Camera update delay

[ALEX95GOGO]

I am moving my camera along a cylindrical trajectory to take the picture, but I found that I have to call sim.step() multiple times (or set the decimation to 5), otherwise the picture of my camera will be delayed 1 or two steps. Any ideas why this happen or how I should set the decimation value? The following is my code snippet:

env_size = 20
radius = env_size/2.0 * 1.5 # Radius of the cylinder
height = env_size/2.0 * 2. # Height of the cylinder
theta = np.linspace(0,6*np.pi, 200)
num_points = len(theta)
x = radius * np.cos(theta)
y = radius * np.sin(theta)
z = np.linspace(2, height, num_points)
index = 0
grid = OccupancyGrid(20, (1,20,20,20))
# Simulate physics
while simulation_app.is_running():
    # Update camera data
    target_position = np.array([x[index], y[index], z[index]])
    target_position = torch.from_numpy(target_position).unsqueeze(0)
    target_orientation = torch.from_numpy(target_orientation).unsqueeze(0)
    camera.set_world_poses_from_view(target_position.float(), torch.tensor([0,0,2]).unsqueeze(0).float())

    # Step simulation
    sim.step()
    sim.step()
    sim.step()
    sim.step()
    sim.step()
    sim.step()
    camera.update(dt=sim.get_physics_dt())