16-889 Assignment 1: Rendering Basics with PyTorch3D
initial information
1.cd into starter folder
import render_mesh
import dolly_zoom
import render_shapes
import camera_transforms
import render_generic
import extra_credit
import matplotlib.pyplot as plt
cow_path = "../data/cow.obj"
image_size = 256
1. Practicing with Cameras
1.1. 360-degree Renders (5 points)
cow_360_gif_path = "../images/cow_360.gif"
render_mesh.render_cow(cow_path=cow_path, image_size=image_size, output_path=cow_360_gif_path)
1.2 Re-creating the Dolly Zoom (10 points)
num_frames = 10
duration = 3
dolly_gif_path = "../images/dolly.gif"
dolly_zoom.dolly_zoom(
image_size=image_size,
num_frames=num_frames,
duration=duration,
output_file=dolly_gif_path,
)
2. Practicing with Meshes
2.1 Constructing a Tetrahedron (5 points)
tetragedron_gif_path = "../images/tetrahedron.gif"
render_shapes.render_tetrahedon(output_file=tetragedron_gif_path, image_size=image_size)
2.2 Constructing a Cube (5 points)
cube_gif_path = "../images/cube.gif"
render_shapes.render_cube(output_file=cube_gif_path, image_size=image_size)
3. Re-texturing a mesh (10 points)
cow_color_gif_path = "../images/cow_color.gif"
render_shapes.render_cow_color(output_file=cow_color_gif_path, image_size=image_size)
In this case, color1 = [1, 0.5, 0] is saffron and color2 = [0, 0.5, 0] is green.
4. Camera Transformations (20 points)
cow_obj_path = "../data/cow.obj"
cow_transform_img_path = "../images/transform_cow.jpg"
transform 1 - rotate around z axis
R_relative = [[0, 1, 0], [-1, 0, 0], [0, 0, 1]]
T_relative = [0, 0, 0]
output_path = cow_transform_img_path[:-4] + "_Trans_1" + cow_transform_img_path[-4:]
plt.imsave(output_path,
camera_transforms.render_textured_cow(cow_path=cow_obj_path,
image_size=image_size,
R_relative=R_relative,
T_relative=T_relative))
transform 2 - move back - z axis
R_relative = [[0, 1, 0], [-1, 0, 0], [0, 0, 1]]
T_relative = [0, 0, 0]
output_path = cow_transform_img_path[:-4] + "_Trans_2" + cow_transform_img_path[-4:]
plt.imsave(output_path,
camera_transforms.render_textured_cow(cow_path=cow_obj_path,
image_size=image_size,
R_relative=R_relative,
T_relative=T_relative))
transform 3 - slight translation along x and y
R_relative = [[1, 0, 0], [0, 1, 0], [0, 0, 1]]
T_relative = [0.5, -0.5, 0]
output_path = cow_transform_img_path[:-4] + "_Trans_3" + cow_transform_img_path[-4:]
plt.imsave(output_path,
camera_transforms.render_textured_cow(cow_path=cow_obj_path,
image_size=image_size,
R_relative=R_relative,
T_relative=T_relative))
transform 4 - rotate around y axis and shift along x and z axis
R_relative = [[0, 0, 1], [0, 1, 0], [-1, 0, 0]]
T_relative = [-3, 0, 3]
output_path = cow_transform_img_path[:-4] + "_Trans_4" + cow_transform_img_path[-4:]
plt.imsave(output_path,
camera_transforms.render_textured_cow(cow_path=cow_obj_path,
image_size=image_size,
R_relative=R_relative,
T_relative=T_relative))
5. Rendering Generic 3D Representations
5.1 Rendering Point Clouds from RGB-D Images (10 points)
# transform 1 - rotate around z axis
R_relative = [[0, 1, 0], [-1, 0, 0], [0, 0, 1]]
T_relative = [0, 0, 0]
#
output_path = cow_transform_img_path[:-4] + "_Trans_1" + cow_transform_img_path[-4:]
plt.imsave(output_path,
camera_transforms.render_textured_cow(cow_path=cow_obj_path,
image_size=image_size,
R_relative=R_relative,
T_relative=T_relative))
# # transform 2 - move back - z axis
R_relative = [[1, 0, 0], [0, 1, 0], [0, 0, 1]]
T_relative = [0, 0, 2]
output_path = cow_transform_img_path[:-4] + "_Trans_2" + cow_transform_img_path[-4:]
plt.imsave(output_path,
camera_transforms.render_textured_cow(cow_path=cow_obj_path,
image_size=image_size,
R_relative=R_relative,
T_relative=T_relative))
# transform 3 - slight translation along x and y
R_relative = [[1, 0, 0], [0, 1, 0], [0, 0, 1]]
T_relative = [0.5, -0.5, 0]
output_path = cow_transform_img_path[:-4] + "_Trans_3" + cow_transform_img_path[-4:]
plt.imsave(output_path,
camera_transforms.render_textured_cow(cow_path=cow_obj_path,
image_size=image_size,
R_relative=R_relative,
T_relative=T_relative))
# transform 4 - rotate around y axis and shift
R_relative = [[0, 0, 1], [0, 1, 0], [-1, 0, 0]]
T_relative = [-3, 0, 3]
output_path = cow_transform_img_path[:-4] + "_Trans_4" + cow_transform_img_path[-4:]
plt.imsave(output_path,
camera_transforms.render_textured_cow(cow_path=cow_obj_path,
image_size=image_size,
R_relative=R_relative,
T_relative=T_relative))
left, right and union
5.2 Parametric Functions (10 points)
output_torus_file = '../images/torus_param.gif'
num_samples = 200
render_generic.render_torus(image_size=image_size, num_samples=num_samples, device=None, output_file=output_torus_file)
In your writeup, include a 360-degree gif of your torus point cloud, and make sure the hole is visible. You may choose to texture your point cloud however you wish.
5.3 Implicit Surfaces (15 points)
output_torus_file = '../images/torus_impl.gif'
voxel_size = 64
render_generic.render_torus_mesh(image_size=image_size, voxel_size=voxel_size,
device=None, output_file=output_torus_file)
In your writeup, include a 360-degree gif of your torus mesh, and make sure the hole is visible. In addition, discuss some of the tradeoffs between rendering as a mesh vs a point cloud. Things to consider might include rendering speed, rendering quality, ease of use, memory usage, etc.
6. Do Something Fun (10 points)
out_fun_file = "../images/cow_rainbow.gif"
render_shapes.render_cow_rainbow(cow_path=cow_obj_path,
image_size=256,
device=None,
output_file=out_fun_file)
Here I did a texture based on rainbow colors.
(Extra Credit) 7. Sampling Points on Meshes (10 points)
num_samples = 1000
out_cow_pts_path = "../images/cow_pts.gif"
extra_credit.render_cow_ptc(out_path=out_cow_pts_path, image_size=image_size, num_samples=num_samples)
10, 100, 1000, 10000
For this part, write a function that takes a triangle mesh and the number of samples and outputs a point cloud. Then, using the cow mesh, randomly sample 10, 100, 1000, and 10000 points. Render each pointcloud and the original cow mesh side-by-side, and include the gif in your writeup.*