The advantages / disavantages of Point based approach

The advantages is:

- No noise.
- Faster computation times. (No ray tracing.)
- The geometric primitives do not need to be visible for ray tracing; this can significantly reduce the memory consumed during rendering.
- Color bleeding is nearly as fast as occlusion. (No evaluation of shaders at ray hit points.)
- Computing (HDRI) environment map illumination is just as fast as only computing occlusion.
- Displacement mapped surfaces take no more time than non-displaced surfaces.
- It runs on standard CPUs -- no special hardware required.


The disadvantages are:

- The area (and optionally color) point cloud has to be generated in a pre-pass, making this a two-pass approach.
- During rendering, the entire point cloud has to be read in and kept in memory until the frame is completed.
- The fundamental algorithm tends to compute too much occlusion and color bleeding. This is because occlusion (and colors) are simply added up, even though some geometry really should block other geometry. (The "clamp" parameter eliminates most of the over-occlusion and over-bleeding. Nevertheless, experimenting with various parameter settings might be necessary to get the desired look.)