Assignment 1

1. Practicing with Cameras

1.1. 360-degree Renders

360-degree render

1.2 Re-creating the Dolly Zoom

Dolly zoom

2. Practicing with Meshes

2.1 Constructing a Tetrahedron

4 vertices, 4 triangle faces.

Tetrahedron

2.2 Constructing a Cube

8 vertices, 12 triangle faces.

Cube

3. Re-texturing a mesh

color1 = [0, 0, 1] (blue at the front)

color2 = [1, 0, 0] (red at the back)

Retexturing cow

4. Camera Transformations

1. Rotating the camera about the z-axis 90°, not changing the camera position.

\[ R_{\mathrm{relative}} = \begin{bmatrix} 0 & 1 & 0 \\ -1 & 0 & 0 \\ 0 & 0 & 1 \end{bmatrix},\qquad T_{\mathrm{relative}} = \begin{bmatrix}0 \\ 0 \\ 0\end{bmatrix} \]

1

2. Rotating the camera 90° about the y-axis, using the translation to keep the cow centered in the image.

\[ R_{\mathrm{relative}} = \begin{bmatrix} 0 & 0 & 1 \\ 0 & 1 & 0 \\ -1 & 0 & 0 \end{bmatrix},\qquad T_{\mathrm{relative}} = \begin{bmatrix}-3 \\ 0 \\ 3\end{bmatrix} \]

2

3. Not changing the camera rotation, moving 2 units farther from the cow.

\[ R_{\mathrm{relative}} = \begin{bmatrix} 1 & 0 & 0 \\ 0 & 1 & 0 \\ 0 & 0 & 1 \end{bmatrix},\qquad T_{\mathrm{relative}} = \begin{bmatrix}0 \\ 0 \\ 2\end{bmatrix} \]

3

4. Not changing the camera rotation, moving 0.5 units along -x and 0.5 units along +y (both in world frame).

\[ R_{\mathrm{relative}} = \begin{bmatrix} 1 & 0 & 0 \\ 0 & 1 & 0 \\ 0 & 0 & 1 \end{bmatrix},\qquad T_{\mathrm{relative}} = \begin{bmatrix}0.5 \\ -0.5 \\ 0\end{bmatrix} \]

4

5. Rendering Generic 3D Representations

5.1 Rendering Point Clouds from RGB-D Images

First point cloud Second point cloud Union point cloud

5.2 Parametric Functions

A torus point cloud:

Parametric torus

A bowl-shaped point cloud.

Parametric bowl

5.3 Implicit Surfaces

A torus mesh:

Implicit torus

A bowl-shaped mesh.

Implicit bowl

  1. Rendering speed

    Point-cloud rendering depends on the number of points, and mesh rendering depends on the number of triangle faces. For continuous and regular surfaces (like cubes), mesh may be rendered faster (at similar rendering quality) since it requires fewer triangle faces than points.

  2. Rendering quality

    Point cloud is not connected, so if the points are sparse, there might be gaps or holes between points. Mesh has innate connectivity. The vertices and faces form a surface. It can show surface structure and shape more directly than point cloud can, and can be rendered with lighting, textures, and continuous shading.

  3. Ease of use

    Point cloud is easier to construct since it only needs points (and colors). Meshes also require connectivity information. For an implicit surface, rendering as a point cloud requires sampling points near the zero level set, while rendering as a mesh requires marching cubes.

  4. Memory usage Theoretically, storing mesh requires more memory space than point cloud when the vertices and points are the same, since mesh needs to store the connectivity information. When representing smooth surfaces or simple 3D shapes (like cubes), mesh can sometimes use less memory because it is better at capturing regular surface structure. Point cloud memory depends on resolution.

6. Do Something Fun

I want to show a toy example of our solar system, with just the Sun, Earth, and Moon. I combined several parametric spheres (for showing planets) and torus (for showing orbits).

Solar system view 1 Solar system view 2 Solar system view 3 Solar system view 4

7. Sampling Points on Meshes

10 points:

10

100 points:

100

1000 points:

1000

10000 points:

10000