16-825 Learning for 3D Vision

Assignment 1: Rendering Basics with PyTorch3D

Name: Joshua Tsai

Andrew ID: joshuat3

1. Practicing with Cameras

1.1 360-degree Renders

1.2 Re-creating the Dolly Zoom

2. Practicing with Meshes

2.1 Constructing a Tetrahedron

4 vertices, 4 triangular faces.

2.2 Constructing a Cube

8 vertices, 12 triangular faces.

3. Re-texturing a Mesh

color1 = [0, 0, 0] (black), color2 = [1, 0, 0] (red).

4. Camera Transformations

Transformation 1

Rotate around the viewing axis. No relative translation.

Transformation 2

Identity rotation. Move farther away along the depth direction.

Transformation 3

Translate in the image plane while keeping the same orientation.

Transformation 4

Rotate around the vertical axis to obtain a side view and translate to keep the cow centered.

5. Rendering Generic 3D Representations

5.1 Rendering Point Clouds from RGB-D Images

Left: first RGB-D view. Middle: second RGB-D view. Right: union of both point clouds.

5.2 Parametric Functions

Torus

Extra Credit: Klein Bottle

5.3 Implicit Surfaces

Implicit Torus

Mesh vs. Point Cloud

Meshes usually look smoother and more complete because triangle connectivity defines the surface, but they require extra face information and processing. Point clouds are simpler, but sparse points can create visible holes and dense point clouds require more memory and computation.

Extra Credit: Gyroid

6. Extra Credit: Do Something Fun

Intergalactic spaceship mesh with procedural vertex coloring and a 360-degree rendering.

Model source: Free3D - Intergalactic Spaceship

7. Extra Credit: Sampling Points on Meshes

Original mesh and point clouds with 10, 100, 1000, and 10000 sampled points.