Post on 19-Jan-2016
description
Lighting modelsLighting models& optimization& optimizationPavel ZemčíkPavel ZemčíkDepartment of Computer Science and Engineering,Department of Computer Science and Engineering,
Faculty of Electrical Engineering and Computer Faculty of Electrical Engineering and Computer Science,Science,
Technical Univeristy of Brno,Technical Univeristy of Brno,
Czech RepublicCzech Republic
zemcik@dcse.fee.vutbr.czzemcik@dcse.fee.vutbr.cz
What are lighting models?What are lighting models?
Lighting models are models of light Lighting models are models of light behaviour on the objectsbehaviour on the objects’ surface’ surface
• Global modelsGlobal models
• Local modelsLocal models
• Behaviour at the edgesBehaviour at the edges
Global modelsGlobal models
Generally used in radiation methodsGenerally used in radiation methods
• Physics laws (preservs model’s Physics laws (preservs model’s energy)energy)
• Simple light propagation (form Simple light propagation (form factors)factors)
Local modelsLocal models
Approximate light propagation Approximate light propagation locallylocally
• Not necessarily physics basedNot necessarily physics based
• Measurement based (empirical)Measurement based (empirical)
• Often just a Often just a ‘good looking’ ‘good looking’ guessguess
Local models geometryLocal models geometry
Phong model geometryPhong model geometry
Phong model equationsPhong model equations
• GeneralGeneral
• Details (note that IDetails (note that Ibb is constant) is constant)
rdb IIII
)(cos nl dldld kk III n
rln
rlr kk )(cos re III nnllr )(2
Phong model parameters-Phong model parameters-nn
• The image showsThe image showsthe effect of the effect of nn
nn== 15 7 15 7 3 1 3 1
nnrlrlr kk )(cos re III
Phong model parameters-Phong model parameters-kkdd,k,krr
• The image showsThe image showsthe effect of the effect of kkdd,k,krr
kkdd,k,krr = = 0.3,0.6 0.5,0.4 0.3,0.6 0.5,0.4 0.7,0.2 0.9,0 0.7,0.2 0.9,0
nrl
nrlr kk )(cos re III
)(cos nl dldld kk III
Surface textureSurface texture
• The image wasThe image wasrendered usingrendered usingPhong model withPhong model withsuperimposedsuperimposedtexture modifyingtexture modifyingthe coefficientsthe coefficients
Normal vector textureNormal vector texture
• The image wasThe image wasrendered usingrendered usingPhong model withPhong model withsuperimposedsuperimposedtexture modifyingtexture modifyingthe normal vectorthe normal vector
Mirror model geometryMirror model geometry
Mirror model equationsMirror model equations
• GeneralGeneral
• Details (reflection direction)Details (reflection direction)
rrk II
nneer )(2
Mirror model exampleMirror model example
Glass model geometryGlass model geometry
Glass model energyGlass model energy
Glass model equationsGlass model equations
• DirectionDirection
• Energy distribution approximationEnergy distribution approximation
2211 sinsin nn
4)90/(1.0 R
Glass model exampleGlass model example
Other lighting modelsOther lighting models
• Torrance-Sparrow (rough surfaces)Torrance-Sparrow (rough surfaces)
• Blinn, Strauss (half-transparent Blinn, Strauss (half-transparent objects)objects)
• Metals (fluorescent effects)Metals (fluorescent effects)
• etc.etc.
Optimisation of ray Optimisation of ray tracingtracing
Optimisation of ray Optimisation of ray tracingtracing
Three general approachesThree general approaches
• Reduction of number of evaluated Reduction of number of evaluated pixelspixels
• Bounding volumesBounding volumes
• Space subdivisionSpace subdivision
Reduction of number of Reduction of number of pixelspixels
Mostly using adaptive subsamplingMostly using adaptive subsampling
• naive would benaive would be
40x3040x30
• optimised 5x5optimised 5x5
16x12+10x2116x12+10x21
402/1200402/1200-66%-66%
Bounding volumesBounding volumes
Bounding volume must bound all the Bounding volume must bound all the real objects but be as small as real objects but be as small as possiblepossible
• complexcomplex
• done manuallydone manually
• speedup >10speedup >10
Space subdivisionSpace subdivision
Scene is divided into several smaller Scene is divided into several smaller units that are evaluated separatelyunits that are evaluated separately
• simple principlesimple principle
• automatic butautomatic butquite high costquite high cost
• speedup >10speedup >10
CSG tree pruningCSG tree pruning
CSG status treeCSG status tree
Referenecs Referenecs (in addition to the previous (in addition to the previous lecture)lecture)
• Bronsvoort W F: Techniques for Reducing Boolean Evaluation Time in CSG scan-line algorithms, Computer-aided Design, vol. 18, no. 10, 1986, Great Britain, pp. 533-536
• Fujimoto A, Tanaka T, Iwata K: ARTS Accelerated Ray Tracing System, IEEE Computer Graphics & Applications, April 1986, USA, pp. 16-26
• Glassner A S: Efficient Boolean Evaluation of CSG Models for Ray Tracing, The Ray Tracing News, vol. 1, no. 1, September 1987, USA, pp. 3-7
• Strauss P S: A Realistic Model for Computer Animators, IEEE Computer Graphics & Applications, November 1990, USA, pp. 56-64
The endThe end