Subversion Repository Public Repository

Divide-Framework

This repository has no backups
This repository's network speed is throttled to 100KB/sec

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
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
#include "config.h"

#include "Headers/GFXDevice.h"

#include "Core/Headers/TaskPool.h"
#include "Core/Headers/ParamHandler.h"
#include "Core/Time/Headers/ProfileTimer.h"
#include "Managers/Headers/SceneManager.h"
#include "Managers/Headers/RenderPassManager.h"

#include "Rendering/Headers/Renderer.h"
#include "Rendering/PostFX/Headers/PostFX.h"
#include "Rendering/Camera/Headers/FreeFlyCamera.h"

#include "Platform/Video/Headers/IMPrimitive.h"
#include "Platform/Video/Headers/RenderStateBlock.h"
#include "Platform/Video/Textures/Headers/Texture.h"
#include "Platform/Video/Shaders/Headers/ShaderProgram.h"
#include "Platform/Video/Buffers/ShaderBuffer/Headers/ShaderBuffer.h"

namespace Divide {

GFXDevice::GFXDevice()
    : _api(nullptr), 
    _commandPool(Config::MAX_DRAW_COMMANDS_IN_FLIGHT),
    _renderStage(RenderStage::DISPLAY),
    _prevRenderStage(RenderStage::COUNT),
    _commandBuildTimer(Time::ADD_TIMER("Command Generation Timer"))
{
    // Hash values
    _state2DRenderingHash = 0;
    _defaultStateBlockHash = 0;
    _defaultStateNoDepthHash = 0;
    _stateDepthOnlyRenderingHash = 0;
    // Pointers
    _axisGizmo = nullptr;
    _gfxDataBuffer = nullptr;
    _HIZConstructProgram = nullptr;
    _HIZCullProgram = nullptr;
    _renderTargetDraw = nullptr;
    _previewDepthMapShader = nullptr;
    _displayShader = nullptr;
    _activeRenderTarget = nullptr;
    // Integers
    FRAME_COUNT = 0;
    FRAME_DRAW_CALLS = 0;
    FRAME_DRAW_CALLS_PREV = FRAME_DRAW_CALLS;
    _graphicResources = 0;
    // Floats
    _interpolationFactor = 1.0;
    // Cameras
    _2DCamera = nullptr;
    _cubeCamera = nullptr;
    _dualParaboloidCamera = nullptr;
    // Booleans
    _2DRendering = false;
    _drawDebugAxis = false;
    _viewportUpdate = false;
    _isPrePassStage = false;
    // Enumerated Types
    _shadowDetailLevel = RenderDetailLevel::HIGH;
    _GPUVendor = GPUVendor::COUNT;
    _GPURenderer = GPURenderer::COUNT;
    _API_ID = RenderAPI::COUNT;
    // Clipping planes
    _clippingPlanes.resize(Config::MAX_CLIP_PLANES, Plane<F32>(0, 0, 0, 0));
    // To allow calls to "setBaseViewport"
    _viewport.push(vec4<I32>(-1));

    _lastCommandCount.fill(0);
    _lastNodeCount.fill(0);
    // Red X-axis
    _axisLines.push_back(
        Line(VECTOR3_ZERO, WORLD_X_AXIS * 2, vec4<U8>(255, 0, 0, 255), 3.0f));
    // Green Y-axis
    _axisLines.push_back(
        Line(VECTOR3_ZERO, WORLD_Y_AXIS * 2, vec4<U8>(0, 255, 0, 255), 3.0f));
    // Blue Z-axis
    _axisLines.push_back(
        Line(VECTOR3_ZERO, WORLD_Z_AXIS * 2, vec4<U8>(0, 0, 255, 255), 3.0f));

    VertexBuffer::AttribFlags flags;
    flags.fill(true);
    VertexBuffer::setAttribMasks(flags);

    // Don't (currently) need these for shadow passes
    flags[to_const_uint(VertexBuffer::VertexAttribute::ATTRIB_COLOR)] = false;
    flags[to_const_uint(VertexBuffer::VertexAttribute::ATTRIB_TANGENT)] = false;
    VertexBuffer::setAttribMask(RenderStage::Z_PRE_PASS, flags);
    flags[to_const_uint(VertexBuffer::VertexAttribute::ATTRIB_NORMAL)] = false;
    VertexBuffer::setAttribMask(RenderStage::SHADOW, flags);
}

