16-825 Learning for 3D Vision

Yiwen Zhao's Project Page

Assignment 1

1. Practicing with Cameras (15 Points)

1.1. 360-degree Renders (5 points)

360 cow

1.2 Re-creating the Dolly Zoom (10 points) dolly zoom

2. Practicing with Meshes (10 Points)

2.1 Constructing a Tetrahedron (5 points)

tetrahedron

2.2 Constructing a Cube (5 points)

cube

3. Re-texturing a mesh (10 points)

tetrahedron

color1=[0, 0, 1], color2=[1, 0, 0]

4. Camera Transformations (10 points)

tetrahedron

R_relative, rotate -90 degrees about the z-axis, T_relative, do nothing.

tetrahedron

R_relative, do nothing, T_relative, move 3 units forward along the negative z-axis (the object's front direction).

tetrahedron

R_relative, do nothing, T_relative, move 0.5 units forward along the positive x-axis and 0.5 units forward along the negative y-axis.

tetrahedron

R_relative, rotate 90 degrees about the y-axis, T_relative, move 3 units forward along the negative x-axis and 3 units forward along the positive y-axis.

5. Rendering Generic 3D Representations (45 Points)

5.1 Rendering Point Clouds from RGB-D Images (10 points)

image image image

5.2 Parametric Functions (10 + 5 points)

image image

5.3 Implicit Surfaces (15 + 5 points)

image image

Mesh has faster speed but lower quality than point cloud, because it introduces more approximation and less sample points. Point cloud is easier to use since its a natrul format in data collection. Rendering as a point cloud can capture fine details without requiring explicit surface reconstruction, but generally requires more memory

6. Do Something Fun (10 points)

image

A CMU color heart shape.