mirror of
https://github.com/TomHarte/CLK.git
synced 2025-01-13 22:32:03 +00:00
More definitively removed the old sample-offset blending approach to filtering.
This commit is contained in:
parent
163c0f1b44
commit
ef03c84b21
@ -331,7 +331,7 @@ void IntermediateShader::set_filter_coefficients(float sampling_rate, float cuto
|
||||
GLfloat offsets[5];
|
||||
unsigned int taps = 11;
|
||||
// unsigned int taps = 21;
|
||||
while(1) {
|
||||
// while(1) {
|
||||
float coefficients[21];
|
||||
SignalProcessing::FIRFilter luminance_filter(taps, sampling_rate, 0.0f, cutoff_frequency, SignalProcessing::FIRFilter::DefaultAttenuation);
|
||||
luminance_filter.get_coefficients(coefficients);
|
||||
@ -341,12 +341,12 @@ void IntermediateShader::set_filter_coefficients(float sampling_rate, float cuto
|
||||
memset(weights, 0, sizeof(float)*12);
|
||||
memset(offsets, 0, sizeof(float)*5);
|
||||
|
||||
int halfSize = (taps >> 1);
|
||||
for(int c = 0; c < taps; c++) {
|
||||
if(c < 5) offsets[c] = (halfSize - c);
|
||||
unsigned int half_size = (taps >> 1);
|
||||
for(unsigned int c = 0; c < taps; c++) {
|
||||
if(c < 5) offsets[c] = (half_size - c);
|
||||
weights[c] = coefficients[c];
|
||||
}
|
||||
break;
|
||||
// break;
|
||||
|
||||
// int halfSize = (taps >> 1);
|
||||
// while(c < halfSize && sample < 5) {
|
||||
@ -368,8 +368,8 @@ void IntermediateShader::set_filter_coefficients(float sampling_rate, float cuto
|
||||
// }
|
||||
// break;
|
||||
// }
|
||||
taps -= 2;
|
||||
}
|
||||
// taps -= 2;
|
||||
// }
|
||||
|
||||
set_uniform("weights", 4, 3, weights);
|
||||
set_uniform("offsets", 1, 5, offsets);
|
||||
|
Loading…
x
Reference in New Issue
Block a user