GFXDevice::~GFXDevice()
{
}

/// Generate a cube texture and store it in the provided RenderTarget
void GFXDevice::generateCubeMap(RenderTarget& cubeMap,
                                const U32 arrayOffset,
                                const vec3<F32>& pos,
                                const vec2<F32>& zPlanes,
                                RenderStage renderStage,
                                U32 passIndex) {
    // Only the first colour attachment or the depth attachment is used for now
    // and it must be a cube map texture
    const RTAttachment& colourAttachment = cubeMap.getAttachment(RTAttachment::Type::Colour, 0, false);
    const RTAttachment& depthAttachment = cubeMap.getAttachment(RTAttachment::Type::Depth, 0, false);
    // Colour attachment takes precedent over depth attachment
    bool hasColour = colourAttachment.used();
    bool hasDepth = depthAttachment.used();
    // Everyone's innocent until proven guilty
    bool isValidFB = true;
    if (hasColour) {
        // We only need the colour attachment
        isValidFB = (colourAttachment.descriptor().isCubeTexture());
    } else {
        // We don't have a colour attachment, so we require a cube map depth
        // attachment
        isValidFB = hasDepth && depthAttachment.descriptor().isCubeTexture();
    }
    // Make sure we have a proper render target to draw to
    if (!isValidFB) {
        // Future formats must be added later (e.g. cube map arrays)
        Console::errorfn(Locale::get(_ID("ERROR_GFX_DEVICE_INVALID_FB_CUBEMAP")));
        return;
    }
    // No dual-paraboloid rendering here. Just draw once for each face.
    static vec3<F32> TabUp[6] = {WORLD_Y_NEG_AXIS, WORLD_Y_NEG_AXIS,
                                 WORLD_Z_AXIS,     WORLD_Z_NEG_AXIS,
                                 WORLD_Y_NEG_AXIS, WORLD_Y_NEG_AXIS};
    // Get the center and up vectors for each cube face
    vec3<F32> TabCenter[6] = {vec3<F32>(pos.x + 1.0f, pos.y, pos.z),
                              vec3<F32>(pos.x - 1.0f, pos.y, pos.z),
                              vec3<F32>(pos.x, pos.y + 1.0f, pos.z),
                              vec3<F32>(pos.x, pos.y - 1.0f, pos.z),
                              vec3<F32>(pos.x, pos.y, pos.z + 1.0f),
                              vec3<F32>(pos.x, pos.y, pos.z - 1.0f)};

    Kernel& kernel = Application::instance().kernel();
    // Set a 90 degree vertical FoV perspective projection
    _cubeCamera->setProjection(1.0f, 90.0f, zPlanes);
    // Set the cube camera as the currently active one
    kernel.getCameraMgr().pushActiveCamera(_cubeCamera);
    // Set the desired render stage, remembering the previous one
    RenderStage prevRenderStage = setRenderStage(renderStage);
    // Enable our render target
    cubeMap.begin(RenderTarget::defaultPolicy());
    // For each of the environment's faces (TOP, DOWN, NORTH, SOUTH, EAST, WEST)

    RenderPassManager& passMgr = RenderPassManager::instance();
    RenderPassManager::PassParams params;
    params.doPrePass = renderStage != RenderStage::SHADOW;
    params.occlusionCull = params.doPrePass;
    params.camera = _cubeCamera;
    params.stage = renderStage;

    for (U8 i = 0; i < 6; ++i) {
        // Draw to the current cubemap face
        cubeMap.drawToFace(hasColour ? RTAttachment::Type::Colour
                                     : RTAttachment::Type::Depth,
                           0,
                           i + arrayOffset);
        // Point our camera to the correct face
        _cubeCamera->lookAt(pos, TabCenter[i], TabUp[i]);
        // Pass our render function to the renderer
        params.pass = passIndex + i;
        passMgr.doCustomPass(params);
    }

