private GaussianBlurArrayComponent[] GenerateForASampleSize(int num)
{
//In the filter we take advantage of 'free' gpu linear filtering optimiser, to create same effect of a larger sample set.
//We need to calculate the weights and offsets of the larger set first
var numSamplesOnOneSideIncludingCentre = num + 1;
var expandSampleSize = ((numSamplesOnOneSideIncludingCentre - 1) * 2) + 1;
var expandedWeightsAndOffsets = CalculateWeightsAndOffsetsOfLargerEquivalentSet(expandSampleSize);
var weightsAndOffsets = new GaussianBlurArrayComponent[SAMPLE_ARRAY_SIZE];
weightsAndOffsets[0] = expandedWeightsAndOffsets[0];
for (var n = 1; n < numSamplesOnOneSideIncludingCentre; n++)
{
var expandedIndex0 = 1 + ((n - 1) * 2);
var expandedIndex1 = expandedIndex0 + 1;
weightsAndOffsets[n].Weight = expandedWeightsAndOffsets[expandedIndex0].Weight + expandedWeightsAndOffsets[expandedIndex1].Weight;
weightsAndOffsets[n].Offset = ((expandedWeightsAndOffsets[expandedIndex0].Weight * expandedWeightsAndOffsets[expandedIndex0].Offset) +
(expandedWeightsAndOffsets[expandedIndex1].Weight * expandedWeightsAndOffsets[expandedIndex1].Offset)) / weightsAndOffsets[n].Weight;
}
return(weightsAndOffsets);
}
private GaussianBlurArrayComponent[] CalculateWeightsAndOffsetsOfLargerEquivalentSet(int numSamples)
{
var tapNumber = (2 * (numSamples - 1)) + 1;
var PascalTriangleRowIndexToUse = tapNumber + 3;
//We use the row minus the two edge numbers on each side
var PascalsTriangleRowSum = (float)Math.Pow(2.0f, PascalTriangleRowIndexToUse);
var firstValue = 1;
var secondValue = CalculateValueInPascalTriangle(PascalTriangleRowIndexToUse, 1);
var adjustedRowSum = PascalsTriangleRowSum - (2 * (firstValue + secondValue));
var halfRow = new float[numSamples];
for (var n = 0; n < numSamples; n++)
{
var index = n + 2; // we skip the first two values in the pascals triangle
halfRow[n] = CalculateValueInPascalTriangle(PascalTriangleRowIndexToUse, index);
}
var weightsAndOffsets = new GaussianBlurArrayComponent[numSamples];
for (var n = 0; n < numSamples; n++)
{
weightsAndOffsets[n].Offset = n;
var reverse = numSamples - n - 1;
weightsAndOffsets[reverse].Weight = (1.0f * halfRow[n]) / (1.0f * adjustedRowSum);
}
return(weightsAndOffsets);
}