Images