    // Resolve our render target
    cubeMap.end();
    // Return to our previous rendering stage
    setRenderStage(prevRenderStage);
    // Restore our previous camera
    kernel.getCameraMgr().popActiveCamera();
}

void GFXDevice::generateDualParaboloidMap(RenderTarget& targetBuffer,
                                          const U32 arrayOffset,
                                          const vec3<F32>& pos,
                                          const vec2<F32>& zPlanes,
                                          RenderStage renderStage,
                                          U32 passIndex)
{
    const RTAttachment& colourAttachment = targetBuffer.getAttachment(RTAttachment::Type::Colour, 0, false);
    const RTAttachment& depthAttachment = targetBuffer.getAttachment(RTAttachment::Type::Depth, 0, false);
    // Colour attachment takes precedent over depth attachment
    bool hasColour = colourAttachment.used();
    bool hasDepth = depthAttachment.used();
    bool isValidFB = true;
    if (hasColour) {
        // We only need the colour attachment
        isValidFB = colourAttachment.descriptor().isArrayTexture();
    } else {
        // We don't have a colour attachment, so we require a cube map depth attachment
        isValidFB = hasDepth && depthAttachment.descriptor().isArrayTexture();
    }
    // Make sure we have a proper render target to draw to
    if (!isValidFB) {
        // Future formats must be added later (e.g. cube map arrays)
        Console::errorfn(Locale::get(_ID("ERROR_GFX_DEVICE_INVALID_FB_DP")));
        return;
    }
    Kernel& kernel = Application::instance().kernel();
    // Set a 90 degree vertical FoV perspective projection
    _dualParaboloidCamera->setProjection(1.0f, 180.0f, zPlanes);
    // Set the cube camera as the currently active one
    kernel.getCameraMgr().pushActiveCamera(_dualParaboloidCamera);
    // Set the desired render stage, remembering the previous one
    RenderStage prevRenderStage = setRenderStage(renderStage);

    RenderPassManager& passMgr = RenderPassManager::instance();
    RenderPassManager::PassParams params;
    params.doPrePass = renderStage != RenderStage::SHADOW;
    params.occlusionCull = params.doPrePass;
    params.camera = _dualParaboloidCamera;
    params.stage = renderStage;

    // Enable our render target
    targetBuffer.begin(RenderTarget::defaultPolicy());
        for (U8 i = 0; i < 2; ++i) {
            targetBuffer.drawToLayer(hasColour ? RTAttachment::Type::Colour
                                               : RTAttachment::Type::Depth,
                                     0,
                                     i + arrayOffset);
            // Point our camera to the correct face
            _dualParaboloidCamera->lookAt(pos, (i == 0 ? WORLD_Z_NEG_AXIS : WORLD_Z_AXIS) + pos, WORLD_Y_AXIS);
            // And generated required matrices
            // Pass our render function to the renderer
            params.pass = passIndex + i;
            passMgr.doCustomPass(params);
        }
    targetBuffer.end();
    // Return to our previous rendering stage
    setRenderStage(prevRenderStage);
    // Restore our previous camera
    kernel.getCameraMgr().popActiveCamera();
}

void GFXDevice::increaseResolution() {
    const WindowManager& winManager = Application::instance().windowManager();
    const vec2<U16>& resolution = winManager.getActiveWindow().getDimensions();
    const vectorImpl<GPUState::GPUVideoMode>& displayModes = _state.getDisplayModes(winManager.targetDisplay());

    vectorImpl<GPUState::GPUVideoMode>::const_reverse_iterator it;
    for (it = std::rbegin(displayModes); it != std::rend(displayModes); ++it) {
        const vec2<U16>& tempResolution = it->_resolution;

        if (resolution.width < tempResolution.width &&
            resolution.height < tempResolution.height) {
            WindowManager& winMgr = Application::instance().windowManager();
            winMgr.handleWindowEvent(WindowEvent::RESOLUTION_CHANGED,
                                     winMgr.getActiveWindow().getGUID(),
                                     to_int(tempResolution.width),
                                     to_int(tempResolution.height));
            return;
        }
    }
}

void GFXDevice::decreaseResolution() {
    const WindowManager& winManager = Application::instance().windowManager();
    const vec2<U16>& resolution = winManager.getActiveWindow().getDimensions();
    const vectorImpl<GPUState::GPUVideoMode>& displayModes = _state.getDisplayModes(winManager.targetDisplay());
    
