Skip to content

Commit

Permalink
Merge pull request #423 from ferdnyc/optimized-blur
Browse files Browse the repository at this point in the history 
Optimized, parallelized Blur effect
  • Loading branch information
@jonoomph
jonoomph authored Feb 27, 2020
2 parents f9b4fe5 + fd663c4 commit 6666702
Show file tree
Hide file tree
Showing 2 changed files with 61 additions and 145 deletions.
1 change: 0 additions & 1 deletion include/effects/Blur.h
View file
Original file line number Diff line number Diff line change
Expand Up @@ -67,7 +67,6 @@ namespace openshot
void init_effect_details();

/// Internal blur methods (inspired and credited to http://blog.ivank.net/fastest-gaussian-blur.html)
int* initBoxes(float sigma, int n);
void boxBlurH(unsigned char *scl, unsigned char *tcl, int w, int h, int r);
void boxBlurT(unsigned char *scl, unsigned char *tcl, int w, int h, int r);

Expand Down
205 changes: 61 additions & 144 deletions src/effects/Blur.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -74,179 +74,96 @@ std::shared_ptr<Frame> Blur::GetFrame(std::shared_ptr<Frame> frame, int64_t fram
float sigma_value = sigma.GetValue(frame_number);
int iteration_value = iterations.GetInt(frame_number);

int w = frame_image->width();
int h = frame_image->height();

// Declare arrays for each color channel
unsigned char *red = new unsigned char[frame_image->width() * frame_image->height()]();
unsigned char *green = new unsigned char[frame_image->width() * frame_image->height()]();
unsigned char *blue = new unsigned char[frame_image->width() * frame_image->height()]();
unsigned char *alpha = new unsigned char[frame_image->width() * frame_image->height()]();
// Create empty target RGBA arrays (for the results of our blur)
unsigned char *blur_red = new unsigned char[frame_image->width() * frame_image->height()]();
unsigned char *blur_green = new unsigned char[frame_image->width() * frame_image->height()]();
unsigned char *blur_blue = new unsigned char[frame_image->width() * frame_image->height()]();
unsigned char *blur_alpha = new unsigned char[frame_image->width() * frame_image->height()]();

// Loop through pixels and split RGBA channels into separate arrays
unsigned char *pixels = (unsigned char *) frame_image->bits();
for (int pixel = 0, byte_index=0; pixel < frame_image->width() * frame_image->height(); pixel++, byte_index+=4)
{
// Get the RGBA values from each pixel
unsigned char R = pixels[byte_index];
unsigned char G = pixels[byte_index + 1];
unsigned char B = pixels[byte_index + 2];
unsigned char A = pixels[byte_index + 3];

// Split channels into their own arrays
red[pixel] = R;
green[pixel] = G;
blue[pixel] = B;
alpha[pixel] = A;
}

// Init target RGBA arrays
for (int i = 0; i < (frame_image->width() * frame_image->height()); i++) blur_red[i] = red[i];
for (int i = 0; i < (frame_image->width() * frame_image->height()); i++) blur_green[i] = green[i];
for (int i = 0; i < (frame_image->width() * frame_image->height()); i++) blur_blue[i] = blue[i];
for (int i = 0; i < (frame_image->width() * frame_image->height()); i++) blur_alpha[i] = alpha[i];
// Grab two copies of the image pixel data
QImage image_copy = frame_image->copy();
std::shared_ptr<QImage> frame_image_2 = std::make_shared<QImage>(image_copy);

// Loop through each iteration
for (int iteration = 0; iteration < iteration_value; iteration++)
for (int iteration = 0; iteration < iteration_value; ++iteration)
{
// HORIZONTAL BLUR (if any)
if (horizontal_radius_value > 0.0) {
// Init boxes for computing blur
int *bxs = initBoxes(sigma_value, horizontal_radius_value);

