Global Illumination. COMP 575/770 Spring 2013

Transcription

1 Global Illumination COMP 575/770 Spring 2013

2 Final Exam and Projects COMP 575 Final Exam Friday, May 3 4:00 pm COMP 770 (and 575 extra credit) Projects Final report due by end of day, May 1 Presentations: May 2, 9:30 am 12:00pm minute slots

3 Global Illumination Simulation of light transport Or, how light interacts with objects in a scene Including light bouncing between objects Can lead to many complex, subtle effects

4 Local Shading Model Simple analytic model: Diffuse reflection Specular reflection Ambient light Emission Reflections Refractions Shadows

5 Local Shading Model Light transport simulated from: Light to surface point Surface point to camera Special cases: Reflection: recursive ray Shadows: shadow ray to light source

6 Global Shading Model Light transport simulated between: Light to surface point Surface point to surface point Surface points to camera Surface to surface transport: How many surfaces to consider? How many bounces to simulate? What kinds of bounces? (Specular, scattering, etc.)

7 Indirect Light

8 Color Bleeding

9 Caustics

10 Caustics

11 Surface-to-Surface Transport Mirror-like reflections Purely specular surface to diffuse surface Refractions Translucent surface to diffuse surface Diffuse inter-reflections Diffuse surface to diffuse surface

12 Diffuse Inter-reflections

13 Radiosity An algorithm for simulating diffuse interreflections No specular reflection or refraction Handles any number of diffuse bounces of light Solution depends only on light position And not on camera position

14 Radiosity What is radiosity? The amount of light energy leaving a surface Lots of complex math behind all this Covered in more advanced courses Divide the scene into surface patches Find the radiosity at each patch

15 Radiosity Assume: Surface patches are purely diffuse, opaque Surfaces don t move Scene is in a vacuum Radiosity of one patch depends on radiosity of every other patch Build a huge system of linear equations Radiosity at each patch is a variable Coefficients form a large matrix

16 Radiosity

17 Radiosity Examples

18 Radiosity Examples

19 Radiosity Examples

20 Radiosity Examples

21 Progressive Radiosity Solve for radiosity after each bounce of diffuse inter-reflections

22 Path Tracing How can we model diffuse inter-reflections in a recursive ray tracer? A diffuse surface reflects light in all directions What other phenomena involve light from multiple directions? Soft shadows Ambient occlusion

23 Path Tracing When a ray hits a diffuse surface, trace rays in all directions Sample a hemisphere of directions at the hit point Alternatively: Sample a single reflected direction at random And sample multiple rays per pixel Each ray sample results in a different reflected direction This approach is called path tracing

24 Path Tracing Need lots of samples per pixel (50+) Samples are chosen based on probability distributions Which in turn depend on surface properties Can simulate any kind of surface Diffuse, specular, glossy, etc.

25 Path Tracing Examples

26 Path Tracing Examples

27 Path Tracing

28 Path Tracing Best for large lights If the lights are small, few rays hit them causing the image to look noisy

29 Path Tracing + Shadow Rays

30 Path Tracing + Shadow Rays Works well for specular surfaces

31 Light Tracing

32 Bidirectional Path Tracing

33 Bidirectional Path Tracing Light Tracing Bi-Directional Path Tracing Path Tracing

34 Bidirectional Path Tracing Light Tracing Bi-Directional Path Tracing Path Tracing

35 Bidirectional Path Tracing

36 Metropolis Light Transport Named after the physicist Nicholas Metropolis Nothing to do with cities, public transit, or superheroes Path tracing needs lots of samples in certain cases Small lights Glossy surfaces If you find such a path, look for other similar ones nearby Sample more paths near important paths

37 Metropolis Light Transport All the light in the room comes in through a crack in the doorway.

38 Metropolis Light Transport Bi-Directional Path Tracing Metropolis Light Transport

39 Photon Mapping A two-pass algorithm First pass: construct a photon map Trace photons from light Reflect them off surfaces (diffuse or specular) Store the photons at surfaces (probabilistically) Second pass: render from photon map Trace rays from camera Collect photons from photon map at the hit point Compute lighting at the surface

40 Photon Mapping

41 Photon Mapping

42 Precomputed Radiance Transfer Somewhat similar to radiosity But stores interactions between each pair of patches Allows lights to move in real-time! Used in many top-quality games

43 Precomputed Radiance Transfer

44 Precomputed Radiance Transfer Two step process Precomputation Derive a matrix representing the optical properties of a scene Runtime Use the matrix to quickly perform global illumination