mutter/cogl/cogl-meta-texture.c

637 lines
21 KiB
C
Raw Normal View History

/*
* Cogl
*
* An object oriented GL/GLES Abstraction/Utility Layer
*
* Copyright (C) 2011 Intel Corporation.
*
* 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/>.
*
*
* Authors:
* Robert Bragg <robert@linux.intel.com>
*/
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
#include "cogl-texture.h"
#include "cogl-matrix.h"
#include "cogl-spans.h"
#include "cogl-meta-texture.h"
#include "cogl-texture-rectangle-private.h"
#include <string.h>
#include <math.h>
typedef struct _ForeachData
{
float meta_region_coords[4];
CoglPipelineWrapMode wrap_s;
CoglPipelineWrapMode wrap_t;
CoglMetaTextureCallback callback;
void *user_data;
int width;
int height;
CoglTexture *padded_textures[9];
const float *grid_slice_texture_coords;
float slice_offset_s;
float slice_offset_t;
float slice_range_s;
float slice_range_t;
} ForeachData;
static void
padded_grid_repeat_cb (CoglTexture *slice_texture,
const float *slice_texture_coords,
const float *meta_coords,
void *user_data)
{
ForeachData *data;
float mapped_coords[4];
/* Ignore padding slices for the current grid */
if (!slice_texture)
return;
data = user_data;
/* NB: the slice_texture_coords[] we get here will always be
* normalized.
*
* We now need to map the normalized slice_texture_coords[] we have
* here back to the real slice coordinates we saved in the previous
* stage...
*/
mapped_coords[0] =
slice_texture_coords[0] * data->slice_range_s + data->slice_offset_s;
mapped_coords[1] =
slice_texture_coords[1] * data->slice_range_t + data->slice_offset_t;
mapped_coords[2] =
slice_texture_coords[2] * data->slice_range_s + data->slice_offset_s;
mapped_coords[3] =
slice_texture_coords[3] * data->slice_range_t + data->slice_offset_t;
data->callback (slice_texture,
mapped_coords, meta_coords, data->user_data);
}
static int
setup_padded_spans (CoglSpan *spans,
float start,
float end,
float range,
int *real_index)
{
int span_index = 0;
if (start > 0)
{
spans[0].start = 0;
spans[0].size = start;
spans[0].waste = 0;
span_index++;
spans[1].start = spans[0].size;
}
else
spans[span_index].start = 0;
spans[span_index].size = end - start;
spans[span_index].waste = 0;
*real_index = span_index;
span_index++;
if (end < range)
{
spans[span_index].start =
spans[span_index - 1].start + spans[span_index - 1].size;
spans[span_index].size = range - end;
spans[span_index].waste = 0;
span_index++;
}
return span_index;
}
/* This handles each sub-texture within the range [0,1] of our
* original meta texture and repeats each one separately across the
* users requested virtual texture coordinates.
*
* A notable advantage of this approach is that we will batch
* together callbacks corresponding to the same underlying slice
* together.
