Tao Pang
Training data: posed images
(x,y,z): point in the 3D volume containing the scene.
d: direction at which the camera looks at the point.
Goal: render new views.
Intensity:
Learns
Color(RGB):
Volumetric rendering
The color of a pixel is the weighted sum of the colors along the ray.
The weight wi of a point is high if:
Depth can also be estimated from NeRF:
NeRF:
(similar to α)
camera ray
Pixel (a)
camera ray
Pixel (b)
Pre-trained
C(x,y,z)
C(x,y,z,d)
Pre-trained
C(x,y,z)
C(x,y,z,d)
Brighter
darker
Pixel (a)
Pixel (b)
Pre-trained
C(x,y,z)
C(x,y,z,d)
Pre-trained
Position only
Position and direction
Pixel a
Pixel b