mirror of
https://github.com/BinomialLLC/basis_universal.git
synced 2026-06-08 00:23:52 +00:00
Adding -compare_plot command, and defaulting to not displaying plots in -compare mode
This commit is contained in:
Richard Geldreich
309
basisu_tool.cpp
309
basisu_tool.cpp
@@ -145,6 +145,7 @@ static void print_usage()
|
||||
" -etc1_only: Only unpack to ETC1, skipping the other texture formats during -unpack\n"
|
||||
" -disable_hierarchical_endpoint_codebooks: Disable hierarchical endpoint codebook usage, slower but higher quality on some compression levels\n"
|
||||
" -compare_ssim: Compute and display SSIM of image comparison (slow)\n"
|
||||
" -compare_plot: Display histogram plots in -compare mode\n"
|
||||
" -bench: UASTC benchmark mode, for development only\n"
|
||||
" -resample X Y: Resample all input textures to XxY pixels using a box filter\n"
|
||||
" -resample_factor X: Resample all input textures by scale factor X using a box filter\n"
|
||||
@@ -296,6 +297,7 @@ public:
|
||||
m_etc1_only(false),
|
||||
m_fuzz_testing(false),
|
||||
m_compare_ssim(false),
|
||||
m_compare_plot(false),
|
||||
m_bench(false),
|
||||
m_parallel_compression(false)
|
||||
{
|
||||
@@ -364,6 +366,8 @@ public:
|
||||
m_mode = cVersion;
|
||||
else if (strcasecmp(pArg, "-compare_ssim") == 0)
|
||||
m_compare_ssim = true;
|
||||
else if (strcasecmp(pArg, "-compare_plot") == 0)
|
||||
m_compare_plot = true;
|
||||
else if (strcasecmp(pArg, "-bench") == 0)
|
||||
m_mode = cBench;
|
||||
else if (strcasecmp(pArg, "-comp_size") == 0)
|
||||
@@ -868,6 +872,7 @@ public:
|
||||
bool m_etc1_only;
|
||||
bool m_fuzz_testing;
|
||||
bool m_compare_ssim;
|
||||
bool m_compare_plot;
|
||||
bool m_bench;
|
||||
bool m_parallel_compression;
|
||||
};
|
||||
@@ -2749,180 +2754,184 @@ static bool compare_mode(command_line_params &opts)
|
||||
save_png("delta_img_a.png", delta_img, cImageSaveGrayscale, 3);
|
||||
printf("Wrote delta_img_a.png\n");
|
||||
|
||||
uint32_t bins[5][512];
|
||||
clear_obj(bins);
|
||||
|
||||
running_stat delta_stats[5];
|
||||
basisu::rand rm;
|
||||
|
||||
double avg[5];
|
||||
clear_obj(avg);
|
||||
|
||||
for (uint32_t y = 0; y < a.get_height(); y++)
|
||||
if (opts.m_compare_plot)
|
||||
{
|
||||
for (uint32_t x = 0; x < a.get_width(); x++)
|
||||
uint32_t bins[5][512];
|
||||
clear_obj(bins);
|
||||
|
||||
running_stat delta_stats[5];
|
||||
basisu::rand rm;
|
||||
|
||||
double avg[5];
|
||||
clear_obj(avg);
|
||||
|
||||
for (uint32_t y = 0; y < a.get_height(); y++)
|
||||
{
|
||||
//color_rgba& d = delta_img(x, y);
|
||||
|
||||
for (int c = 0; c < 4; c++)
|
||||
for (uint32_t x = 0; x < a.get_width(); x++)
|
||||
{
|
||||
int delta = a(x, y)[c] - b(x, y)[c];
|
||||
|
||||
//delta = clamp<int>((int)std::round(rm.gaussian(70.0f, 10.0f)), -255, 255);
|
||||
|
||||
