1
0
Fork 0
mutter-performance-source/clutter/clutter/clutter-blur.c
Georges Basile Stavracas Neto 3440fbd358 clutter: Add private ClutterBlur
ClutterBlur is a small helper structure that implements
blurring a texture.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1627>
2020-12-14 09:43:08 -03:00

422 lines
14 KiB
C

/*
* Copyright (C) 2020 Endless OS Foundation, LLC
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2 of the License, or (at your option) any later version.
*
* This library is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with this library. If not, see <http://www.gnu.org/licenses/>.
*/
#include "clutter-blur-private.h"
#include "clutter-backend.h"
/**
* SECTION:clutter-blur
* @short_description: Blur textures
*
* #ClutterBlur is a moderately fast gaussian blur implementation.
*
* # Optimizations
*
* There are a number of optimizations in place to make this blur implementation
* real-time. All in all, the implementation performs best when using large
* blur-radii that allow downscaling the texture to smaller sizes, at small
* radii where no downscaling is possible this can easily halve the framerate.
*
* ## Multipass
*
* It is implemented in 2 passes: vertical and horizontal.
*
* ## Downscaling
*
* #ClutterBlur uses dynamic downscaling to speed up blurring. Downscaling
* happens in factors of 2 (the image is downscaled either by 2, 4, 8, 16, …)
* and depends on the blur radius, the texture size, among others.
*
* The texture is drawn into a downscaled framebuffer; the blur passes are
* applied on the downscaled texture contents; and finally, the blurred
* contents are drawn
* upscaled again.
*
* ## Hardware Interpolation
*
* This blur implementation cuts down the number of sampling operations by
* exploiting the hardware interpolation that is performed when sampling between
* pixel boundaries. This technique is described at:
*
* http://rastergrid.com/blog/2010/09/efficient-gaussian-blur-with-linear-sampling/
*
* ## Incremental gauss-factor calculation
*
* The kernel values for the gaussian kernel are computed incrementally instead
* of running the expensive calculations multiple times inside the blur shader.
* The implementation is based on the algorithm presented by K. Turkowski in
* GPU Gems 3, chapter 40:
*
* https://developer.nvidia.com/gpugems/GPUGems3/gpugems3_ch40.html
*
*/
static const char *gaussian_blur_glsl_declarations =
"uniform float sigma; \n"
"uniform float pixel_step; \n"
"uniform int vertical; \n";
static const char *gaussian_blur_glsl =
" int horizontal = 1 - vertical; \n"
" \n"
" vec2 uv = vec2 (cogl_tex_coord.st); \n"
" \n"
" vec3 gauss_coefficient; \n"
" gauss_coefficient.x = 1.0 / (sqrt (2.0 * 3.14159265) * sigma); \n"
" gauss_coefficient.y = exp (-0.5 / (sigma * sigma)); \n"
" gauss_coefficient.z = gauss_coefficient.y * gauss_coefficient.y; \n"
" \n"
" float gauss_coefficient_total = gauss_coefficient.x; \n"
" \n"
" vec4 ret = texture2D (cogl_sampler, uv) * gauss_coefficient.x; \n"
" gauss_coefficient.xy *= gauss_coefficient.yz; \n"
" \n"
" int n_steps = int (ceil (3 * sigma)); \n"
" \n"
" for (int i = 1; i < n_steps; i += 2) { \n"
" float coefficient_subtotal = gauss_coefficient.x; \n"
" gauss_coefficient.xy *= gauss_coefficient.yz; \n"
" coefficient_subtotal += gauss_coefficient.x; \n"
" \n"
" float gauss_ratio = gauss_coefficient.x / coefficient_subtotal; \n"
" \n"
" float foffset = float (i) + gauss_ratio; \n"
" vec2 offset = vec2 (foffset * pixel_step * float (horizontal), \n"
" foffset * pixel_step * float (vertical)); \n"
" \n"
" ret += texture2D (cogl_sampler, uv + offset) * coefficient_subtotal; \n"
" ret += texture2D (cogl_sampler, uv - offset) * coefficient_subtotal; \n"
" \n"
" gauss_coefficient_total += 2.0 * coefficient_subtotal; \n"
" gauss_coefficient.xy *= gauss_coefficient.yz; \n"
" } \n"
" \n"
" cogl_texel = ret / gauss_coefficient_total; \n";
#define MIN_DOWNSCALE_SIZE 256.f
#define MAX_SIGMA 6.f
enum
{
VERTICAL,
HORIZONTAL,
};
typedef struct
{
CoglFramebuffer *framebuffer;
CoglPipeline *pipeline;
CoglTexture *texture;
