sij/mapnik
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Fix apply_color_blind_filter to use correct color-space and avoid NAN which results in SIGILL when compiled with DEBUG_UNDEFINED=yes (#3861)

Browse source
This commit is contained in:
Artem Pavlenko 2018-02-20 11:17:18 +01:00
parent d43c4c1812
commit bac24cd844
2 changed files with 38 additions and 56 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

92
include/mapnik/image_filter.hpp
View file

@ -28,7 +28,6 @@
#include <mapnik/image_filter_types.hpp>
#include <mapnik/image_util.hpp>
#include <mapnik/util/hsl.hpp>
#pragma GCC diagnostic push
#include <mapnik/warning_ignore.hpp>
#include <boost/gil/gil_all.hpp>
@ -687,36 +686,37 @@ void apply_color_blind_filter(Src & src, ColorBlindFilter const& op)
rgba8_view_t src_view = rgba8_view(src);
bool premultiplied = src.get_premultiplied();
static constexpr double gamma = 2.2;
static constexpr double inv_gamma = 1.0/gamma;
for (std::ptrdiff_t y = 0; y < src_view.height(); ++y)
{
rgba8_view_t::x_iterator src_it = src_view.row_begin(static_cast<long>(y));
for (std::ptrdiff_t x = 0; x < src_view.width(); ++x)
{
// formula taken from boost/gil/color_convert.hpp:rgb_to_luminance
uint8_t & r = get_color(src_it[x], red_t());
uint8_t & g = get_color(src_it[x], green_t());
uint8_t & b = get_color(src_it[x], blue_t());
uint8_t & a = get_color(src_it[x], alpha_t());
double dr = static_cast<double>(r)/255.0;
double dg = static_cast<double>(g)/255.0;
double db = static_cast<double>(b)/255.0;
double da = static_cast<double>(a)/255.0;
// demultiply
if (da <= 0.0)
if (a == 0)
{
r = g = b = 0;
continue;
}
else if (premultiplied)
{
dr /= da;
dg /= da;
db /= da;
std::uint32_t cr = (r * 255) / a;
std::uint32_t cg = (g * 255) / a;
std::uint32_t cb = (b * 255) / a;
r = static_cast<uint8_t>((cr > 255) ? 255 : cr);
g = static_cast<uint8_t>((cg > 255) ? 255 : cg);
b = static_cast<uint8_t>((cb > 255) ? 255 : cb);
}
// Convert source color into XYZ color space
double pow_r = std::pow(dr, 2.2);
double pow_g = std::pow(dg, 2.2);
double pow_b = std::pow(db, 2.2);
double pow_r = std::pow(r, gamma);
double pow_g = std::pow(g, gamma);
double pow_b = std::pow(b, gamma);
double X = (0.412424 * pow_r) + (0.357579 * pow_g) + (0.180464 * pow_b);
double Y = (0.212656 * pow_r) + (0.715158 * pow_g) + (0.0721856 * pow_b);
double Z = (0.0193324 * pow_r) + (0.119193 * pow_g) + (0.950444 * pow_b);
@ -733,10 +733,6 @@ void apply_color_blind_filter(Src & src, ColorBlindFilter const& op)
if (std::abs(m_div2) < (std::numeric_limits<double>::epsilon())) continue;
double deviate_x = (op.yint - yint) / (m - op.m);
double deviate_y = (m * deviate_x) + yint;
if (std::abs(deviate_y) < (std::numeric_limits<double>::epsilon()))
{
deviate_y = std::numeric_limits<double>::epsilon() * 2.0;
}
// Compute the simulated color's XYZ coords
X = deviate_x * Y / deviate_y;
Z = (1.0 - (deviate_x + deviate_y)) * Y / deviate_y;
