1 / 54

Master Thesis Lighting and materials for real-time game engines

Master Thesis Lighting and materials for real-time game engines. Tobias Kruseborn Supervisors: Per Einarsson, Daniel Johansson. Disposition. Introduction Methods Skin rendering Diffuse and specular reflection , SSS Modified Translucent Shadow Maps Shadows Environment lighting

raymond-hernandez
Download Presentation

Master Thesis Lighting and materials for real-time game engines

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Master ThesisLighting and materials for real-time game engines Tobias Kruseborn Supervisors: Per Einarsson, Daniel Johansson

  2. Disposition • Introduction • Methods • Skin rendering • Diffuse and specularreflection, SSS • Modified Translucent Shadow Maps • Shadows • Environment lighting • Median Cut Algorithm • Spherical Harmonics • Wavelets • Conclusions

  3. Introduction Problem definition: to lighten materials with several layers in a physically correct way and in real time Objective: to study and implement advanced real-time rendering techniques for complex materials Methods: skin rendering and environment lighting techniques

  4. Skin rendering Skin rendering 1. Goal 2. Methods • Diffuse and specularreflection, SSS • Modified Translucent Shadow Maps • Shadows 3. Final algorithm 4. Results

  5. Skin rendering Goal To obtain a look as seenbelow (the speed should be an order of magnitude faster) (Doug Jones demo, NVIDIA)

  6. Skin rendering: Diffuse and specularreflection I Specular reflection • The specular light in human skin only reflects 6% of the whole light spectrum • The oil layer doesn't give out a mirror like reflection • The phong lobe doesn’t give an accurate specular falloff • Kelemen-Szmirnay-Kalos, Cook-Torrance

  7. Skin rendering: Diffuse and specularreflection II Diffuse reflection In order to represent diffuse reflection we need to solve the double integral:

  8. Skin rendering: Diffuse and specularreflection III Diffuse profile The diffuse profile provides an approximation for the manner in which light scatters underneath the surface of a highly scattering translucent material.

  9. Skin rendering: Diffuse and specularreflection IV A sum-of-Gaussians The dipole curve plotted for the diffuse profile, can be approximated by summarizing a number of Gaussians functions

  10. Skin rendering: Diffuse and specularreflection V Implementation of diffuse reflection • Render the illumination to a light map • Perform the Gaussian convolution

  11. Skin rendering: Diffuse and specularreflection VI By carefully choosing a sampling pattern, we can represent each Gaussian blur with a 12 jitter sample: blurJitteredWeights[13] = { { 0.220441, 0.437000, 0.635000 }, { 0.076356, 0.064487, 0.039097 }, { 0.116515, 0.103222, 0.064912 }, { 0.064844, 0.086388, 0.062272 }, { 0.131798, 0.151695, 0.103676 }, { 0.025690, 0.042728, 0.033003 }, { 0.048593, 0.064740, 0.046131 }, { 0.048092, 0.003042, 0.000400 }, { 0.048845, 0.005406, 0.001222 }, { 0.051322, 0.006034, 0.001420 }, { 0.061428, 0.009152, 0.002511 }, { 0.030936, 0.002868, 0.000652 }, { 0.073580, 0.023239, 0.009703 }, }; blurJitteredSamples[13] = { { 0.000000, 0.000000 }, { 1.633992, 0.036795 }, { 0.177801, 1.717593 }, { -0.194906, 0.091094 }, { -0.239737, -0.220217 }, { -0.003530, -0.118219 }, { 1.320107, -0.181542 }, { 5.970690, 0.253378 }, { -1.089250, 4.958349 }, { -4.015465, 4.156699 }, { -4.063099, -4.110150 }, { -0.638605, -6.297663 }, { 2.542348, -3.245901 }, };

  12. Skin rendering: TSM Modified Translucent Shadow Maps How do we capture scattering throughthin regions in the Euclidean space?

  13. Skin rendering: Final algorithm Final Algorithm (1) Use the median cut algorithm to get light source positions(2) convert the diffuse light from an environment map to spherical harmonics(3) for each light(4) render a shadow map and apply Gauss filter to it(5) render the shadows and the diffuse light to the a light map(6) apply SSS to the light map(7) apply SSS for translucency(7) read the diffuse light + shadow from the light map(8) Add the rest of the mesh texture to the diffuse light(8) Calculate the specular light from the same positions as the shadows(9) Combine the specular and diffuse light to a final color

  14. Skin rendering Results

  15. Skin rendering: Specularreflection Kelemen-Szmirnay-Kalos vs Phong

  16. Skin rendering: TSM

  17. Skin rendering: Shadows

  18. Shadows

  19. Environment lighting Environment lighting 1. Goal 2. Why environment lighting is difficult 3. Methods: • Median Cut Algorithm • Spherical Harmonics • Wavelets 4. Results

  20. Environment lighting Goal • To be able to representboth diffuse and specular reflection from an environmentmap in real time.