int orientation;
int sigma_uniform;
int pixel_step_uniform;
int vertical_uniform;
} BlurPass;
struct _ClutterBlur
{
CoglTexture *source_texture;
unsigned int sigma;
float downscale_factor;
BlurPass pass[2];
};
static CoglPipeline*
create_blur_pipeline (void)
{
static CoglPipelineKey blur_pipeline_key = "clutter-blur-pipeline-private";
CoglContext *ctx =
clutter_backend_get_cogl_context (clutter_get_default_backend ());
CoglPipeline *blur_pipeline;
blur_pipeline =
cogl_context_get_named_pipeline (ctx, &blur_pipeline_key);
if (G_UNLIKELY (blur_pipeline == NULL))
{
CoglSnippet *snippet;
blur_pipeline = cogl_pipeline_new (ctx);
cogl_pipeline_set_layer_null_texture (blur_pipeline, 0);
cogl_pipeline_set_layer_filters (blur_pipeline,
0,
COGL_PIPELINE_FILTER_LINEAR,
COGL_PIPELINE_FILTER_LINEAR);
cogl_pipeline_set_layer_wrap_mode (blur_pipeline,
0,
COGL_PIPELINE_WRAP_MODE_CLAMP_TO_EDGE);
snippet = cogl_snippet_new (COGL_SNIPPET_HOOK_TEXTURE_LOOKUP,
gaussian_blur_glsl_declarations,
NULL);
cogl_snippet_set_replace (snippet, gaussian_blur_glsl);
cogl_pipeline_add_layer_snippet (blur_pipeline, 0, snippet);
cogl_object_unref (snippet);
cogl_context_set_named_pipeline (ctx, &blur_pipeline_key, blur_pipeline);
}
return cogl_pipeline_copy (blur_pipeline);
}
static void
update_blur_uniforms (ClutterBlur *blur,
BlurPass *pass)
{
gboolean vertical = pass->orientation == VERTICAL;
pass->pixel_step_uniform =
cogl_pipeline_get_uniform_location (pass->pipeline, "pixel_step");
if (pass->pixel_step_uniform > -1)
{
float pixel_step;
if (vertical)
pixel_step = 1.f / cogl_texture_get_height (pass->texture);
else
pixel_step = 1.f / cogl_texture_get_width (pass->texture);
cogl_pipeline_set_uniform_1f (pass->pipeline,
pass->pixel_step_uniform,
pixel_step);
}
pass->sigma_uniform =
cogl_pipeline_get_uniform_location (pass->pipeline, "sigma");
if (pass->sigma_uniform > -1)
{
cogl_pipeline_set_uniform_1f (pass->pipeline,
pass->sigma_uniform,
blur->sigma / blur->downscale_factor);
}
pass->vertical_uniform =
cogl_pipeline_get_uniform_location (pass->pipeline, "vertical");
if (pass->vertical_uniform > -1)
{
cogl_pipeline_set_uniform_1i (pass->pipeline,
pass->vertical_uniform,
vertical);
}
}
static gboolean
create_fbo (ClutterBlur *blur,
BlurPass *pass)
{
CoglContext *ctx =
clutter_backend_get_cogl_context (clutter_get_default_backend ());
graphene_matrix_t projection;
float scaled_height;
float scaled_width;
float height;
float width;
g_clear_pointer (&pass->texture, cogl_object_unref);
g_clear_object (&pass->framebuffer);
width = cogl_texture_get_width (blur->source_texture);
height = cogl_texture_get_height (blur->source_texture);
scaled_width = floorf (width / blur->downscale_factor);
scaled_height = floorf (height / blur->downscale_factor);
pass->texture = COGL_TEXTURE (cogl_texture_2d_new_with_size (ctx,
scaled_width,
scaled_height));
if (!pass->texture)
return FALSE;
pass->framebuffer =
COGL_FRAMEBUFFER (cogl_offscreen_new_with_texture (pass->texture));
if (!pass->framebuffer)
{
g_warning ("%s: Unable to create an Offscreen buffer", G_STRLOC);
return FALSE;
}
graphene_matrix_init_translate (&projection,
&GRAPHENE_POINT3D_INIT (-scaled_width / 2.f,
-scaled_height / 2.f,
0.f));
graphene_matrix_scale (&projection,
2.f / scaled_width,
-2.f / scaled_height,
1.f);
cogl_framebuffer_set_projection_matrix (pass->framebuffer, &projection);
return TRUE;
}
static gboolean
setup_blur_pass (ClutterBlur *blur,
BlurPass *pass,
int orientation,
CoglTexture *texture)
{
pass->orientation = orientation;
pass->pipeline = create_blur_pipeline ();
cogl_pipeline_set_layer_texture (pass->pipeline, 0, texture);
if (!create_fbo (blur, pass))
return FALSE;
update_blur_uniforms (blur, pass);
return TRUE;
}
static float
calculate_downscale_factor (float width,
float height,
float sigma)
{
float downscale_factor = 1.f;
float scaled_width = width;
float scaled_height = height;
float scaled_sigma = sigma;
/* This is the algorithm used by Firefox; keep downscaling until either the
* blur radius is lower than the threshold, or the downscaled texture is too
* small.