@ -746,25 +742,14 @@ void apply_color_blind_filter(Src & src, ColorBlindFilter const& op)
// Difference between simulated color and neutral grey
double diff_X = neutral_X - X;
double diff_Z = neutral_Z - Z;
double diff_r = diff_X * 3.24071 + diff_Z * -0.498571; // XYZ->RGB (sRGB:D65)
double diff_g = diff_X * -0.969258 + diff_Z * 0.0415557;
double diff_b = diff_X * 0.0556352 + diff_Z * 1.05707;
if (std::abs(diff_r) < (std::numeric_limits<double>::epsilon()))
{
diff_r = std::numeric_limits<double>::epsilon() * 2.0;
}
if (std::abs(diff_g) < (std::numeric_limits<double>::epsilon()))
{
diff_g = std::numeric_limits<double>::epsilon() * 2.0;
}
if (std::abs(diff_b) < (std::numeric_limits<double>::epsilon()))
{
diff_b = std::numeric_limits<double>::epsilon() * 2.0;
}
// Convert to RGB color space
dr = X * 3.24071 + Y * -1.53726 + Z * -0.498571; // XYZ->RGB (sRGB:D65)
dg = X * -0.969258 + Y * 1.87599 + Z * 0.0415557;
db = X * 0.0556352 + Y * -0.203996 + Z * 1.05707;
// XYZ->RGB (sRGB:D65)
double diff_r = diff_X * 3.2407100 + diff_Z *-0.4985710;
double diff_g = diff_X *-0.9692580 + diff_Z * 0.0415557;
double diff_b = diff_X * 0.0556352 + diff_Z * 1.0570700;
// XYZ->RGB (sRGB:D65)
double dr = X * 3.2407100 + Y *-1.537260 + Z *-0.4985710;
double dg = X *-0.9692580 + Y * 1.875990 + Z * 0.0415557;
double db = X * 0.0556352 + Y *-0.203996 + Z * 1.0570700;
// Compensate simulated color towards a neutral fit in RGB space
double fit_r = ((dr < 0.0 ? 0.0 : 1.0) - dr) / diff_r;
double fit_g = ((dg < 0.0 ? 0.0 : 1.0) - dg) / diff_g;
@ -774,27 +759,24 @@ void apply_color_blind_filter(Src & src, ColorBlindFilter const& op)
);
adjust = std::max((fit_b > 1.0 || fit_b < 0.0) ? 0.0 : fit_b, adjust);
// Shift proportional to the greatest shift
dr = dr + (adjust * diff_r);
dg = dg + (adjust * diff_g);
db = db + (adjust * diff_b);
dr += adjust * diff_r;
dg += adjust * diff_g;
db += adjust * diff_b;
// Apply gamma correction
dr = std::pow(dr, 1.0 / 2.2);
dg = std::pow(dg, 1.0 / 2.2);
db = std::pow(db, 1.0 / 2.2);
// premultiply
dr *= da;
dg *= da;
db *= da;
dr = std::pow(dr, inv_gamma);
dg = std::pow(dg, inv_gamma);
db = std::pow(db, inv_gamma);
// Clamp values
if(dr < 0.0) dr = 0.0;
if(dr > 1.0) dr = 1.0;
if(dg < 0.0) dg = 0.0;
if(dg > 1.0) dg = 1.0;
if(db < 0.0) db = 0.0;
if(db > 1.0) db = 1.0;
r = static_cast<uint8_t>(dr * 255.0);
g = static_cast<uint8_t>(dg * 255.0);
b = static_cast<uint8_t>(db * 255.0);
if(dr < 0.0 || std::isnan(dr)) dr = 0.0;
if(dr > 255.0) dr = 255.0;
if(dg < 0.0 || std::isnan(dg)) dg = 0.0;
if(dg > 255.0) dg = 255.0;
if(db < 0.0 || std::isnan(db)) db = 0.0;
if(db > 255.0) db = 255.0;
// premultiply
r = (static_cast<uint8_t>(dr) * a + 255) >> 8;
g = (static_cast<uint8_t>(dg) * a + 255) >> 8;
b = (static_cast<uint8_t>(db) * a + 255) >> 8;
}
}
// set as premultiplied

2
test/data-visual

@ -1 +1 @@
Subproject commit d975838e0f6fa75cb3b3574b43ae205979508f67
Subproject commit 052135e8a403cdea6835f084881b38c988710ba0