*/
Add -Wmissing-declarations to maintainer flags and fix problems This option to GCC makes it give a warning whenever a global function is defined without a declaration. This should catch cases were we've defined a function but forgot to put it in a header. In that case it is either only used within one file so we should make it static or we should declare it in a header. The following changes where made to fix problems: • Some functions were made static • cogl-path.h (the one containing the 1.0 API) was split into two files, one defining the functions and one defining the enums so that cogl-path.c can include the enum and function declarations from the 2.0 API as well as the function declarations from the 1.0 API. • cogl2-clip-state has been removed. This only had one experimental function called cogl_clip_push_from_path but as this is unstable we might as well remove it favour of the equivalent cogl_framebuffer_* API. • The GLX, SDL and WGL winsys's now have a private header to define their get_vtable function instead of directly declaring in the C file where it is called. • All places that were calling COGL_OBJECT_DEFINE need to have the cogl_is_whatever function declared so these have been added either as a public function or in a private header. • Some files that were not including the header containing their function declarations have been fixed to do so. • Any unused error quark functions have been removed. If we later want them we should add them back one by one and add a declaration for them in a header. • _cogl_is_framebuffer has been renamed to cogl_is_framebuffer and made a public function with a declaration in cogl-framebuffer.h • Similarly for CoglOnscreen. • cogl_vdraw_indexed_attributes is called cogl_framebuffer_vdraw_indexed_attributes in the header. The definition has been changed to match the header. • cogl_index_buffer_allocate has been removed. This had no declaration and I'm not sure what it's supposed to do. • CoglJournal has been changed to use the internal CoglObject macro so that it won't define an exported cogl_is_journal symbol. • The _cogl_blah_pointer_from_handle functions have been removed. CoglHandle isn't used much anymore anyway and in the few places where it is used I think it's safe to just use the implicit cast from void* to the right type. • The test-utils.h header for the conformance tests explicitly disables the -Wmissing-declaration option using a pragma because all of the tests declare their main function without a header. Any mistakes relating to missing declarations aren't really important for the tests. • cogl_quaternion_init_from_quaternion and init_from_matrix have been given declarations in cogl-quaternion.h Reviewed-by: Robert Bragg <robert@linux.intel.com>
2012-03-06 13:21:28 -05:00
static void
create_grid_and_repeat_cb (CoglTexture *slice_texture,
const float *slice_texture_coords,
const float *meta_coords,
void *user_data)
{
ForeachData *data = user_data;
CoglSpan x_spans[3];
int n_x_spans;
int x_real_index;
CoglSpan y_spans[3];
int n_y_spans;
int y_real_index;
/* NB: This callback is called for each slice of the meta-texture
* in the range [0,1].
*
* We define a "padded grid" for each slice of the meta-texture in
* the range [0,1]. The x axis and y axis grid lines are defined
* using CoglSpans.
*
* The padded grid maps over the meta-texture coordinates in the
* range [0,1] but only contains one valid cell that corresponds to
* current slice being iterated and all the surrounding cells just
* provide padding.
*
* Once we've defined our padded grid we then repeat that across
* the user's original region, calling their callback whenever
* we see our current slice - ignoring padding.
*
* NB: we can assume meta_coords[] are normalized at this point
* since TextureRectangles aren't iterated with this code-path.
*
* NB: spans are always defined using non-normalized coordinates
*/
n_x_spans = setup_padded_spans (x_spans,
meta_coords[0] * data->width,
meta_coords[2] * data->width,
data->width,
&x_real_index);
n_y_spans = setup_padded_spans (y_spans,
meta_coords[1] * data->height,
meta_coords[3] * data->height,
data->height,
&y_real_index);
data->padded_textures[n_y_spans * y_real_index + x_real_index] =
slice_texture;
/* Our callback is going to be passed normalized slice texture
* coordinates, and we will need to map the range [0,1] to the real
* slice_texture_coords we have here... */
data->grid_slice_texture_coords = slice_texture_coords;
data->slice_range_s = fabs (data->grid_slice_texture_coords[2] -
data->grid_slice_texture_coords[0]);
data->slice_range_t = fabs (data->grid_slice_texture_coords[3] -
data->grid_slice_texture_coords[1]);
data->slice_offset_s = MIN (data->grid_slice_texture_coords[0],
data->grid_slice_texture_coords[2]);
data->slice_offset_t = MIN (data->grid_slice_texture_coords[1],
data->grid_slice_texture_coords[3]);
/* Now actually iterate the region the user originally requested
* using the current padded grid */
_cogl_texture_spans_foreach_in_region (x_spans,
n_x_spans,
y_spans,
n_y_spans,
data->padded_textures,
data->meta_region_coords,
data->width,
data->height,
data->wrap_s,
data->wrap_t,
padded_grid_repeat_cb,
data);
/* Clear the padded_textures ready for the next iteration */
data->padded_textures[n_y_spans * y_real_index + x_real_index] = NULL;
}
#define SWAP(A,B) do { float tmp = B; B = A; A = tmp; } while (0)
typedef struct _ClampData
{
float start;
float end;
CoglBool s_flipped;
CoglBool t_flipped;
CoglMetaTextureCallback callback;
void *user_data;
} ClampData;
static void
clamp_s_cb (CoglTexture *sub_texture,
const float *sub_texture_coords,
const float *meta_coords,
void *user_data)
{
ClampData *clamp_data = user_data;
float mapped_meta_coords[4] = {
clamp_data->start,
meta_coords[1],
clamp_data->end,
meta_coords[3]
};
if (clamp_data->s_flipped)
SWAP (mapped_meta_coords[0], mapped_meta_coords[2]);
/* NB: we never need to flip the t coords when dealing with
* s-axis clamping so no need to check if ->t_flipped */
clamp_data->callback (sub_texture,
sub_texture_coords, mapped_meta_coords,
clamp_data->user_data);
}
static void
clamp_t_cb (CoglTexture *sub_texture,
const float *sub_texture_coords,
const float *meta_coords,
void *user_data)
{
ClampData *clamp_data = user_data;
float mapped_meta_coords[4] = {
meta_coords[0],
clamp_data->start,
meta_coords[2],
clamp_data->end,
};
if (clamp_data->s_flipped)
SWAP (mapped_meta_coords[0], mapped_meta_coords[2]);
if (clamp_data->t_flipped)
SWAP (mapped_meta_coords[1], mapped_meta_coords[3]);
clamp_data->callback (sub_texture,
sub_texture_coords, mapped_meta_coords,
clamp_data->user_data);
}
static CoglBool
foreach_clamped_region (CoglMetaTexture *meta_texture,
float *tx_1,
float *ty_1,
float *tx_2,
float *ty_2,
CoglPipelineWrapMode wrap_s,
CoglPipelineWrapMode wrap_t,
CoglMetaTextureCallback callback,
void *user_data)
{
float width = cogl_texture_get_width (COGL_TEXTURE (meta_texture));
ClampData clamp_data;
/* Consider that *tx_1 may be > *tx_2 and to simplify things
* we just flip them around if that's the case and keep a note
* of the fact that they are flipped. */
if (*tx_1 > *tx_2)
{
SWAP (*tx_1, *tx_2);
clamp_data.s_flipped = TRUE;
}
else
clamp_data.s_flipped = FALSE;
/* The same goes for ty_1 and ty_2... */
if (*ty_1 > *ty_2)
{
SWAP (*ty_1, *ty_2);
clamp_data.t_flipped = TRUE;
}
else
clamp_data.t_flipped = FALSE;
clamp_data.callback = callback;
clamp_data.user_data = user_data;
if (wrap_s == COGL_PIPELINE_WRAP_MODE_CLAMP_TO_EDGE)
{
float max_s_coord;
float half_texel_width;
/* Consider that rectangle textures have non-normalized
* coordinates... */
if (cogl_is_texture_rectangle (meta_texture))
max_s_coord = width;
else
max_s_coord = 1.0;
half_texel_width = max_s_coord / (width * 2);
/* Handle any left clamped region */
if (*tx_1 < 0)
{
clamp_data.start = *tx_1;
clamp_data.end = MIN (0, *tx_2);;
cogl_meta_texture_foreach_in_region (meta_texture,
half_texel_width, *ty_1,
half_texel_width, *ty_2,
COGL_PIPELINE_WRAP_MODE_REPEAT,
wrap_t,
clamp_s_cb,
&clamp_data);
/* Have we handled everything? */
if (tx_2 <= 0)
return TRUE;
/* clamp tx_1 since we've handled everything with x < 0 */
*tx_1 = 0;
}
/* Handle any right clamped region - including the corners */
if (*tx_2 > max_s_coord)
{
clamp_data.start = MAX (max_s_coord, *tx_1);
clamp_data.end = *tx_2;
cogl_meta_texture_foreach_in_region (meta_texture,
max_s_coord - half_texel_width,
*ty_1,
max_s_coord - half_texel_width,
*ty_2,
COGL_PIPELINE_WRAP_MODE_REPEAT,
wrap_t,
clamp_s_cb,
&clamp_data);
/* Have we handled everything? */
if (*tx_1 >= max_s_coord)
return TRUE;
/* clamp tx_2 since we've handled everything with x >
* max_s_coord */
*tx_2 = max_s_coord;
}
}
if (wrap_t == COGL_PIPELINE_WRAP_MODE_CLAMP_TO_EDGE)
{
float height = cogl_texture_get_height (COGL_TEXTURE (meta_texture));
float max_t_coord;
float half_texel_height;
/* Consider that rectangle textures have non-normalized
* coordinates... */
if (cogl_is_texture_rectangle (meta_texture))
max_t_coord = height;
else
max_t_coord = 1.0;
half_texel_height = max_t_coord / (height * 2);
/* Handle any top clamped region */
if (*ty_1 < 0)
{
clamp_data.start = *ty_1;
clamp_data.end = MIN (0, *ty_2);;
cogl_meta_texture_foreach_in_region (meta_texture,
*tx_1, half_texel_height,
*tx_2, half_texel_height,
wrap_s,
COGL_PIPELINE_WRAP_MODE_REPEAT,
clamp_t_cb,
&clamp_data);
/* Have we handled everything? */
if (tx_2 <= 0)
return TRUE;
/* clamp ty_1 since we've handled everything with y < 0 */
*ty_1 = 0;
}
/* Handle any bottom clamped region */
if (*ty_2 > max_t_coord)
{
clamp_data.start = MAX (max_t_coord, *ty_1);;
clamp_data.end = *ty_2;
cogl_meta_texture_foreach_in_region (meta_texture,
*tx_1,
max_t_coord - half_texel_height,
*tx_2,
max_t_coord - half_texel_height,
wrap_s,
COGL_PIPELINE_WRAP_MODE_REPEAT,
clamp_t_cb,
&clamp_data);
/* Have we handled everything? */
if (*ty_1 >= max_t_coord)
return TRUE;
/* clamp ty_2 since we've handled everything with y >
* max_t_coord */
*ty_2 = max_t_coord;
}
}
if (clamp_data.s_flipped)
SWAP (*tx_1, *tx_2);
if (clamp_data.t_flipped)
SWAP (*ty_1, *ty_2);
return FALSE;
}
typedef struct _NormalizeData
{
CoglMetaTextureCallback callback;
void *user_data;
float s_normalize_factor;
float t_normalize_factor;
} NormalizeData;
static void
normalize_meta_coords_cb (CoglTexture *slice_texture,
const float *slice_coords,
const float *meta_coords,
void *user_data)
{
NormalizeData *data = user_data;
float normalized_meta_coords[4] = {
meta_coords[0] * data->s_normalize_factor,
meta_coords[1] * data->t_normalize_factor,
meta_coords[2] * data->s_normalize_factor,
meta_coords[3] * data->t_normalize_factor
};
data->callback (slice_texture,
slice_coords, normalized_meta_coords,
data->user_data);
}
typedef struct _UnNormalizeData
{
CoglMetaTextureCallback callback;
void *user_data;
float width;
float height;
} UnNormalizeData;
static void
un_normalize_slice_coords_cb (CoglTexture *slice_texture,
const float *slice_coords,
const float *meta_coords,
void *user_data)
{
UnNormalizeData *data = user_data;
float un_normalized_slice_coords[4] = {
slice_coords[0] * data->width,
slice_coords[1] * data->height,
slice_coords[2] * data->width,
slice_coords[3] * data->height
};
data->callback (slice_texture,
un_normalized_slice_coords, meta_coords,
data->user_data);
}
void
cogl_meta_texture_foreach_in_region (CoglMetaTexture *meta_texture,
float tx_1,
float ty_1,
float tx_2,
float ty_2,
CoglPipelineWrapMode wrap_s,
CoglPipelineWrapMode wrap_t,
CoglMetaTextureCallback callback,
void *user_data)
{
CoglTexture *texture = COGL_TEXTURE (meta_texture);
float width = cogl_texture_get_width (texture);
float height = cogl_texture_get_height (texture);
NormalizeData normalize_data;
if (wrap_s == COGL_PIPELINE_WRAP_MODE_AUTOMATIC)
wrap_s = COGL_PIPELINE_WRAP_MODE_CLAMP_TO_EDGE;
if (wrap_t == COGL_PIPELINE_WRAP_MODE_AUTOMATIC)
wrap_t = COGL_PIPELINE_WRAP_MODE_CLAMP_TO_EDGE;
if (wrap_s == COGL_PIPELINE_WRAP_MODE_CLAMP_TO_EDGE ||
wrap_t == COGL_PIPELINE_WRAP_MODE_CLAMP_TO_EDGE)
{
CoglBool finished = foreach_clamped_region (meta_texture,
&tx_1, &ty_1, &tx_2, &ty_2,
wrap_s, wrap_t,
callback,
user_data);
if (finished)
return;
/* Since clamping has been handled we now want to normalize our
* wrap modes we se can assume from this point on we don't
* need to consider CLAMP_TO_EDGE. (NB: The spans code will
* assert that CLAMP_TO_EDGE isn't requested) */
if (wrap_s == COGL_PIPELINE_WRAP_MODE_CLAMP_TO_EDGE)
wrap_s = COGL_PIPELINE_WRAP_MODE_REPEAT;
if (wrap_t == COGL_PIPELINE_WRAP_MODE_CLAMP_TO_EDGE)
wrap_t = COGL_PIPELINE_WRAP_MODE_REPEAT;
}
/* It makes things simpler to deal with non-normalized region
* coordinates beyond this point and only re-normalize just before
* calling the user's callback... */
if (!cogl_is_texture_rectangle (COGL_TEXTURE (meta_texture)))
{
normalize_data.callback = callback;
normalize_data.user_data = user_data;
normalize_data.s_normalize_factor = 1.0f / width;
normalize_data.t_normalize_factor = 1.0f / height;
callback = normalize_meta_coords_cb;
user_data = &normalize_data;
tx_1 *= width;
ty_1 *= height;
tx_2 *= width;
ty_2 *= height;
}
/* XXX: at some point this wont be routed through the CoglTexture
* vtable, instead there will be a separate CoglMetaTexture
* interface vtable. */
if (texture->vtable->foreach_sub_texture_in_region)
{
ForeachData data;
data.meta_region_coords[0] = tx_1;
data.meta_region_coords[1] = ty_1;
data.meta_region_coords[2] = tx_2;
data.meta_region_coords[3] = ty_2;
data.wrap_s = wrap_s;
data.wrap_t = wrap_t;
data.callback = callback;
data.user_data = user_data;
data.width = width;
data.height = height;
memset (data.padded_textures, 0, sizeof (data.padded_textures));
/*
* 1) We iterate all the slices of the meta-texture only within
* the range [0,1].
*
* 2) We define a "padded grid" for each slice of the
* meta-texture in the range [0,1].
*
* The padded grid maps over the meta-texture coordinates in
* the range [0,1] but only contains one valid cell that
* corresponds to current slice being iterated and all the
* surrounding cells just provide padding.
*
* 3) Once we've defined our padded grid we then repeat that
* across the user's original region, calling their callback
* whenever we see our current slice - ignoring padding.
*
* A notable benefit of this design is that repeating a texture
* made of multiple slices will result in us repeating each
* slice in-turn so the user gets repeat callbacks for the same
* texture batched together. For manual emulation of texture
* repeats done by drawing geometry this makes it more likely
* that we can batch geometry.
*/
texture->vtable->foreach_sub_texture_in_region (texture,
0, 0, 1, 1,
create_grid_and_repeat_cb,
&data);
}
else
{
CoglSpan x_span = { 0, width, 0 };
CoglSpan y_span = { 0, height, 0 };
float meta_region_coords[4] = { tx_1, ty_1, tx_2, ty_2 };
UnNormalizeData un_normalize_data;
/* If we are dealing with a CoglTextureRectangle then we need a shim
* callback that un_normalizes the slice coordinates we get from
* _cogl_texture_spans_foreach_in_region before passing them to
* the user's callback. */
if (cogl_is_texture_rectangle (meta_texture))
{
un_normalize_data.callback = callback;
un_normalize_data.user_data = user_data;
un_normalize_data.width = width;
un_normalize_data.height = height;
callback = un_normalize_slice_coords_cb;
user_data = &un_normalize_data;
}
_cogl_texture_spans_foreach_in_region (&x_span, 1,
&y_span, 1,
&texture,
meta_region_coords,
width,
height,
wrap_s,
wrap_t,
callback,
user_data);
}
}
#undef SWAP