Surface Space Ambient Occlusion

Demonstrate the usage of surface space ambient occlusion.

Surface space ambient occlusion (SSAO) can approximate shadows more efficiently than ray-tracing and produce similar results. Use this when you wish to plot the occlusion effect that nearby meshes have on each other by blocking nearby light sources.

See Kitware: Screen-Space Ambient Occlusion for more details

# First, let's create several cubes nearby each other

import pyvista as pv
from pyvista import examples

grid = pv.ImageData(dimensions=(5, 5, 5)).explode(0.2)

Plot with defaults

Let’s plot this without SSAO. Note how the lighting is identical for each cube.

pl = pv.Plotter()
pl.add_mesh(grid)
pl.show()

Plot with SSAO

Now plot this with SSAO. Note how adjacent cubes affect the lighting of each other to make it look less artificial.

With a low kernel_size, the image will be rendered quickly at the expense of quality.

pl = pv.Plotter()
pl.add_mesh(grid)
pl.enable_ssao(kernel_size=32)
pl.show()

Improve the SSAO rendering

Here we’ve increased the kernel_size to improve the quality of the SSAO and also enabled SSAA anti-aliasing to smooth out any of the artifacts created from SSAO.

pl = pv.Plotter()
pl.add_mesh(grid)
pl.enable_ssao(kernel_size=128)
pl.enable_anti_aliasing('ssaa')
pl.show()

Plot a CAD model without SSAO

Here’s another example without SSAO. This is a CAD model of a Raspberry PI case.

mesh = examples.download_cad_model_case()

pl = pv.Plotter()
pl.add_mesh(mesh, smooth_shading=True, split_sharp_edges=True)
pl.enable_anti_aliasing('ssaa')
pl.camera.zoom(1.7)
pl.show()

Plot with SSAO

Here’s the same CAD model with SSAO. Note how we had to increase both radius and bias due to the relative scene size.

Note that the occlusion still seems quite small. In the next example we will increase the radius to increase the effect of the occlusion.

pl = pv.Plotter()
pl.add_mesh(mesh, smooth_shading=True, split_sharp_edges=True)
pl.enable_ssao(radius=2, bias=0.5)
pl.enable_anti_aliasing('ssaa')
pl.camera.zoom(1.7)
pl.show()

Increase the Radius

Here we’ve increased the radius to the point where the case occlusion now seems realistic without it becoming overwhelming.

pl = pv.Plotter()
pl.add_mesh(mesh, smooth_shading=True, split_sharp_edges=True)
pl.enable_ssao(radius=15, bias=0.5)
pl.enable_anti_aliasing('ssaa')
pl.camera.zoom(1.7)
pl.show()