    vectorImpl<GPUState::GPUVideoMode>::const_iterator it;
    for (it = std::begin(displayModes); it != std::end(displayModes); ++it) {
        const vec2<U16>& tempResolution = it->_resolution;
        if (resolution.width > tempResolution.width &&
            resolution.height > tempResolution.height) {
            WindowManager& winMgr = Application::instance().windowManager();
            winMgr.handleWindowEvent(WindowEvent::RESOLUTION_CHANGED,
                                     winMgr.getActiveWindow().getGUID(),
                                     to_int(tempResolution.width),
                                     to_int(tempResolution.height));
            return;
        }
    }
}

void GFXDevice::toggleFullScreen() {
    WindowManager& winManager = Application::instance().windowManager();
    switch (winManager.getActiveWindow().type()) {
        case WindowType::WINDOW:
        case WindowType::SPLASH:
            winManager.getActiveWindow().type(WindowType::FULLSCREEN_WINDOWED);
            break;
        case WindowType::FULLSCREEN_WINDOWED:
            winManager.getActiveWindow().type(WindowType::FULLSCREEN);
            break;
        case WindowType::FULLSCREEN:
            winManager.getActiveWindow().type(WindowType::WINDOW);
            break;
    };
}

/// The main entry point for any resolution change request
void GFXDevice::onChangeResolution(U16 w, U16 h) {
    RenderTarget& screenRT = renderTarget(RenderTargetID(RenderTargetUsage::SCREEN));
    // Update resolution only if it's different from the current one.
    // Avoid resolution change on minimize so we don't thrash render targets
    if ((w == screenRT.getWidth() &&
         h == screenRT.getHeight()) ||
        !(w > 1 && h > 1)) {
        return;
    }

    // Update render targets with the new resolution
    for (RenderTarget* rt : _rtPool.renderTargets(RenderTargetUsage::SCREEN)) {
        if (rt) {
            rt->create(w, h);
        }
    }

    _previousDepthBuffer._rt->create(w, h);

    // Update post-processing render targets and buffers
    PostFX::instance().updateResolution(w, h);
    _gpuBlock._data._invScreenDimension.xy(1.0f / w, 1.0f / h);
    _gpuBlock._updated = true;
    // Update the 2D camera so it matches our new rendering viewport
    _2DCamera->setProjection(vec4<F32>(0, to_float(w), 0, to_float(h)), vec2<F32>(-1, 1));
}

/// Return a GFXDevice specific matrix or a derivative of it
void GFXDevice::getMatrix(const MATRIX& mode, mat4<F32>& mat) const {
    mat.set(getMatrixInternal(mode));
}

mat4<F32>& GFXDevice::getMatrixInternal(const MATRIX& mode) {
    // The matrix names are self-explanatory
    switch (mode) {
        case  MATRIX::VIEW_PROJECTION:
            return _gpuBlock._data._ViewProjectionMatrix;
        case MATRIX::VIEW:
            return _gpuBlock._data._ViewMatrix;
        case MATRIX::PROJECTION:
            return _gpuBlock._data._ProjectionMatrix;
        case MATRIX::VIEW_INV: 
            return _gpuBlock.viewMatrixInv();
        case MATRIX::PROJECTION_INV:
            return _gpuBlock._data._InvProjectionMatrix;
        case MATRIX::VIEW_PROJECTION_INV:
            return _gpuBlock.viewProjectionMatrixInv();
        case MATRIX::TEXTURE:
            Console::errorfn(Locale::get(_ID("ERROR_TEXTURE_MATRIX_ACCESS")));
            break;
        default:
            DIVIDE_ASSERT(false, "GFXDevice error: attempted to query an invalid matrix target!");
            break;
    };
    
    return MAT4_IDENTITY;
}

const mat4<F32>& GFXDevice::getMatrixInternal(const MATRIX& mode) const {
    // The matrix names are self-explanatory
    switch (mode) {
        case  MATRIX::VIEW_PROJECTION:
            return _gpuBlock._data._ViewProjectionMatrix;
        case MATRIX::VIEW:
            return _gpuBlock._data._ViewMatrix;
        case MATRIX::PROJECTION:
            return _gpuBlock._data._ProjectionMatrix;
        case MATRIX::VIEW_INV:
            return _gpuBlock.viewMatrixInv();
        case MATRIX::PROJECTION_INV:
            return _gpuBlock._data._InvProjectionMatrix;
        case MATRIX::VIEW_PROJECTION_INV:
            return _gpuBlock.viewProjectionMatrixInv();
        case MATRIX::TEXTURE:
            Console::errorfn(Locale::get(_ID("ERROR_TEXTURE_MATRIX_ACCESS")));
            break;
        default:
            DIVIDE_ASSERT(false, "GFXDevice error: attempted to query an invalid matrix target!");
            break;
    };