bins[c][delta + 256]++;
|
||||
delta_stats[c].push(delta);
|
||||
//color_rgba& d = delta_img(x, y);
|
||||
|
||||
avg[c] += delta;
|
||||
}
|
||||
|
||||
int y_delta = a(x, y).get_709_luma() - b(x, y).get_709_luma();
|
||||
bins[4][y_delta + 256]++;
|
||||
delta_stats[4].push(y_delta);
|
||||
|
||||
avg[4] += y_delta;
|
||||
for (int c = 0; c < 4; c++)
|
||||
{
|
||||
int delta = a(x, y)[c] - b(x, y)[c];
|
||||
|
||||
} // x
|
||||
} // y
|
||||
//delta = clamp<int>((int)std::round(rm.gaussian(70.0f, 10.0f)), -255, 255);
|
||||
|
||||
for (uint32_t i = 0; i <= 4; i++)
|
||||
avg[i] /= a.get_total_pixels();
|
||||
bins[c][delta + 256]++;
|
||||
delta_stats[c].push(delta);
|
||||
|
||||
printf("\n");
|
||||
avg[c] += delta;
|
||||
}
|
||||
|
||||
//bins[2][256+-255] = 100000;
|
||||
//bins[2][256-56] = 50000;
|
||||
int y_delta = a(x, y).get_709_luma() - b(x, y).get_709_luma();
|
||||
bins[4][y_delta + 256]++;
|
||||
delta_stats[4].push(y_delta);
|
||||
|
||||
const uint32_t X_SIZE = 128, Y_SIZE = 40;
|
||||
avg[4] += y_delta;
|
||||
|
||||
for (uint32_t c = 0; c <= 4; c++)
|
||||
{
|
||||
std::vector<uint8_t> plot[Y_SIZE + 1];
|
||||
for (uint32_t i = 0; i < Y_SIZE; i++)
|
||||
{
|
||||
plot[i].resize(X_SIZE + 2);
|
||||
memset(plot[i].data(), ' ', X_SIZE + 1);
|
||||
}
|
||||
} // x
|
||||
} // y
|
||||
|
||||
uint32_t max_val = 0;
|
||||
int max_val_bin_index = 0;
|
||||
int lowest_bin_index = INT_MAX, highest_bin_index = INT_MIN;
|
||||
double avg_val = 0;
|
||||
double total_val = 0;
|
||||
running_stat bin_stats;
|
||||
for (uint32_t i = 0; i <= 4; i++)
|
||||
avg[i] /= a.get_total_pixels();
|
||||
|
||||
for (int y = -255; y <= 255; y++)
|
||||
{
|
||||
uint32_t val = bins[c][256 + y];
|
||||
if (!val)
|
||||
continue;
|
||||
|
||||
bin_stats.push(y);
|
||||
|
||||
total_val += (double)val;
|
||||
|
||||
lowest_bin_index = minimum(lowest_bin_index, y);
|
||||
highest_bin_index = maximum(highest_bin_index, y);
|
||||
|
||||
if (val > max_val)
|
||||
{
|
||||
max_val = val;
|
||||
max_val_bin_index = y;
|
||||
}
|
||||
avg_val += y * (double)val;
|
||||
}
|
||||
avg_val /= total_val;
|
||||
|
||||
int lo_limit = -(int)X_SIZE / 2;
|
||||
int hi_limit = X_SIZE / 2;
|
||||
for (int x = lo_limit; x <= hi_limit; x++)
|
||||
{
|
||||
uint32_t total = 0;
|
||||
if (x == lo_limit)
|
||||
{
|
||||
for (int i = -255; i <= lo_limit; i++)
|
||||
total += bins[c][256 + i];
|
||||
}
|
||||
else if (x == hi_limit)
|
||||
{
|
||||
for (int i = hi_limit; i <= 255; i++)
|
||||
total += bins[c][256 + i];
|
||||
}
|
||||
else
|
||||
{
|
||||
total = bins[c][256 + x];
|
||||
}
|
||||
|
||||
uint32_t height = max_val ? (total * Y_SIZE + max_val - 1) / max_val : 0;
|
||||
|
||||
if (height)
|
||||
{
|
||||
for (uint32_t y = (Y_SIZE - 1) - (height - 1); y <= (Y_SIZE - 1); y++)
|
||||