*/
while (scaled_sigma > MAX_SIGMA &&
scaled_width > MIN_DOWNSCALE_SIZE &&
scaled_height > MIN_DOWNSCALE_SIZE)
{
downscale_factor *= 2.f;
scaled_width = width / downscale_factor;
scaled_height = height / downscale_factor;
scaled_sigma = sigma / downscale_factor;
}
return downscale_factor;
}
static void
apply_blur_pass (BlurPass *pass)
{
CoglColor transparent;
cogl_color_init_from_4ub (&transparent, 0, 0, 0, 0);
cogl_framebuffer_clear (pass->framebuffer,
COGL_BUFFER_BIT_COLOR,
&transparent);
cogl_framebuffer_draw_rectangle (pass->framebuffer,
pass->pipeline,
0, 0,
cogl_texture_get_width (pass->texture),
cogl_texture_get_height (pass->texture));
}
static void
clear_blur_pass (BlurPass *pass)
{
g_clear_pointer (&pass->pipeline, cogl_object_unref);
g_clear_pointer (&pass->texture, cogl_object_unref);
g_clear_object (&pass->framebuffer);
}
/**
* clutter_blur_new:
* @texture: a #CoglTexture
* @sigma: blur sigma
*
* Creates a new #ClutterBlur.
*
* Returns: (transfer full) (nullable): A newly created #ClutterBlur
*/
ClutterBlur *
clutter_blur_new (CoglTexture *texture,
unsigned int sigma)
{
ClutterBlur *blur;
unsigned int height;
unsigned int width;
BlurPass *hpass;
BlurPass *vpass;
width = cogl_texture_get_width (texture);
height = cogl_texture_get_height (texture);
blur = g_new0 (ClutterBlur, 1);
blur->sigma = sigma;
blur->source_texture = cogl_object_ref (texture);
blur->downscale_factor = calculate_downscale_factor (width, height, sigma);
vpass = &blur->pass[VERTICAL];
hpass = &blur->pass[HORIZONTAL];
if (!setup_blur_pass (blur, vpass, VERTICAL, texture) ||
!setup_blur_pass (blur, hpass, HORIZONTAL, vpass->texture))
{
clutter_blur_free (blur);
return NULL;
}
return g_steal_pointer (&blur);
}
/**
* clutter_blur_apply:
* @blur: a #ClutterBlur
*
* Applies the blur. The resulting texture can be retrieved by
* clutter_blur_get_texture().
*/
void
clutter_blur_apply (ClutterBlur *blur)
{
apply_blur_pass (&blur->pass[VERTICAL]);
apply_blur_pass (&blur->pass[HORIZONTAL]);
}
/**
* clutter_blur_get_texture:
* @blur: a #ClutterBlur
*
* Retrieves the texture where the blurred contents are stored. The
* contents are undefined until clutter_blur_apply() is called.
*
* Returns: (transfer none): a #CoglTexture
*/
CoglTexture *
clutter_blur_get_texture (ClutterBlur *blur)
{
return blur->pass[HORIZONTAL].texture;
}
/**
* clutter_blur_free:
* @blur: A #ClutterBlur
*
* Frees @blur.
*/
void
clutter_blur_free (ClutterBlur *blur)
{
g_assert (blur);
clear_blur_pass (&blur->pass[VERTICAL]);
clear_blur_pass (&blur->pass[HORIZONTAL]);
cogl_clear_object (&blur->source_texture);
g_free (blur);
}