R_relative and T_relative define how the camera moves between frames by adjusting its pose relative to its current position and orientation. Specifically, R_relative is a rotation matrix that incrementally rotates the camera around the object (changing where it looks from), while T_relative is a translation vector that shifts the camera’s position relative to its local axes (changing how close or far it is).
Image 1:
R_relative = [[0, 1, 0],
[-1, 0, 0],
[0, 0, 1]],
T_relative = [0, 0, 0],
Image 2:
R_relative=[[1, 0, 0],
[0, 1, 0],
[0, 0, 1]],
T_relative=[0, 0, 2],
Image 3:
R_relative=[[1, 0, 0],
[0, 1, 0],
[0, 0, 1]],
T_relative=[0.5, -0.5, 0],
Image 4:
R_relative=[[0, 0, 1],
[0, 1, 0],
[-1, 0, 0]],
T_relative=[-3.0, 0.0, 3.0],
Render from camera 1 on the left, camera 2 at the center and combined cameras at the right.
Meshes give smooth surfaces with lighting and textures, making them look more realistic, but they’re heavier on memory and slower to render. Point clouds are faster and lighter since they’re just unconnected points, but they don’t show surfaces well and can look sparse or noisy. Point clouds are easy to generate, while meshes take more work but give much better visual quality.