// Apply horizontal blur to target RGBA channels
boxBlurH(red, blur_red, frame_image->width(), frame_image->height(), horizontal_radius_value);
boxBlurH(green, blur_green, frame_image->width(), frame_image->height(), horizontal_radius_value);
boxBlurH(blue, blur_blue, frame_image->width(), frame_image->height(), horizontal_radius_value);
boxBlurH(alpha, blur_alpha, frame_image->width(), frame_image->height(), horizontal_radius_value);

// Remove boxes
delete[] bxs;

// Copy blur_<chan> back to <chan> for vertical blur or next iteration
for (int i = 0; i < (frame_image->width() * frame_image->height()); i++) red[i] = blur_red[i];
for (int i = 0; i < (frame_image->width() * frame_image->height()); i++) green[i] = blur_green[i];
for (int i = 0; i < (frame_image->width() * frame_image->height()); i++) blue[i] = blur_blue[i];
for (int i = 0; i < (frame_image->width() * frame_image->height()); i++) alpha[i] = blur_alpha[i];
boxBlurH(frame_image->bits(), frame_image_2->bits(), w, h, horizontal_radius_value);

// Swap output image back to input
frame_image.swap(frame_image_2);
}

// VERTICAL BLUR (if any)
if (vertical_radius_value > 0.0) {
// Init boxes for computing blur
int *bxs = initBoxes(sigma_value, vertical_radius_value);

// Apply vertical blur to target RGBA channels
boxBlurT(red, blur_red, frame_image->width(), frame_image->height(), vertical_radius_value);
boxBlurT(green, blur_green, frame_image->width(), frame_image->height(), vertical_radius_value);
boxBlurT(blue, blur_blue, frame_image->width(), frame_image->height(), vertical_radius_value);
boxBlurT(alpha, blur_alpha, frame_image->width(), frame_image->height(), vertical_radius_value);

// Remove boxes
delete[] bxs;

// Copy blur_<chan> back to <chan> for vertical blur or next iteration
for (int i = 0; i < (frame_image->width() * frame_image->height()); i++) red[i] = blur_red[i];
for (int i = 0; i < (frame_image->width() * frame_image->height()); i++) green[i] = blur_green[i];
for (int i = 0; i < (frame_image->width() * frame_image->height()); i++) blue[i] = blur_blue[i];
for (int i = 0; i < (frame_image->width() * frame_image->height()); i++) alpha[i] = blur_alpha[i];
}
}
boxBlurT(frame_image->bits(), frame_image_2->bits(), w, h, vertical_radius_value);

// Copy RGBA channels back to original image
for (int pixel = 0, byte_index=0; pixel < frame_image->width() * frame_image->height(); pixel++, byte_index+=4)
{
// Get the RGB values from the pixel
unsigned char R = blur_red[pixel];
unsigned char G = blur_green[pixel];
unsigned char B = blur_blue[pixel];
unsigned char A = blur_alpha[pixel];

// Split channels into their own arrays
pixels[byte_index] = R;
pixels[byte_index + 1] = G;
pixels[byte_index + 2] = B;
pixels[byte_index + 3] = A;
// Swap output image back to input
frame_image.swap(frame_image_2);
}
}

// Delete channel arrays
delete[] red;
delete[] green;
delete[] blue;
delete[] alpha;
delete[] blur_red;
delete[] blur_green;
delete[] blur_blue;
delete[] blur_alpha;

// return the modified frame
return frame;
}

// Credit: http://blog.ivank.net/fastest-gaussian-blur.html (MIT License)
int* Blur::initBoxes(float sigma, int n) // standard deviation, number of boxes
{
float wIdeal = sqrt((12.0 * sigma * sigma / n) + 1.0); // Ideal averaging filter width
int wl = floor(wIdeal);
if (wl % 2 == 0) wl--;
int wu = wl + 2;

float mIdeal = (12.0 * sigma * sigma - n * wl * wl - 4 * n * wl - 3 * n) / (-4.0 * wl - 4);
int m = round(mIdeal);

int *sizes = new int[n]();
for (int i = 0; i < n; i++) sizes[i] = i < m ? wl : wu;
return sizes;
}

