Readings I Did
- Approximating Dynamic Global Illumination in Screen Space
- Read the paper.
- SSDO (SSAO but with directional shadows and color bleeding) can be done without adding much computational complexity to SSAO. This allows for an effect that better simulates global illumination in real time.
- SSDO is applied as a post process (or series of post processes) where samples are taken of surrounding pixels to determine occlusion and lighting bounces.
- SSDO can be enhanced by shadow maps and geometry-sensitive blurs to allow for shadows from distant geometry.
- SSDO is sensitive to a max sampling radius that is hard to set automatically, but can be estimated as the average distance between two objects in a scene.
- Storing multiple depth values for a pixel or using multiple cameras can enhance the effect by reducing false occluders or illuminators.
- Being an approximation, discrepancies can occur.
- Section 10.9 - Image Processing in RTR 3rd Edition
- Read the first few paragraphs, glanced at the example effects
- Explanation of how to do post processing was given - render the scene to a texture, place that texture on a screen-filling quadrilateral, then render that quadrilateral to do another pass. Repeat for as many post processes are desired.
Readings I Want to Look At
- Something on environment mapping
- Something on the SSDO improvements above (shadow maps and blurs)