DIST Rendering Deep Implicit Signed Distance Function with Differentiable Sphere Tracing
Published:
optimize forward and backward pass of neural rendering
- tldr, coarse-to-fine, aggressive ray marching
- apply to any sdf model
- sample per 4 pixel and double each 3 iterations
- at 6 iter, back to origin res
- aggressive ray marching
- x = x + 1.5 * sdf(x) * rd
- sample points under surface
- converge if a pixel donot confuse with its neighbor
- refer to supp. for details
- gradient bp
- tbh, i donot understand it