DIST: Rendering Deep Implicit Signed Distance Function with Differentiable Sphere Tracing
This work addresses the computational bottleneck in inverse graphics for 3D reconstruction, offering a more efficient and differentiable rendering method for researchers and practitioners in computer vision and graphics.
The paper tackles the problem of efficiently rendering deep implicit signed distance functions (SDFs) for 3D shape reconstruction by proposing a differentiable sphere tracing algorithm, which reduces function queries and memory usage, enabling accurate 3D shape reconstruction from inputs like sparse depth and multi-view images with improved generalization and robustness.
We propose a differentiable sphere tracing algorithm to bridge the gap between inverse graphics methods and the recently proposed deep learning based implicit signed distance function. Due to the nature of the implicit function, the rendering process requires tremendous function queries, which is particularly problematic when the function is represented as a neural network. We optimize both the forward and backward passes of our rendering layer to make it run efficiently with affordable memory consumption on a commodity graphics card. Our rendering method is fully differentiable such that losses can be directly computed on the rendered 2D observations, and the gradients can be propagated backwards to optimize the 3D geometry. We show that our rendering method can effectively reconstruct accurate 3D shapes from various inputs, such as sparse depth and multi-view images, through inverse optimization. With the geometry based reasoning, our 3D shape prediction methods show excellent generalization capability and robustness against various noises.