    return MAT4_IDENTITY;
}

/// Update the internal GPU data buffer with the clip plane values
void GFXDevice::updateClipPlanes() {
    GPUBlock::GPUData& data = _gpuBlock._data;
    for (U8 i = 0; i < Config::MAX_CLIP_PLANES; ++i) {
        data._clipPlanes[i] = _clippingPlanes[i].getEquation();
    }
    _gpuBlock._updated = true;
}

/// Update the internal GPU data buffer with the updated viewport dimensions
void GFXDevice::updateViewportInternal(const vec4<I32>& viewport) {
    // Change the viewport on the Rendering API level
    _api->changeViewport(viewport);
    // Update the buffer with the new value
    _gpuBlock._data._ViewPort.set(viewport.x, viewport.y, viewport.z, viewport.w);
    _gpuBlock._updated = true;
}

void GFXDevice::updateViewportInternal(I32 x, I32 y, I32 width, I32 height) {
    updateViewportInternal(vec4<I32>(x, y, width, height));
}

void GFXDevice::renderFromCamera(Camera& camera) {
    camera.updateLookAt();
    // Tell the Rendering API to draw from our desired PoV
    GPUBlock::GPUData& data = _gpuBlock._data;

    const mat4<F32>& viewMatrix = camera.getViewMatrix();
    bool viewMatUpdated = viewMatrix != _gpuBlock._data._ViewMatrix;
    if (viewMatUpdated) {
        data._ViewMatrix.set(viewMatrix);
    }
    F32 FoV = camera.getVerticalFoV();
    data._ProjectionMatrix.set(camera.getProjectionMatrix());
    data._cameraPosition.set(camera.getEye(), camera.getAspectRatio());
    data._renderProperties.zw(FoV, std::tan(FoV * 0.5f));
    data._ZPlanesCombined.xy(camera.getZPlanes());
    _gpuBlock.update(viewMatUpdated);
}

/// Enable or disable 2D rendering mode 
/// (orthographic projection, no depth reads)
void GFXDevice::toggle2D(bool state) {
    // Remember the previous state hash
    static size_t previousStateBlockHash = 0;
    // Prevent double 2D toggle to the same state (e.g. in a loop)
    if (state == _2DRendering) {
        return;
    }
    Kernel& kernel = Application::instance().kernel();
    _2DRendering = state;
    // If we need to enable 2D rendering
    if (state) {
        // Activate the 2D render state block
        previousStateBlockHash = _api->setStateBlock(_state2DRenderingHash);
        // Push the 2D camera
        kernel.getCameraMgr().pushActiveCamera(_2DCamera);
        // Upload 2D camera matrices to the GPU
        renderFromCamera(*_2DCamera);
    } else {
        // Reverting to 3D implies popping the 2D camera
        kernel.getCameraMgr().popActiveCamera();
        // And restoring the previous state block
        _api->setStateBlock(previousStateBlockHash);
    }
}

/// Update the rendering viewport
void GFXDevice::setViewport(const vec4<I32>& viewport) {
    // Avoid redundant changes
    _viewportUpdate = !viewport.compare(_viewport.top());

    if (_viewportUpdate) {
        // Push the new viewport
        _viewport.push(viewport);
        // Activate the new viewport
        updateViewportInternal(viewport);
    }
}

/// Restore the viewport to it's previous value
void GFXDevice::restoreViewport() {
    // If we didn't push a new viewport, there's nothing to pop
    if (!_viewportUpdate) {
        return;
    }
    // Restore the viewport
    _viewport.pop();
    // Activate the new top viewport
    updateViewportInternal(_viewport.top());
    _viewportUpdate = false;
}

/// Set a new viewport clearing the previous stack first
void GFXDevice::setBaseViewport(const vec4<I32>& viewport) {
    while (!_viewport.empty()) {
        _viewport.pop();
    }
    _viewport.push(viewport);

