mm package

Subpackages

• mm.optimize package
  • Submodules
  • mm.optimize.camera module
  • mm.optimize.depth module
  • mm.optimize.derivative module
  • mm.optimize.image module
  • Module contents
• mm.utils package
  • Submodules
  • mm.utils.io module
  • mm.utils.mesh module
  • mm.utils.opengl module
  • mm.utils.transform module
  • mm.utils.visualize module
  • Module contents

Submodules

mm.deprecated module

mm.deprecated.calcZBuffer(vertexCoord)

Assumes that the nose will have smaller z values

mm.deprecated.gaussNewton(P, model, target, targetLandmarks, sourceLandmarkInds, NN, jacobi=True, calcId=True)

Energy function to be minimized for fitting.

mm.deprecated.getTexture(dirName, pose=0)

Get the RGB values referenced by the texture vertices in the .obj file for each tester. There are 20 poses to choose from from 0 to 19.

mm.deprecated.shBasis(alb, n)

SH basis functions lm

1/np.sqrt(4*np.pi) Y00 np.sqrt(3/(4*np.pi))*nz Y10 np.sqrt(3/(4*np.pi))*nx Y11e np.sqrt(3/(4*np.pi))*ny Y11o 1/2*np.sqrt(5/(4*np.pi))*(3*nz^2 - 1) Y20 3*np.sqrt(5/(12*np.pi))*nx*nz Y21e 3*np.sqrt(5/(12*np.pi))*ny*nz Y21o 3/2*np.sqrt(5/(12*np.pi))*(nx^2 - ny^2) Y22e 3*np.sqrt(5/(12*np.pi))*nx*ny Y22o

For a sphere, the Lambertian kernel has most of its energy in the first three bands of the spherical harmonic basis functions (above). This implies that Lambertian reflectance functions can be well-approximated by these low-order SH bases.

mm.deprecated.textureCostV(texCoef, x, mask, model, w=(1, 1))

Energy formulation for fitting texture

mm.deprecated.textureGradV(texCoef, x, mask, model, w=(1, 1))

Jacobian for texture energy

mm.deprecated.textureLightingCostV(texParam, x, mask, B, model, w=(1, 1), option='tl', constCoef=None)

Energy formulation for fitting texture and spherical harmonic lighting coefficients

mm.deprecated.textureLightingGradV(texParam, x, mask, B, model, w=(1, 1), option='tl', constCoef=None)

Jacobian for texture and spherical harmonic lighting coefficients

mm.models module

class mm.models.MeshModel(modelFile, numIdEvecs=80, numExpEvecs=76, numTexEvecs=80)

Bases: object

A 3D Morphable Model class object

Parameters:

• modelFile (str) – Filename of .npz file containing 3DMM
• numIdEvecs (int) – Number of the shape identity eigenvectors with the highest eigenvalues from the 3DMM to keep
• numExpEvecs (int) – Number of the shape facial expression eigenvectors with the highest eigenvalues from the 3DMM to keep
• numTexEvecs (int) – Number of the texture eigenvectors with the highest eigenvalues from the 3DMM to keep

numId

int – number of shape identity eigenvectors

numExp

int – number of shape facial expression eigenvectors

numTex

int – number of texture eigenvectors

numVertices

int – number of vertices in the 3DMM

numFaces

int – number of triangular faces in the 3DMM

face

ndarray – array containing the vertex indices for each face, (numFaces, 3)

vertex2face

ndarray – array containing the face index of each vertex, (numVertices,)

idMean

ndarray – shape identity mean, (3, numVertices)

idEvec

ndarray – shape identity eigenvectors, (3, numVertices, numId)

idEval

ndarray – shape identity eigenvalues, (numId,)

expEvec

ndarray – shape facial expression eigenvectors, (3, numVertices, numExp)

expEval

ndarray – shape facial expression eigenvalues, (numExp)

texMean

ndarray – texture mean, (3, numVertices)

texEvec

ndarray – texture eigenvectors, (3, numVertices, numTex)

texEval

ndarray – texture eigenvalues, (numTex,)

targetLMInd

ndarray – landmark indices for OpenPose

sourceLMInd

ndarray – vertex indices of the 3DMM that correspond to targetLMInd

Module contents