// Credit: http://blog.ivank.net/fastest-gaussian-blur.html (MIT License)
// Modified to process all four channels in a pixel array
void Blur::boxBlurH(unsigned char *scl, unsigned char *tcl, int w, int h, int r) {
float iarr = 1.0 / (r + r + 1);
for (int i = 0; i < h; i++) {
int ti = i * w, li = ti, ri = ti + r;
int fv = scl[ti], lv = scl[ti + w - 1], val = (r + 1) * fv;
for (int j = 0; j < r; j++) val += scl[ti + j];
for (int j = 0; j <= r; j++) {
val += scl[ri++] - fv;
tcl[ti++] = round(val * iarr);
}
for (int j = r + 1; j < w - r; j++) {
val += scl[ri++] - scl[li++];
tcl[ti++] = round(val * iarr);
}
for (int j = w - r; j < w; j++) {
val += lv - scl[li++];
tcl[ti++] = round(val * iarr);

#pragma omp parallel for shared (scl, tcl)
for (int i = 0; i < h; ++i) {
for (int ch = 0; ch < 4; ++ch) {
int ti = i * w, li = ti, ri = ti + r;
int fv = scl[ti * 4 + ch], lv = scl[(ti + w - 1) * 4 + ch], val = (r + 1) * fv;
for (int j = 0; j < r; ++j) {
val += scl[(ti + j) * 4 + ch];
}
for (int j = 0; j <= r; ++j) {
val += scl[ri++ * 4 + ch] - fv;
tcl[ti++ * 4 + ch] = round(val * iarr);
}
for (int j = r + 1; j < w - r; ++j) {
val += scl[ri++ * 4 + ch] - scl[li++ * 4 + ch];
tcl[ti++ * 4 + ch] = round(val * iarr);
}
for (int j = w - r; j < w; ++j) {
val += lv - scl[li++ * 4 + ch];
tcl[ti++ * 4 + ch] = round(val * iarr);
}
}
}
}

void Blur::boxBlurT(unsigned char *scl, unsigned char *tcl, int w, int h, int r) {
float iarr = 1.0 / (r + r + 1);

#pragma omp parallel for shared (scl, tcl)
for (int i = 0; i < w; i++) {
int ti = i, li = ti, ri = ti + r * w;
int fv = scl[ti], lv = scl[ti + w * (h - 1)], val = (r + 1) * fv;
for (int j = 0; j < r; j++) val += scl[ti + j * w];
for (int j = 0; j <= r; j++) {
val += scl[ri] - fv;
tcl[ti] = round(val * iarr);
ri += w;
ti += w;
}
for (int j = r + 1; j < h - r; j++) {
val += scl[ri] - scl[li];
tcl[ti] = round(val * iarr);
li += w;
ri += w;
ti += w;
}
for (int j = h - r; j < h; j++) {
val += lv - scl[li];
tcl[ti] = round(val * iarr);
li += w;
ti += w;
for (int ch = 0; ch < 4; ++ch) {
int ti = i, li = ti, ri = ti + r * w;
int fv = scl[ti * 4 + ch], lv = scl[(ti + w * (h - 1)) * 4 + ch], val = (r + 1) * fv;
for (int j = 0; j < r; j++) val += scl[(ti + j * w) * 4 + ch];
for (int j = 0; j <= r; j++) {
val += scl[ri * 4 + ch] - fv;
tcl[ti * 4 + ch] = round(val * iarr);
ri += w;
ti += w;
}
for (int j = r + 1; j < h - r; j++) {
val += scl[ri * 4 + ch] - scl[li * 4 + ch];
tcl[ti * 4 + ch] = round(val * iarr);
li += w;
ri += w;
ti += w;
}
for (int j = h - r; j < h; j++) {
val += lv - scl[li * 4 + ch];
tcl[ti * 4 + ch] = round(val * iarr);
li += w;
ti += w;
}
}
}
}
Expand Down

0 comments on commit 6666702

Please sign in to comment.