mirror of
https://github.com/amyinspace/MagicSetEditor2.git
synced 2026-06-10 04:57:00 -04:00
Lymia Aluysia
297 lines
12 KiB
C++
297 lines
12 KiB
C++
//+----------------------------------------------------------------------------+
|
|
//| Description: Magic Set Editor - Program to make Magic (tm) cards |
|
|
//| Copyright: (C) 2001 - 2010 Twan van Laarhoven and Sean Hunt |
|
|
//| License: GNU General Public License 2 or later (see file COPYING) |
|
|
//+----------------------------------------------------------------------------+
|
|
|
|
// ----------------------------------------------------------------------------- : Includes
|
|
|
|
#include <util/prec.hpp>
|
|
#include <gfx/gfx.hpp>
|
|
#include <util/error.hpp>
|
|
|
|
// ----------------------------------------------------------------------------- : Resample passes
|
|
|
|
// bitshift for fixed point numbers
|
|
// higher is less error
|
|
// we will get errors if 2^shift * imagesize becomes too large
|
|
const int shift = 32-10-8; // => max size = 1024, max alpha = 255
|
|
|
|
// Resample an image only in a single direction, either horizontally or vertically
|
|
/* Terms are based on x resampling (keeping the same number of lines):
|
|
* offset = number of elements to skip at the start
|
|
* length = length of a line
|
|
* delta = number of elements between pixels in a lines
|
|
* lines = number of lines
|
|
* line_delta = number of elements between the the first pixel of two lines
|
|
* 1 element = 3 bytes in data, 1 byte in alpha
|
|
*/
|
|
void resample_pass(const Image& img_in, Image& img_out, int offset_in, int offset_out,
|
|
int length_in, int delta_in, int length_out, int delta_out,
|
|
int lines, int line_delta_in, int line_delta_out)
|
|
{
|
|
bool alpha = img_in.HasAlpha();
|
|
if (alpha && !img_out.HasAlpha()) img_out.InitAlpha();
|
|
int out_fact = (length_out << shift) / length_in; // how much to output for 256 input = 1 pixel
|
|
int out_rest = (length_out << shift) % length_in;
|
|
// for each line
|
|
for (int l = 0 ; l < lines ; ++l) {
|
|
Byte* in = img_in .GetData() + 3 * (offset_in + l * line_delta_in);
|
|
Byte* out = img_out.GetData() + 3 * (offset_out + l * line_delta_out);
|
|
UInt in_rem = out_fact + out_rest; // remaining to input from the current input pixel
|
|
|
|
if (alpha) {
|
|
Byte* in_a = img_in .GetAlpha() + (offset_in + l * line_delta_in);
|
|
Byte* out_a = img_out.GetAlpha() + (offset_out + l * line_delta_out);
|
|
|
|
for (int x = 0 ; x < length_out ; ++x) {
|
|
UInt out_rem = 1 << shift;
|
|
UInt totR = 0, totG = 0, totB = 0, totA = 0;
|
|
while (out_rem >= in_rem) {
|
|
// eat a whole input pixel
|
|
totR += in[0] * in_rem * in_a[0]; // multiply by alpha
|
|
totG += in[1] * in_rem * in_a[0];
|
|
totB += in[2] * in_rem * in_a[0];
|
|
totA += in_a[0] * in_rem;
|
|
out_rem -= in_rem;
|
|
in_rem = out_fact;
|
|
in += 3*delta_in; in_a += delta_in;
|
|
}
|
|
if (out_rem > 0) {
|
|
// eat a partial input pixel
|
|
totR += in[0] * out_rem * in_a[0];
|
|
totG += in[1] * out_rem * in_a[0];
|
|
totB += in[2] * out_rem * in_a[0];
|
|
totA += in_a[0] * out_rem;
|
|
in_rem -= out_rem;
|
|
}
|
|
// store
|
|
if (totA) {
|
|
out[0] = totR / totA;
|
|
out[1] = totG / totA;
|
|
out[2] = totB / totA;
|
|
out_a[0] = totA >> shift;
|
|
} else {
|
|
out[0] = out[1] = out[2] = out_a[0] = 0; // div by 0 is bad
|
|
}
|
|
out += 3*delta_out; out_a += delta_out;
|
|
}
|
|
|
|
} else {
|
|
// no alpha
|
|
for (int x = 0 ; x < length_out ; ++x) {
|
|
UInt out_rem = 1 << shift;
|
|
UInt totR = 0, totG = 0, totB = 0;
|
|
while (out_rem >= in_rem) {
|
|
// eat a whole input pixel
|
|
totR += in[0] * in_rem;
|
|
totG += in[1] * in_rem;
|
|
totB += in[2] * in_rem;
|
|
out_rem -= in_rem;
|
|
in_rem = out_fact;
|
|
in += 3*delta_in;
|
|
}
|
|
if (out_rem > 0) {
|
|
// eat a partial input pixel
|
|
totR += in[0] * out_rem;