plot[y][x + X_SIZE / 2] = '*';
|
||||
}
|
||||
}
|
||||
|
||||
printf("%c delta histogram: total samples: %5.0f, max bin value: %u index: %i (%3.3f%% of total), range %i [%i,%i], weighted mean: %f\n", "RGBAY"[c], total_val, max_val, max_val_bin_index, max_val * 100.0f / total_val, highest_bin_index - lowest_bin_index + 1, lowest_bin_index, highest_bin_index, avg_val);
|
||||
printf("bin mean: %f, bin std deviation: %f, non-zero bins: %u\n", bin_stats.get_mean(), bin_stats.get_std_dev(), bin_stats.get_num());
|
||||
printf("delta mean: %f, delta std deviation: %f\n", delta_stats[c].get_mean(), delta_stats[c].get_std_dev());
|
||||
printf("\n");
|
||||
|
||||
for (uint32_t y = 0; y < Y_SIZE; y++)
|
||||
printf("%s\n", (char*)plot[y].data());
|
||||
//bins[2][256+-255] = 100000;
|
||||
//bins[2][256-56] = 50000;
|
||||
|
||||
char tics[1024];
|
||||
tics[0] = '\0';
|
||||
char tics2[1024];
|
||||
tics2[0] = '\0';
|
||||
const uint32_t X_SIZE = 128, Y_SIZE = 40;
|
||||
|
||||
for (int x = 0; x <= (int)X_SIZE; x++)
|
||||
for (uint32_t c = 0; c <= 4; c++)
|
||||
{
|
||||
char buf[64];
|
||||
if (x == X_SIZE / 2)
|
||||
std::vector<uint8_t> plot[Y_SIZE + 1];
|
||||
for (uint32_t i = 0; i < Y_SIZE; i++)
|
||||
{
|
||||
while ((int)strlen(tics) < x)
|
||||
strcat(tics, ".");
|
||||
|
||||
while ((int)strlen(tics2) < x)
|
||||
strcat(tics2, " ");
|
||||
|
||||
sprintf(buf, "0");
|
||||
strcat(tics, buf);
|
||||
plot[i].resize(X_SIZE + 2);
|
||||
memset(plot[i].data(), ' ', X_SIZE + 1);
|
||||
}
|
||||
else if (((x & 7) == 0) || (x == X_SIZE))
|
||||
{
|
||||
while ((int)strlen(tics) < x)
|
||||
strcat(tics, ".");
|
||||
|
||||
while ((int)strlen(tics2) < x)
|
||||
strcat(tics2, " ");
|
||||
|
||||
int v = (x - (int)X_SIZE / 2);
|
||||
sprintf(buf, "%i", v / 10);
|
||||
strcat(tics, buf);
|
||||
|
||||
if (v < 0)
|
||||
uint32_t max_val = 0;
|
||||
int max_val_bin_index = 0;
|
||||
int lowest_bin_index = INT_MAX, highest_bin_index = INT_MIN;
|
||||
double avg_val = 0;
|
||||
double total_val = 0;
|
||||
running_stat bin_stats;
|
||||
|
||||
for (int y = -255; y <= 255; y++)
|
||||
{
|
||||
uint32_t val = bins[c][256 + y];
|
||||
if (!val)
|
||||
continue;
|
||||
|
||||
bin_stats.push(y);
|
||||
|
||||
total_val += (double)val;
|
||||
|
||||
lowest_bin_index = minimum(lowest_bin_index, y);
|
||||
highest_bin_index = maximum(highest_bin_index, y);
|
||||
|
||||
if (val > max_val)
|
||||
{
|
||||
if (-v < 10)
|
||||
sprintf(buf, "%i", v % 10);
|
||||
else
|
||||
sprintf(buf, " %i", -v % 10);
|
||||
max_val = val;
|
||||
max_val_bin_index = y;
|
||||
}
|
||||
avg_val += y * (double)val;
|
||||
}
|
||||
avg_val /= total_val;
|
||||
|
||||
int lo_limit = -(int)X_SIZE / 2;
|
||||
int hi_limit = X_SIZE / 2;
|
||||
for (int x = lo_limit; x <= hi_limit; x++)
|
||||
{
|
||||
