Zheren_Zhu
vertices = torch.tensor([[0, -0.5, -1.0], [0.866, -0.5, 0.5], [-0.866, -0.5, 0.5], [0, 1.0, 0]])
faces = torch.tensor([[0, 1, 3], [1, 2, 3], [0, 2, 3], [0, 1, 2]])
Vertices: 4 Faces: 4
vertices = torch.tensor([[0, -0.5, 0], [1, -0.5, 0], [1, 0.5, 0], [0, 0.5, 0], [0, 0.5, 1], [1, 0.5, 1], [1, -0.5, 1], [0, -0.5, 1]])
faces = torch.tensor([[0, 2, 1], [0, 3, 2], [2, 3, 4], [2, 4, 5], [1, 2, 5], [1, 5, 6], [0, 7, 4], [0, 4, 3], [5, 4, 7], [5, 7, 6], [0, 6, 7], [0, 1, 6]])
Vertices: 8 Faces: 12
color: blue to red
color1 = [0, 0, 1] color2 = [1, 0, 0]
Case1:
rotate the camera -90 degrees along the z-axis
set R_relative to R_relative = [[0, 1, 0], [-1, 0, 0], [0, 0, 1]], thus x_new = -y; y_new = x
no change in distance between camera and object: T_relative=[0, 0, 0]
Case2:
translate the camera 2 units away from the object along the z-axis
set T_relative to T_relative = [0, 0, 2]
the relative rotation remains R_relative = [[1, 0, 0], [0, 1, 0], [0, 0, 1]]
Case3:
rotate the camera 90 degrees along the y-axis: R_r = [[0, 0, 1], [0, 1, 0], [-1, 0, 0]]
move the camera: T_r = [-3, 0, 3]
Case4:
rotate the camera 10 degrees along the y-axis: R_r = R_r=[[np.cos(a2r(10)), 0, np.sin(a2r(10))], [0, 1, 0], [-np.sin(a2r(10)), 0, np.cos(a2r(10))]]
move the camera up-left and close to the object: T_r = [-0.2, -0.4, -0.1]
The point cloud corresponding to the first image
The point cloud corresponding to the second image
The point cloud formed by the union of the first 2 point clouds.
Speed: rendering a point-cloud is much faster than rendering a mesh. Though the rendering speed still depends on number of samples for point-cloud and mesh quality for mesh-rendering, point-cloud may still run faster because there's no need of connectivity info
Quality: The mesh-rendering is better for its connectivity
Point cloud is easier in use and consumes less memory
Pinky Ball
