private void recalcOldDiff(int c)
{
gradValues grad = new gradValues();
grad.a = A[c];
grad.b = B[c];
grad.c = C[c];
grad.d = D[c];
setGradDiff(c, grad);
oldDiff[c] = grad.diff;
}
private void setGradDiff(int c, gradValues grad)
{
grad.diff = 0f;
for (int s = 0; s < sampleCount; s++)
{
float point = ( (float)s + ( jitter ? Random.value : .5f ) ) / (float)sampleCount;
float sVal = gradients.GetPixelBilinear(point, gradientPosition)[c];
float gVal = 0f;
switch (gradientFunction)
{
case gFuncs.trigonometric: gVal = getTrigGradValue( point, grad.a, grad.b, grad.c, grad.d ); break;
case gFuncs.asymetricPolynomial: gVal = getAsymetricPolyGradValue( point, grad.a, grad.b, grad.c, grad.d ); break;
case gFuncs.symetricPolynomial: gVal = getSymetricPolyGradValue( point, grad.a, grad.b, grad.c, grad.d ); break;
case gFuncs.exponential: gVal = getExponentialGradValue( point, grad.a, grad.b, grad.c, grad.d ); break;
}
if(clamp) gVal = Mathf.Clamp01(gVal);
grad.diff += Mathf.Abs(sVal - gVal);
}
grad.diff /= (float)sampleCount;
}
private void stepChannel(int c)
{
List<gradValues> grads = new List<gradValues>();
for (int v = 0; v < variationCount; v++)
{
gradValues grad = new gradValues();
grad.a = A[c] + Random.Range( -oldDiff[c]/1.5f, oldDiff[c]/1.5f );
grad.b = B[c] + Random.Range( -oldDiff[c]/1.5f, oldDiff[c]/1.5f );
grad.c = C[c] + Random.Range( -oldDiff[c]/5.0f, oldDiff[c]/5.0f );
grad.d = D[c] + Random.Range( -oldDiff[c]/1.5f, oldDiff[c]/1.5f );
setGradDiff(c, grad);
grads.Add(grad);
}
grads.Sort((a,b)=>a.diff.CompareTo(b.diff));
if (grads[0].diff < oldDiff[c])
{
oldDiff[c] = grads[0].diff;
A[c] = grads[0].a;
B[c] = grads[0].b;
C[c] = grads[0].c;
D[c] = grads[0].d;
}
}
