Assignment 3

1. Differentiable Volume Rendering

Grid/ray visualization:

1.4. Point sampling (10 points)

Visualization of the point samples from the first camera:

1.5. Volume rendering (30 points)

Visualization:

2. Optimizing a basic implicit volume

2.2. Loss and training (5 points)

Box center: (0.25, 0.25, 0.00)

Box side lengths: (2.01, 1.50 1.50)

2.3. Visualization

Visualization:

3. Optimizing a Neural Radiance Field (NeRF) (30 points)

Visualization:

4. NeRF Extras (Choose at least one! More than one is extra credit)

4.1 View Dependence (10 pts)

Visualization:

Adding view dependence makes the model capture more details of the object. But increasing view dependence may lead the model to overfitting training images.

Here are some visualizations of models with and without view dependence. The first image is the ground truth lego model. The second one is rendered by the model with view dependence and the third one is rendered by the model without view dependence.

We can find that the tail light in the second image is darker than the third one and the circle in the second image is more complete than the third one. Thus, removing view dependence prevents the model from capturing some details of the object.