uint32_t total = 0;
|
||||
if (x == lo_limit)
|
||||
{
|
||||
for (int i = -255; i <= lo_limit; i++)
|
||||
total += bins[c][256 + i];
|
||||
}
|
||||
else if (x == hi_limit)
|
||||
{
|
||||
for (int i = hi_limit; i <= 255; i++)
|
||||
total += bins[c][256 + i];
|
||||
}
|
||||
else
|
||||
sprintf(buf, "%i", v % 10);
|
||||
strcat(tics2, buf);
|
||||
}
|
||||
else
|
||||
{
|
||||
while ((int)strlen(tics) < x)
|
||||
strcat(tics, ".");
|
||||
}
|
||||
}
|
||||
printf("%s\n", tics);
|
||||
printf("%s\n", tics2);
|
||||
{
|
||||
total = bins[c][256 + x];
|
||||
}
|
||||
|
||||
printf("\n");
|
||||
}
|
||||
uint32_t height = max_val ? (total * Y_SIZE + max_val - 1) / max_val : 0;
|
||||
|
||||
if (height)
|
||||
{
|
||||
for (uint32_t y = (Y_SIZE - 1) - (height - 1); y <= (Y_SIZE - 1); y++)
|
||||
plot[y][x + X_SIZE / 2] = '*';
|
||||
}
|
||||
}
|
||||
|
||||
printf("%c delta histogram: total samples: %5.0f, max bin value: %u index: %i (%3.3f%% of total), range %i [%i,%i], weighted mean: %f\n", "RGBAY"[c], total_val, max_val, max_val_bin_index, max_val * 100.0f / total_val, highest_bin_index - lowest_bin_index + 1, lowest_bin_index, highest_bin_index, avg_val);
|
||||
printf("bin mean: %f, bin std deviation: %f, non-zero bins: %u\n", bin_stats.get_mean(), bin_stats.get_std_dev(), bin_stats.get_num());
|
||||
printf("delta mean: %f, delta std deviation: %f\n", delta_stats[c].get_mean(), delta_stats[c].get_std_dev());
|
||||
printf("\n");
|
||||
|
||||
for (uint32_t y = 0; y < Y_SIZE; y++)
|
||||
printf("%s\n", (char*)plot[y].data());
|
||||
|
||||
char tics[1024];
|
||||
tics[0] = '\0';
|
||||
char tics2[1024];
|
||||
tics2[0] = '\0';
|
||||
|
||||
for (int x = 0; x <= (int)X_SIZE; x++)
|
||||
{
|
||||
char buf[64];
|
||||
if (x == X_SIZE / 2)
|
||||
{
|
||||
while ((int)strlen(tics) < x)
|
||||
strcat(tics, ".");
|
||||
|
||||
while ((int)strlen(tics2) < x)
|
||||
strcat(tics2, " ");
|
||||
|
||||
sprintf(buf, "0");
|
||||
strcat(tics, buf);
|
||||
}
|
||||
else if (((x & 7) == 0) || (x == X_SIZE))
|
||||
{
|
||||
while ((int)strlen(tics) < x)
|
||||
strcat(tics, ".");
|
||||
|
||||
while ((int)strlen(tics2) < x)
|
||||
strcat(tics2, " ");
|
||||
|
||||
int v = (x - (int)X_SIZE / 2);
|
||||
sprintf(buf, "%i", v / 10);
|
||||
strcat(tics, buf);
|
||||
|
||||
if (v < 0)
|
||||
{
|
||||
if (-v < 10)
|
||||
sprintf(buf, "%i", v % 10);
|
||||
else
|
||||
sprintf(buf, " %i", -v % 10);
|
||||
}
|
||||
else
|
||||
sprintf(buf, "%i", v % 10);
|
||||
strcat(tics2, buf);
|
||||
}
|
||||
else
|
||||
{
|
||||
while ((int)strlen(tics) < x)
|
||||
strcat(tics, ".");
|
||||
}
|
||||
}
|
||||
printf("%s\n", tics);
|
||||
printf("%s\n", tics2);
|
||||
|
||||
printf("\n");
|
||||
}
|
||||
|
||||
} // display_plot
|
||||
|
||||
return true;
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user