|
|
totG += in[1] * out_rem;
|
|
totB += in[2] * out_rem;
|
|
in_rem -= out_rem;
|
|
}
|
|
// store
|
|
out[0] = totR >> shift;
|
|
out[1] = totG >> shift;
|
|
out[2] = totB >> shift;
|
|
out += 3*delta_out;
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
// ----------------------------------------------------------------------------- : Resample
|
|
|
|
/* The algorithm first resizes in horizontally, then vertically,
|
|
* the two passes are essentially the same:
|
|
* - for each row:
|
|
* - each input pixel becomes a fixed amount of output (in 1<<shift fixed point math)
|
|
* - for each output pixel:
|
|
* - _('eat') input pixels until the total is 1<<shift
|
|
* - write the total to the output pixel
|
|
* - to ensure the sum of all the pixel amounts is exacly width<<shift an extra rest amount
|
|
* is _('eaten') from the first pixel;
|
|
*
|
|
* Uses fixed point numbers
|
|
*/
|
|
void resample(const Image& img_in, Image& img_out) {
|
|
resample_and_clip(img_in, img_out, wxRect(0, 0, img_in.GetWidth(), img_in.GetHeight()));
|
|
}
|
|
Image resample(const Image& img_in, int width, int height) {
|
|
if (img_in.GetWidth() == width && img_in.GetHeight() == height) {
|
|
return img_in; // already the right size
|
|
} else {
|
|
Image img_out(width,height,false);
|
|
resample(img_in, img_out);
|
|
return img_out;
|
|
}
|
|
}
|
|
|
|
void resample_and_clip(const Image& img_in, Image& img_out, wxRect rect) {
|
|
// mask to alpha
|
|
if (img_in.HasMask() && !img_in.HasAlpha()) {
|
|
const_cast<Image&>(img_in).InitAlpha();
|
|
}
|
|
// starting position in data
|
|
int offset_in = (rect.x + img_in.GetWidth() * rect.y);
|
|
if (img_out.GetHeight() == rect.height) {
|
|
// no resizing vertically
|
|
resample_pass(img_in, img_out, offset_in, 0, rect.width, 1, img_out .GetWidth(), 1, rect .GetHeight(), img_in.GetWidth(), img_out .GetWidth());
|
|
} else {
|
|
Image img_temp(img_out.GetWidth(), rect.height, false);
|
|
resample_pass(img_in, img_temp, offset_in, 0, rect.width, 1, img_temp.GetWidth(), 1, rect .GetHeight(), img_in.GetWidth(), img_temp.GetWidth());
|
|
resample_pass(img_temp, img_out, 0, 0, rect.height, img_temp.GetWidth(), img_out .GetHeight(), img_temp.GetWidth(), img_temp.GetWidth(), 1, 1);
|
|
}
|
|
}
|
|
|
|
|
|
// ----------------------------------------------------------------------------- : Aspect ratio preserving
|
|
|
|
// fill an image with 100% transparent
|
|
void fill_transparent(Image& img) {
|
|
if (!img.HasAlpha()) img.InitAlpha();
|
|
memset(img.GetAlpha(), 0, img.GetWidth() * img.GetHeight());
|
|
}
|
|
|
|
void resample_preserve_aspect(const Image& img_in, Image& img_out) {
|
|
int rheight = img_in.GetHeight() * img_out.GetWidth() / img_in.GetWidth();
|
|
int rwidth = img_in.GetWidth() * img_out.GetHeight() / img_in.GetHeight();
|
|
// actual size of output
|
|
if (rheight < img_out.GetHeight()) rwidth = img_out.GetWidth();
|
|
else if (rwidth < img_out.GetWidth()) rheight = img_out.GetHeight();
|
|
else {rwidth = img_out.GetWidth(); rheight = img_out.GetHeight();}
|
|
int dx = (img_out.GetWidth() - rwidth) / 2;
|
|
int dy = (img_out.GetHeight() - rheight) / 2;
|
|
// transparent background
|
|
fill_transparent(img_out);
|
|
// resample
|
|
int offset_out = dx + img_out.GetWidth() * dy;
|
|
Image img_temp(rwidth, img_in.GetHeight(), false);
|
|
img_temp.InitAlpha();
|
|
resample_pass(img_in, img_temp, 0, 0, img_in.GetWidth(), 1, rwidth, 1, img_in.GetHeight(), img_in.GetWidth(), img_temp.GetWidth());
|
|
resample_pass(img_temp, img_out, 0, offset_out, img_in.GetHeight(), img_temp.GetWidth(), rheight, img_out.GetWidth(), rwidth, 1, 1);
|
|
}
|
|
|
|
Image resample_preserve_aspect(const Image& img_in, int width, int height) {
|
|
if (img_in.GetWidth() == width && img_in.GetHeight() == height) {
|
|
return img_in; // already the right size
|
|
} else {
|
|