    // Set the new viewport
    updateViewportInternal(viewport);
    // The forced viewport can't be popped
    _viewportUpdate = false;
}

void GFXDevice::onCameraUpdate(Camera& camera) {
    if (drawDebugAxis()) {
        // We need to transform the gizmo so that it always remains axis aligned
        // Create a world matrix using a look at function with the eye position
        // backed up from the camera's view direction
        _axisGizmo->worldMatrix(
            mat4<F32>(-camera.getForwardDir() * 2, VECTOR3_ZERO, camera.getUpDir()) *
            _gpuBlock._data._ViewMatrix.getInverse());
        _axisGizmo->paused(false);
    } else {
        _axisGizmo->paused(true);
    }
}

void GFXDevice::onCameraChange(Camera& camera) {
    GPUBlock::GPUData& data = _gpuBlock._data;

    const mat4<F32>& projectionMatrix = camera.getProjectionMatrix();
    const vec2<F32>& planes = camera.getZPlanes();
    F32 aspectRatio = camera.getAspectRatio();
    F32 FoV = camera.getVerticalFoV();

    data._ProjectionMatrix.set(projectionMatrix);
    data._ZPlanesCombined.xy(planes);
    data._cameraPosition.w = aspectRatio;
    data._renderProperties.z = FoV;
    data._renderProperties.w = std::tan(FoV * 0.5f);
    _gpuBlock.update(false);
}

/// Depending on the context, either immediately call the function, or pass it
/// to the loading thread via a queue
bool GFXDevice::loadInContext(const CurrentContext& context, const DELEGATE_CBK_PARAM<bool>& callback) {
    // Skip invalid callbacks
    if (callback) {
        if (context == CurrentContext::GFX_LOADING_CTX && Config::USE_GPU_THREADED_LOADING) {
            CreateTask(callback)._task->startTask(Task::TaskPriority::HIGH, to_const_uint(Task::TaskFlags::SYNC_WITH_GPU));
        } else {
            if (Application::isMainThread()) {
                callback(false);
            } else {
                WriteLock w_lock(_GFXLoadQueueLock);
                _GFXLoadQueue.push_back(callback);
            }
        }

        // The callback is valid and has been processed
        return true;
    }
    
    return false;
}

/// Transform our depth buffer to a HierarchicalZ buffer (for occlusion queries and screen space reflections)
/// Based on RasterGrid implementation: http://rastergrid.com/blog/2010/10/hierarchical-z-map-based-occlusion-culling/
/// Modified with nVidia sample code: https://github.com/nvpro-samples/gl_occlusion_culling
void GFXDevice::constructHIZ(RenderTarget& depthBuffer) {
    static bool firstRun = true;
    static RTDrawDescriptor depthOnlyTarget;

    if (firstRun) {
        // We use a special shader that downsamples the buffer
        // We will use a state block that disables colour writes as we will render only a depth image,
        // disables depth testing but allows depth writes
        // Set the depth buffer as the currently active render target
        depthOnlyTarget.disableState(RTDrawDescriptor::State::CLEAR_COLOUR_BUFFERS);
        depthOnlyTarget.disableState(RTDrawDescriptor::State::CLEAR_DEPTH_BUFFER);
        depthOnlyTarget.disableState(RTDrawDescriptor::State::CHANGE_VIEWPORT);
        depthOnlyTarget.drawMask().disableAll();
        depthOnlyTarget.drawMask().setEnabled(RTAttachment::Type::Depth, 0, true);

        firstRun = false;
    }

    // The depth buffer's resolution should be equal to the screen's resolution
    RenderTarget& screenTarget = depthBuffer;
    U16 width = screenTarget.getWidth();
    U16 height = screenTarget.getHeight();
    U16 level = 0;
    U16 dim = width > height ? width : height;
    U16 twidth = width;
    U16 theight = height;
    bool wasEven = false;

    // Store the current width and height of each mip
    vec4<I32> previousViewport(_viewport.top());

    GenericDrawCommand triangleCmd;
    triangleCmd.primitiveType(PrimitiveType::TRIANGLES);
    triangleCmd.drawCount(1);
    triangleCmd.stateHash(_stateDepthOnlyRenderingHash);
    triangleCmd.shaderProgram(_HIZConstructProgram);

