SphericalHarmonicsLightProposals

Type Members

1. case class RobustSHLightSolverProposal(modelRenderer: ParametricImageRenderer[RGBA] with ParametricModel, shOpt: SphericalHarmonicsOptimizer, target: PixelImage[RGBA], pixelEvaluator: PairEvaluator[RGB] = ..., nSamples: Int = 30, sigmaThreshold: Double = 2, percentage: Double = 0.4, iterations: Int = 500, nSamplesIllumination: Int = 1000)(implicit rnd: Random) extends ProposalGenerator[RenderParameter] with TransitionProbability[RenderParameter] with Product with Serializable

   

   This proposal performs RANSAC to try to estimate illumination excluding outliers/occlusions The face model is used as appearance prior

   This proposal performs RANSAC to try to estimate illumination excluding outliers/occlusions The face model is used as appearance prior

   The procedure is explained in: Occlusion-aware 3D Morphable Face Models, Bernhard Egger, Andreas Schneider, Clemens Blumer, Andreas Morel-Forster, Sandro Schönborn, Thomas Vetter IN: British Machine Vision Conference (BMVC), September 2016 https://dx.doi.org/10.5244/C.30.64

2. case class RobustSHLightSolverProposalWithLabel(modelRenderer: ParametricImageRenderer[RGBA] with ParametricModel, shOpt: SphericalHarmonicsOptimizer, target: PixelImage[RGBA], pixelEvaluator: PairEvaluator[RGB] = ..., nSamples: Int = 30, sigmaThreshold: Double = 2, percentage: Double = 0.4, iterations: Int = 500, nSamplesIllumination: Int = 1000)(implicit rnd: Random) extends ProposalGenerator[(RenderParameter, PixelImage[Int])] with TransitionProbability[(RenderParameter, PixelImage[Int])] with Product with Serializable

   

   This proposal performs RANSAC to try to estimate illumination excluding outliers/occlusions The face model is used as appearance prior

   This proposal performs RANSAC to try to estimate illumination excluding outliers/occlusions The face model is used as appearance prior

   The procedure is explained in: Occlusion-aware 3D Morphable Face Models, Bernhard Egger, Andreas Schneider, Clemens Blumer, Andreas Morel-Forster, Sandro Schönborn, Thomas Vetter IN: British Machine Vision Conference (BMVC), September 2016 https://dx.doi.org/10.5244/C.30.64

3. case class SHLightBandEnergyMixer(logSdev: Double)(implicit rnd: Random) extends ProposalGenerator[SphericalHarmonicsLight] with TransitionProbability[SphericalHarmonicsLight] with Product with Serializable

   

   Randomly changes the light intensity of the shl bands according to std of a normal distribution.

   Randomly changes the light intensity of the shl bands according to std of a normal distribution.

   logSdev

   log of expected factor of variation for each component

4. case class SHLightColorProposal(sdev: Double)(implicit rnd: Random) extends ProposalGenerator[SphericalHarmonicsLight] with SymmetricTransitionRatio[SphericalHarmonicsLight] with TransitionProbability[SphericalHarmonicsLight] with Product with Serializable

   

   vary the color distribution of the illumination while keeping the intensity constant

5. case class SHLightIntensityPreservingProposal(proposalGenerator: ProposalGenerator[SphericalHarmonicsLight] with TransitionProbability[SphericalHarmonicsLight])(implicit rnd: Random) extends ProposalGenerator[SphericalHarmonicsLight] with TransitionProbability[SphericalHarmonicsLight] with Product with Serializable

   

   Draw from a proposalGenerator in a way such that it has the same illumination intensity as a given sample.

   Draw from a proposalGenerator in a way such that it has the same illumination intensity as a given sample. We adjust the intensity of a sample drawn from proposalGenerator, so that it has the same intensity as the given proposal.

   proposalGenerator

   underlying proposal generator to use, proposal of it will get a rescaled light

6. case class SHLightIntensityProposal(logSdev: Double)(implicit rnd: Random) extends ProposalGenerator[SphericalHarmonicsLight] with TransitionProbability[SphericalHarmonicsLight] with Product with Serializable

   

   vary the light intensity

   vary the light intensity

   logSdev

   log of expected variation factor

7. case class SHLightPerturbationProposal(sdev: Double, fixIntensity: Boolean)(implicit rnd: Random) extends ProposalGenerator[SphericalHarmonicsLight] with SymmetricTransitionRatio[SphericalHarmonicsLight] with TransitionProbability[SphericalHarmonicsLight] with Product with Serializable

   

   independent Gaussian perturbation of the SH illumination parameter

   independent Gaussian perturbation of the SH illumination parameter

   sdev

   standard deviation of proposal per component

   fixIntensity

   if true the intensity is perserved

8. case class SHLightSolverProposal(sphericalHarmonicsOptimizer: SphericalHarmonicsOptimizer, samplingFunction: (TriangleMesh3D) ⇒ IndexedSeq[(TriangleId, BarycentricCoordinates)])(implicit rnd: Random) extends ProposalGenerator[RenderParameter] with TransitionProbability[RenderParameter] with Product with Serializable

   

   use the linear solver to find the best illumination parameters

9. case class SHLightSpatialPerturbation(sdev: Double)(implicit rnd: Random) extends ProposalGenerator[SphericalHarmonicsLight] with TransitionProbability[SphericalHarmonicsLight] with Product with Serializable

   

   Perturbs the shl coefficients, without changing the color.