Q1.3
Q1.4
Q1.5
Q2.2
center of box: (0.25, 0.25, -0.00)
side lengths: (2.01, 1.50, 1.50)
Q2.3
Q3
Q4.1
I have added the view dependency to my code. Increased view dependency improves the generalization quality, but it can also causes the overfit to training images.
Q4.3
With the decrease of the points samples per ray, the training results loses the details. Which means, With more points samples per ray, the result gets better. However, when we increase the points samples per ray bigger than 128, the results seem very similar. Also, when we increase the point samples, the training time becomes longer. So the choice of 128 points samples per ray is a trade-off between the quality of the result and the training time.
High-res w points samples per ray=256:
High-res w points samples per ray=128:
High-res w points samples per ray=64:
High-res w points samples per ray=32:
