1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
|
/*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License
* as published by the Free Software Foundation; either version 2
* of the License, or (at your option) any later version.
*
* This program 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 General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software Foundation,
* Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
*
* Copyright 2021, Blender Foundation.
*/
/** \file
* \ingroup draw_engine
*/
#pragma once
#include "BKE_image_partial_update.hh"
#include "IMB_imbuf_types.h"
#include "BLI_float4x4.hh"
#include "BLI_math_vec_types.hh"
#include "image_batches.hh"
#include "image_private.hh"
#include "image_wrappers.hh"
namespace blender::draw::image_engine {
constexpr float EPSILON_UV_BOUNDS = 0.00001f;
/**
* \brief Screen space method using a single texture spawning the whole screen.
*/
struct OneTextureMethod {
IMAGE_InstanceData *instance_data;
OneTextureMethod(IMAGE_InstanceData *instance_data) : instance_data(instance_data)
{
}
/** \brief Update the texture slot uv and screen space bounds. */
void update_screen_space_bounds(const ARegion *region)
{
/* Create a single texture that covers the visible screen space. */
BLI_rctf_init(
&instance_data->texture_infos[0].clipping_bounds, 0, region->winx, 0, region->winy);
instance_data->texture_infos[0].visible = true;
/* Mark the other textures as invalid. */
for (int i = 1; i < SCREEN_SPACE_DRAWING_MODE_TEXTURE_LEN; i++) {
BLI_rctf_init_minmax(&instance_data->texture_infos[i].clipping_bounds);
instance_data->texture_infos[i].visible = false;
}
}
void update_screen_uv_bounds()
{
for (int i = 0; i < SCREEN_SPACE_DRAWING_MODE_TEXTURE_LEN; i++) {
update_screen_uv_bounds(instance_data->texture_infos[0]);
}
}
void update_screen_uv_bounds(TextureInfo &info)
{
/* Although this works, computing an inverted matrix adds some precision issues and leads to
* tearing artifacts. This should be modified to use the scaling and transformation from the
* not inverted matrix.*/
float4x4 mat(instance_data->ss_to_texture);
float4x4 mat_inv = mat.inverted();
float3 min_uv = mat_inv * float3(0.0f, 0.0f, 0.0f);
float3 max_uv = mat_inv * float3(1.0f, 1.0f, 0.0f);
rctf new_clipping_bounds;
BLI_rctf_init(&new_clipping_bounds, min_uv[0], max_uv[0], min_uv[1], max_uv[1]);
if (!BLI_rctf_compare(&info.clipping_uv_bounds, &new_clipping_bounds, EPSILON_UV_BOUNDS)) {
info.clipping_uv_bounds = new_clipping_bounds;
info.dirty = true;
}
}
};
using namespace blender::bke::image::partial_update;
template<typename TextureMethod> class ScreenSpaceDrawingMode : public AbstractDrawingMode {
private:
DRWPass *create_image_pass() const
{
DRWState state = static_cast<DRWState>(DRW_STATE_WRITE_COLOR | DRW_STATE_DEPTH_ALWAYS |
DRW_STATE_BLEND_ALPHA_PREMUL);
return DRW_pass_create("Image", state);
}
DRWPass *create_depth_pass() const
{
/* Depth is needed for background overlay rendering. Near depth is used for
* transparency checker and Far depth is used for indicating the image size. */
DRWState state = static_cast<DRWState>(DRW_STATE_WRITE_DEPTH | DRW_STATE_DEPTH_LESS_EQUAL);
return DRW_pass_create("Depth", state);
}
void add_shgroups(const IMAGE_InstanceData *instance_data) const
{
const ShaderParameters &sh_params = instance_data->sh_params;
GPUShader *shader = IMAGE_shader_image_get();
DefaultTextureList *dtxl = DRW_viewport_texture_list_get();
DRWShadingGroup *shgrp = DRW_shgroup_create(shader, instance_data->passes.image_pass);
DRW_shgroup_uniform_vec2_copy(shgrp, "farNearDistances", sh_params.far_near);
DRW_shgroup_uniform_vec4_copy(shgrp, "shuffle", sh_params.shuffle);
DRW_shgroup_uniform_int_copy(shgrp, "drawFlags", sh_params.flags);
DRW_shgroup_uniform_bool_copy(shgrp, "imgPremultiplied", sh_params.use_premul_alpha);
DRW_shgroup_uniform_texture(shgrp, "depth_texture", dtxl->depth);
float image_mat[4][4];
unit_m4(image_mat);
for (int i = 0; i < SCREEN_SPACE_DRAWING_MODE_TEXTURE_LEN; i++) {
const TextureInfo &info = instance_data->texture_infos[i];
if (!info.visible) {
continue;
}
DRWShadingGroup *shgrp_sub = DRW_shgroup_create_sub(shgrp);
DRW_shgroup_uniform_texture_ex(shgrp_sub, "imageTexture", info.texture, GPU_SAMPLER_DEFAULT);
DRW_shgroup_call_obmat(shgrp_sub, info.batch, image_mat);
}
}
/**
* \brief add depth drawing calls.
*
* The depth is used to identify if the tile exist or transparent.
*/
void add_depth_shgroups(IMAGE_InstanceData &instance_data,
Image *image,
ImageUser *image_user) const
{
GPUShader *shader = IMAGE_shader_depth_get();
DRWShadingGroup *shgrp = DRW_shgroup_create(shader, instance_data.passes.depth_pass);
float image_mat[4][4];
unit_m4(image_mat);
ImageUser tile_user = {0};
if (image_user) {
tile_user = *image_user;
}
for (int i = 0; i < SCREEN_SPACE_DRAWING_MODE_TEXTURE_LEN; i++) {
const TextureInfo &info = instance_data.texture_infos[i];
if (!info.visible) {
continue;
}
LISTBASE_FOREACH (ImageTile *, image_tile_ptr, &image->tiles) {
const ImageTileWrapper image_tile(image_tile_ptr);
const int tile_x = image_tile.get_tile_x_offset();
const int tile_y = image_tile.get_tile_y_offset();
tile_user.tile = image_tile.get_tile_number();
/* NOTE: `BKE_image_has_ibuf` doesn't work as it fails for render results. That could be a
* bug or a feature. For now we just acquire to determine if there is a texture. */
void *lock;
ImBuf *tile_buffer = BKE_image_acquire_ibuf(image, &tile_user, &lock);
if (tile_buffer == nullptr) {
continue;
}
BKE_image_release_ibuf(image, tile_buffer, lock);
DRWShadingGroup *shsub = DRW_shgroup_create_sub(shgrp);
float4 min_max_uv(tile_x, tile_y, tile_x + 1, tile_y + 1);
DRW_shgroup_uniform_vec4_copy(shsub, "min_max_uv", min_max_uv);
DRW_shgroup_call_obmat(shsub, info.batch, image_mat);
}
}
}
/**
* \brief Update GPUTextures for drawing the image.
*
* GPUTextures that are marked dirty are rebuild. GPUTextures that aren't marked dirty are
* updated with changed region of the image.
*/
void update_textures(IMAGE_InstanceData &instance_data,
Image *image,
ImageUser *image_user) const
{
PartialUpdateChecker<ImageTileData> checker(
image, image_user, instance_data.partial_update.user);
PartialUpdateChecker<ImageTileData>::CollectResult changes = checker.collect_changes();
switch (changes.get_result_code()) {
case ePartialUpdateCollectResult::FullUpdateNeeded:
instance_data.mark_all_texture_slots_dirty();
break;
case ePartialUpdateCollectResult::NoChangesDetected:
break;
case ePartialUpdateCollectResult::PartialChangesDetected:
/* Partial update when wrap repeat is enabled is not supported. */
if (instance_data.flags.do_tile_drawing) {
instance_data.mark_all_texture_slots_dirty();
}
else {
do_partial_update(changes, instance_data);
}
break;
}
do_full_update_for_dirty_textures(instance_data, image_user);
}
void do_partial_update(PartialUpdateChecker<ImageTileData>::CollectResult &iterator,
IMAGE_InstanceData &instance_data) const
{
while (iterator.get_next_change() == ePartialUpdateIterResult::ChangeAvailable) {
/* Quick exit when tile_buffer isn't availble. */
if (iterator.tile_data.tile_buffer == nullptr) {
continue;
}
ensure_float_buffer(*iterator.tile_data.tile_buffer);
const float tile_width = static_cast<float>(iterator.tile_data.tile_buffer->x);
const float tile_height = static_cast<float>(iterator.tile_data.tile_buffer->y);
for (int i = 0; i < SCREEN_SPACE_DRAWING_MODE_TEXTURE_LEN; i++) {
const TextureInfo &info = instance_data.texture_infos[i];
/* Dirty images will receive a full update. No need to do a partial one now. */
if (info.dirty) {
continue;
}
if (!info.visible) {
continue;
}
GPUTexture *texture = info.texture;
const float texture_width = GPU_texture_width(texture);
const float texture_height = GPU_texture_height(texture);
// TODO
// early bound check.
ImageTileWrapper tile_accessor(iterator.tile_data.tile);
float tile_offset_x = static_cast<float>(tile_accessor.get_tile_x_offset());
float tile_offset_y = static_cast<float>(tile_accessor.get_tile_y_offset());
rcti *changed_region_in_texel_space = &iterator.changed_region.region;
rctf changed_region_in_uv_space;
BLI_rctf_init(&changed_region_in_uv_space,
static_cast<float>(changed_region_in_texel_space->xmin) /
static_cast<float>(iterator.tile_data.tile_buffer->x) +
tile_offset_x,
static_cast<float>(changed_region_in_texel_space->xmax) /
static_cast<float>(iterator.tile_data.tile_buffer->x) +
tile_offset_x,
static_cast<float>(changed_region_in_texel_space->ymin) /
static_cast<float>(iterator.tile_data.tile_buffer->y) +
tile_offset_y,
static_cast<float>(changed_region_in_texel_space->ymax) /
static_cast<float>(iterator.tile_data.tile_buffer->y) +
tile_offset_y);
rctf changed_overlapping_region_in_uv_space;
const bool region_overlap = BLI_rctf_isect(&info.clipping_uv_bounds,
&changed_region_in_uv_space,
&changed_overlapping_region_in_uv_space);
if (!region_overlap) {
continue;
}
// convert the overlapping region to texel space and to ss_pixel space...
// TODO: first convert to ss_pixel space as integer based. and from there go back to texel
// space. But perhaps this isn't needed and we could use an extraction offset somehow.
rcti gpu_texture_region_to_update;
BLI_rcti_init(
&gpu_texture_region_to_update,
floor((changed_overlapping_region_in_uv_space.xmin - info.clipping_uv_bounds.xmin) *
texture_width / BLI_rctf_size_x(&info.clipping_uv_bounds)),
floor((changed_overlapping_region_in_uv_space.xmax - info.clipping_uv_bounds.xmin) *
texture_width / BLI_rctf_size_x(&info.clipping_uv_bounds)),
ceil((changed_overlapping_region_in_uv_space.ymin - info.clipping_uv_bounds.ymin) *
texture_height / BLI_rctf_size_y(&info.clipping_uv_bounds)),
ceil((changed_overlapping_region_in_uv_space.ymax - info.clipping_uv_bounds.ymin) *
texture_height / BLI_rctf_size_y(&info.clipping_uv_bounds)));
rcti tile_region_to_extract;
BLI_rcti_init(
&tile_region_to_extract,
floor((changed_overlapping_region_in_uv_space.xmin - tile_offset_x) * tile_width),
floor((changed_overlapping_region_in_uv_space.xmax - tile_offset_x) * tile_width),
ceil((changed_overlapping_region_in_uv_space.ymin - tile_offset_y) * tile_height),
ceil((changed_overlapping_region_in_uv_space.ymax - tile_offset_y) * tile_height));
// Create an image buffer with a size
// extract and scale into an imbuf
const int texture_region_width = BLI_rcti_size_x(&gpu_texture_region_to_update);
const int texture_region_height = BLI_rcti_size_y(&gpu_texture_region_to_update);
ImBuf extracted_buffer;
IMB_initImBuf(
&extracted_buffer, texture_region_width, texture_region_height, 32, IB_rectfloat);
int offset = 0;
ImBuf *tile_buffer = iterator.tile_data.tile_buffer;
for (int y = gpu_texture_region_to_update.ymin; y < gpu_texture_region_to_update.ymax;
y++) {
float yf = y / (float)texture_height;
float v = info.clipping_uv_bounds.ymax * yf + info.clipping_uv_bounds.ymin * (1.0 - yf) -
tile_offset_y;
for (int x = gpu_texture_region_to_update.xmin; x < gpu_texture_region_to_update.xmax;
x++) {
float xf = x / (float)texture_width;
float u = info.clipping_uv_bounds.xmax * xf +
info.clipping_uv_bounds.xmin * (1.0 - xf) - tile_offset_x;
nearest_interpolation_color(tile_buffer,
nullptr,
&extracted_buffer.rect_float[offset * 4],
u * tile_buffer->x,
v * tile_buffer->y);
offset++;
}
}
GPU_texture_update_sub(texture,
GPU_DATA_FLOAT,
extracted_buffer.rect_float,
gpu_texture_region_to_update.xmin,
gpu_texture_region_to_update.ymin,
0,
extracted_buffer.x,
extracted_buffer.y,
0);
imb_freerectImbuf_all(&extracted_buffer);
}
}
}
void do_full_update_for_dirty_textures(IMAGE_InstanceData &instance_data,
const ImageUser *image_user) const
{
for (int i = 0; i < SCREEN_SPACE_DRAWING_MODE_TEXTURE_LEN; i++) {
TextureInfo &info = instance_data.texture_infos[i];
if (!info.dirty) {
continue;
}
if (!info.visible) {
continue;
}
do_full_update_gpu_texture(info, instance_data, image_user);
}
}
void do_full_update_gpu_texture(TextureInfo &info,
IMAGE_InstanceData &instance_data,
const ImageUser *image_user) const
{
ImBuf texture_buffer;
const int texture_width = GPU_texture_width(info.texture);
const int texture_height = GPU_texture_height(info.texture);
IMB_initImBuf(&texture_buffer, texture_width, texture_height, 0, IB_rectfloat);
ImageUser tile_user = {0};
if (image_user) {
tile_user = *image_user;
}
void *lock;
Image *image = instance_data.image;
LISTBASE_FOREACH (ImageTile *, image_tile_ptr, &image->tiles) {
const ImageTileWrapper image_tile(image_tile_ptr);
tile_user.tile = image_tile.get_tile_number();
ImBuf *tile_buffer = BKE_image_acquire_ibuf(image, &tile_user, &lock);
if (tile_buffer == nullptr) {
/* Couldn't load the image buffer of the tile. */
continue;
}
do_full_update_texture_slot(instance_data, info, texture_buffer, *tile_buffer, image_tile);
BKE_image_release_ibuf(image, tile_buffer, lock);
}
GPU_texture_update(info.texture, GPU_DATA_FLOAT, texture_buffer.rect_float);
imb_freerectImbuf_all(&texture_buffer);
}
/**
* \brief Ensure that the float buffer of the given image buffer is available.
*
* Returns true when a float buffer was created. Somehow the VSE cache increases the ref
* counter, but might use a different mechanism for destructing the image, that doesn't free the
* rect_float as the refcounter isn't 0. To work around this we destruct any created local
* buffers ourself.
*/
bool ensure_float_buffer(ImBuf &image_buffer) const
{
if (image_buffer.rect_float == nullptr) {
IMB_float_from_rect(&image_buffer);
return true;
}
return false;
}
void do_full_update_texture_slot(const IMAGE_InstanceData &instance_data,
const TextureInfo &texture_info,
ImBuf &texture_buffer,
ImBuf &tile_buffer,
const ImageTileWrapper &image_tile) const
{
const int texture_width = texture_buffer.x;
const int texture_height = texture_buffer.y;
const bool float_buffer_created = ensure_float_buffer(tile_buffer);
/* TODO(jbakker): Find leak when rendering VSE and don't free here. */
const bool do_free_float_buffer = float_buffer_created &&
instance_data.image->type == IMA_TYPE_R_RESULT;
/* IMB_transform works in a non-consistent space. This should be documented or fixed!.
* Construct a variant of the info_uv_to_texture that adds the texel space
* transformation.*/
float uv_to_texel[4][4];
copy_m4_m4(uv_to_texel, instance_data.ss_to_texture);
float scale[3] = {static_cast<float>(texture_width) / static_cast<float>(tile_buffer.x),
static_cast<float>(texture_height) / static_cast<float>(tile_buffer.y),
1.0f};
rescale_m4(uv_to_texel, scale);
uv_to_texel[3][0] += image_tile.get_tile_x_offset() /
BLI_rctf_size_x(&texture_info.clipping_uv_bounds);
uv_to_texel[3][1] += image_tile.get_tile_y_offset() /
BLI_rctf_size_y(&texture_info.clipping_uv_bounds);
uv_to_texel[3][0] *= texture_width;
uv_to_texel[3][1] *= texture_height;
invert_m4(uv_to_texel);
rctf crop_rect;
rctf *crop_rect_ptr = nullptr;
eIMBTransformMode transform_mode;
if (instance_data.flags.do_tile_drawing) {
transform_mode = IMB_TRANSFORM_MODE_WRAP_REPEAT;
}
else {
BLI_rctf_init(&crop_rect, 0.0, tile_buffer.x, 0.0, tile_buffer.y);
crop_rect_ptr = &crop_rect;
transform_mode = IMB_TRANSFORM_MODE_CROP_SRC;
}
IMB_transform(&tile_buffer,
&texture_buffer,
transform_mode,
IMB_FILTER_NEAREST,
uv_to_texel,
crop_rect_ptr);
if (do_free_float_buffer) {
imb_freerectfloatImBuf(&tile_buffer);
}
}
public:
void cache_init(IMAGE_Data *vedata) const override
{
IMAGE_InstanceData *instance_data = vedata->instance_data;
instance_data->passes.image_pass = create_image_pass();
instance_data->passes.depth_pass = create_depth_pass();
}
void cache_image(IMAGE_Data *vedata, Image *image, ImageUser *iuser) const override
{
const DRWContextState *draw_ctx = DRW_context_state_get();
IMAGE_InstanceData *instance_data = vedata->instance_data;
TextureMethod method(instance_data);
instance_data->partial_update.ensure_image(image);
instance_data->clear_dirty_flag();
// Step: Find out which screen space textures are needed to draw on the screen. Remove the
// screen space textures that aren't needed.
const ARegion *region = draw_ctx->region;
method.update_screen_space_bounds(region);
method.update_screen_uv_bounds();
// Step: Update the GPU textures based on the changes in the image.
instance_data->update_gpu_texture_allocations();
update_textures(*instance_data, image, iuser);
// Step: Add the GPU textures to the shgroup.
instance_data->update_batches();
add_depth_shgroups(*instance_data, image, iuser);
add_shgroups(instance_data);
}
void draw_finish(IMAGE_Data *UNUSED(vedata)) const override
{
}
void draw_scene(IMAGE_Data *vedata) const override
{
IMAGE_InstanceData *instance_data = vedata->instance_data;
DefaultFramebufferList *dfbl = DRW_viewport_framebuffer_list_get();
GPU_framebuffer_bind(dfbl->default_fb);
static float clear_col[4] = {0.0f, 0.0f, 0.0f, 0.0f};
GPU_framebuffer_clear_color_depth(dfbl->default_fb, clear_col, 1.0);
DRW_view_set_active(instance_data->view);
DRW_draw_pass(instance_data->passes.depth_pass);
GPU_framebuffer_bind(dfbl->color_only_fb);
DRW_draw_pass(instance_data->passes.image_pass);
DRW_view_set_active(nullptr);
GPU_framebuffer_bind(dfbl->default_fb);
}
}; // namespace clipping
} // namespace blender::draw::image_engine
|