Image img_out(width,height,false);
|
|
resample_preserve_aspect(img_in, img_out);
|
|
return img_out;
|
|
}
|
|
}
|
|
|
|
// ----------------------------------------------------------------------------- : Sharpening
|
|
|
|
void sharp_downsample(const Image& img_in, Image& img_out, int amount);
|
|
|
|
void sharp_resample(const Image& img_in, Image& img_out, int amount) {
|
|
sharp_resample_and_clip(img_in, img_out, wxRect(0, 0, img_in.GetWidth(), img_in.GetHeight()), amount);
|
|
}
|
|
|
|
void sharp_resample_and_clip(const Image& img_in, Image& img_out, wxRect rect, int amount) {
|
|
Image img_larger(img_out.GetWidth() * 2, img_out.GetHeight() * 2, false);
|
|
resample_and_clip(img_in, img_larger, rect);
|
|
sharp_downsample(img_larger, img_out, amount);
|
|
}
|
|
|
|
// Downsample an image to create a sharp result by applying a sharpening filter
|
|
// img_in must be twice as large as img_out
|
|
void sharp_downsample(const Image& img_in, Image& img_out, int amount) {
|
|
assert(img_in.GetWidth() == img_out.GetWidth() * 2);
|
|
assert(img_in.GetHeight() == img_out.GetHeight() * 2);
|
|
|
|
int width = img_out.GetWidth(), height = img_out.GetHeight();
|
|
int line = width * 6;
|
|
int center_weight = 201;
|
|
int border_weight = amount;
|
|
assert(4 * center_weight - 8 * border_weight > 0);
|
|
|
|
Byte *in = img_in.GetData(), *out = img_out.GetData();
|
|
Byte *al = nullptr, *outa = nullptr;
|
|
if (img_in.HasAlpha()) {
|
|
img_out.InitAlpha();
|
|
al = img_in.GetAlpha();
|
|
outa = img_out.GetAlpha();
|
|
}
|
|
|
|
for (int y = 0 ; y < height ; ++y) {
|
|
for (int x = 0 ; x < width ; ++x) {
|
|
// Filter using a kernel of the form
|
|
/* -1 -1
|
|
* -1 c c -1
|
|
* -1 c c -1
|
|
* -1 -1
|
|
* But when we are near the edge ignore the pixel
|
|
*/
|
|
// when there is alpha, all weights are multiplied by 4*255
|
|
int center_alpha = al ? al[0] + al[1] + al[width*2] + al[width*2+1] : 1;
|
|
int weight = center_weight * center_alpha * 4;
|
|
int sumR = center_weight * center_alpha * (in[0] + in[3] + in[line+0] + in[line+3]);
|
|
int sumG = center_weight * center_alpha * (in[1] + in[4] + in[line+1] + in[line+4]);
|
|
int sumB = center_weight * center_alpha * (in[2] + in[5] + in[line+2] + in[line+5]);
|
|
// edges
|
|
if (x != 0) {
|
|
int a = al ? border_weight * min(2 * (al[-1] + al[width*2-1]), center_alpha)
|
|
: border_weight;
|
|
sumR -= a * (in[-3] + in[line-3]);
|
|
sumG -= a * (in[-2] + in[line-2]);
|
|
sumB -= a * (in[-1] + in[line-1]);
|
|
weight -= a * 2;
|
|
}
|
|
if (x+1 != width) {
|
|
int a = al ? border_weight * min(2 * (al[2] + al[width*2+2]), center_alpha)
|
|
: border_weight;
|
|
sumR -= a * (in[6] + in[line+6]);
|
|
sumG -= a * (in[7] + in[line+7]);
|
|
sumB -= a * (in[8] + in[line+8]);
|
|
weight -= a * 2;
|
|
}
|
|
if (y != 0) {
|
|
int a = al ? border_weight * min(2 * (al[-width*2] + al[-width*2+1]), center_alpha)
|
|
: border_weight;
|
|
sumR -= a * (in[-line+0] + in[-line+3]);
|
|
sumG -= a * (in[-line+1] + in[-line+4]);
|
|
sumB -= a * (in[-line+2] + in[-line+5]);
|
|
weight -= a * 2;
|
|
}
|
|
if (y+1 != height) {
|
|
int a = al ? border_weight * min(2 * (al[width*2*2] + al[width*2*2+1]), center_alpha)
|
|
: border_weight;
|
|
sumR -= a * (in[line*2+0] + in[line*2+3]);
|
|
sumG -= a * (in[line*2+1] + in[line*2+4]);
|
|
sumB -= a * (in[line*2+2] + in[line*2+5]);
|
|
weight -= a * 2;
|
|
}
|
|
// And then avarage the result into a single pixel (downsample by factor 2 in both dimensions)
|
|
if (weight > 0) {
|
|
out[0] = col( sumR / weight );
|
|
out[1] = col( sumG / weight );
|
|
out[2] = col( sumB / weight );
|
|
} else {
|
|
out[0] = out[1] = out[2] = 0;
|
|
}
|
|
if (al) {
|
|
outa[0] = center_alpha / 4;
|
|
outa += 1;
|
|
al += 2;
|
|
}
|
|
// next pixel
|
|
in += 6;
|
|
out += 3;
|
|
}
|
|
// skip a line
|
|
in += line;
|
|
if (al) al += width*2;
|
|
}
|
|
}
|