  21. Why environment lighting is difficult If we represent light with an environment map, all texels in the texture becomes a light source

  22. Environment lighting: Median CutAlgorithm I Median CutAlgorithm • Median Cut Algorithm converts an HDR light probe image to a set of light sources • The algorithmcan represent a complexenvironmentlighting with a fewpointlights

  23. Environment lighting: Median CutAlgorithm II Median CutAlgorithm • Create a SAT with dynamic programming • Add the entire light probe image to the region list • For each region • Subdivide along the longest dimension until the lightenergy is divided xx evenly using a binary search algorithm • For each region • find the centroid using binary search • Calculate the centroids world position

  24. Environment lighting: Sphericalharmonics I Spherical harmonics • Projection of a function into the orthonormal SH basis is simply done by multiplying the integral of the function to the SH basis function • To create an approximation of the signal, is multiplied with the SH basis.

  25. Environment lighting: Sphericalharmonics II The Halo3 method • BRDFExpressiveness Phong/Blinn Cook Torrance, He, etc Project method • Lightingcomplexity Point lights area/env lights • Performance interactive real time [Ramamoorthi02] • storage large minimum mostgamestoday [Kautz02] [Sloan03]

  26. Environment lighting: Sphericalharmonics III • Basis • Separate material into diffuse parts and low and a high freq glossy parts. • SH irradiance env. map for diffuse reflection • New area specular model for low frequency glossy. • BRDF evaluated directly with point lights for high freq. Diffuse Area Specular Analytical Specular

  27. Environment lighting: Sphericalharmonics IV Z Z/N N V Y Y’ V’ X X’ Isotropic BRDF = any coordinate frame 16 m values, and 8 V directions is enough. C (i=0,2,3,6) D (i=0,2,3,6) C,D (i=7,8)

  28. Environment lighting: Wavelets I Wavelets • Wavelets are a set of non-linear bases. • When projecting a function in terms of wavelets, the wavelet basis functions are chosen according to the function being approximated • why use wavelets? Why not non-linear spherical harmonics, or why not just choose pixels in the cubemap non-linearly?

  29. Environment lighting: Wavelets II Waveletsbases

  30. Environment lighting: Wavelets III (1, 1, 1, 1) (1, 1, -1, -1) (1, -1, 0, 0) (0, 0, 1, -1) (1/2, 1/2, 1/2, 1/2) (1/2, 1/2, -1/2, -1/2) (1/√2, -1/√2, 0, 0) (0, 0, 1/ √2, -1/√2) Vector:(4, 2, 5, 5)

  31. Environment lighting: Wavelets IV Soho wavelets • The Haar wavelet lies in the planar domain and leads to distortion when used for functions in other domains • The Soho Wavelets lies in spherical domain

  32. Environment lighting: Wavelets V First implementation • Haar 2D wavelet combined with self occlusion (1) for each vertex(2) draw a visibility cube map and multiply it with a cosine weight(3) tranform the cube map to wavelets and store it in a transport matrix(4) convert the light to wavelets and apply non-linear approximation to it(5) save the coefficient to disc (1) for each vertex(2) multiply the light wavelets with the vertex occlusion wavelets using sparse matrix multiplication

  33. Environment lighting: Wavelets VI Second implementation • Algorithm:Soho approximation algorithm (1) Create the Soho wavelets for the light(2) Sort the coefficents at the finest level and calculate their world position(3) Send the n largest coefficient with its position to the to GPU (This is done in a pre-process)(4) Calculate the BRDF for n position and tranform it to Soho wavelets(5) Multiply the light and the BRDFs wavelets to get the final color

  34. Environment lighting Results

  35. Environment lighting: Median CutAlgorithm

  36. Environment lighting: Halo3 Cook-Torrance Soho Wavelets Kelemen-Szmirnay-Kalos, Soho Wavelets Cook-Torrance Halo3 method

  37. Environment lighting: Diffuse + specularreflection, Soho wavelets 8 16 32 256 512

  38. Conclusions, skin rendering Diffuse reflection: • makes the skin looks more natural • is 10 times faster than the Doug Jones demo • can be used in real-time game, especially for cut scenes Specular reflection - Kelemen-Szmirnay-Kalos and Cook-Torrance : • gave better results that the Phong model • makes the specular reflection more realistic than when using Phong • can catch reflection at grazing angles

  39. Conclusions, skin rendering TSM: • unsatisfying results when implemented with a jitter kernel • if the results would be compared to the extra cost of computing the TSM, the conclusion would be that the modified TSM method was not worth using in the game engine Shadows: • were distinctly improved when applying Gaussian shadow map filter instead of using a uniform shadow map filter • combination of Gaussian shadow maps and subsurface scattering, provided soft shadows without artifacts.

More Related