    // Bind the depth texture to the first texture unit
    screenTarget.bind(to_const_ubyte(ShaderProgram::TextureUsage::DEPTH), RTAttachment::Type::Depth, 0);
    screenTarget.begin(depthOnlyTarget);
    // We skip the first level as that's our full resolution image
    while (dim) {
        if (level) {
            twidth = twidth < 1 ? 1 : twidth;
            theight = theight < 1 ? 1 : theight;
            // Update the viewport with the new resolution
            updateViewportInternal(0, 0, twidth, theight);
            // Bind next mip level for rendering but first restrict fetches only to previous level
            screenTarget.setMipLevel(level, RTAttachment::Type::Depth, 0);
            _HIZConstructProgram->Uniform("depthLoD", level-1);
            _HIZConstructProgram->Uniform("isDepthEven", wasEven);
            // Dummy draw command as the full screen quad is generated completely in the vertex shader
            draw(triangleCmd);
        }

        // Calculate next viewport size
        wasEven = (twidth % 2 == 0) && (theight % 2 == 0);
        dim /= 2;
        twidth /= 2;
        theight /= 2;
        level++;
    }

    updateViewportInternal(previousViewport);
    // Reset mipmap level range for the depth buffer
    screenTarget.resetMipLevel(RTAttachment::Type::Depth, 0);
    // Unbind the render target
    screenTarget.end();
}

/// Extract the pixel data from the main render target's first colour attachment
/// and save it as a TGA image
void GFXDevice::Screenshot(const stringImpl& filename) {
    // Get the screen's resolution
    RenderTarget& screenRT = renderTarget(RenderTargetID(RenderTargetUsage::SCREEN));
    U16 width = screenRT.getWidth();
    U16 height = screenRT.getHeight();
    // Allocate sufficiently large buffers to hold the pixel data
    U32 bufferSize = width * height * 4;
    U8* imageData = MemoryManager_NEW U8[bufferSize];
    // Read the pixels from the main render target (RGBA16F)
    screenRT.readData(GFXImageFormat::RGBA, GFXDataFormat::UNSIGNED_BYTE, imageData);
    // Save to file
    ImageTools::SaveSeries(filename,
                           vec2<U16>(width, height),
                           32,
                           imageData);
    // Delete local buffers
    MemoryManager::DELETE_ARRAY(imageData);
}

void GFXDevice::computeFrustumPlanes(const mat4<F32>& invViewProj, vec4<F32>* planesOut) {
    static const vec4<F32> unitVecs[] = { vec4<F32>(-1, -1, -1, 1),
                                          vec4<F32>(-1 , 1, -1, 1),
                                          vec4<F32>(-1, -1,  1, 1),
                                          vec4<F32>( 1, -1, -1, 1),
                                          vec4<F32>( 1,  1, -1, 1),
                                          vec4<F32>( 1, -1,  1, 1),
                                          vec4<F32>( 1,  1,  1, 1) };

    // Get world-space coordinates for clip-space bounds.
    vec4<F32> lbn(invViewProj * unitVecs[0]);
    vec4<F32> ltn(invViewProj * unitVecs[1]);
    vec4<F32> lbf(invViewProj * unitVecs[2]);
    vec4<F32> rbn(invViewProj * unitVecs[3]);
    vec4<F32> rtn(invViewProj * unitVecs[4]);
    vec4<F32> rbf(invViewProj * unitVecs[5]);
    vec4<F32> rtf(invViewProj * unitVecs[6]);

    vec3<F32> lbn_pos(lbn.xyz() / lbn.w);
    vec3<F32> ltn_pos(ltn.xyz() / ltn.w);
    vec3<F32> lbf_pos(lbf.xyz() / lbf.w);
    vec3<F32> rbn_pos(rbn.xyz() / rbn.w);
    vec3<F32> rtn_pos(rtn.xyz() / rtn.w);
    vec3<F32> rbf_pos(rbf.xyz() / rbf.w);
    vec3<F32> rtf_pos(rtf.xyz() / rtf.w);

    // Get plane equations for all sides of frustum.
    vec3<F32> left_normal(Cross(lbf_pos - lbn_pos, ltn_pos - lbn_pos));
    left_normal.normalize();
    vec3<F32> right_normal(Cross(rtn_pos - rbn_pos, rbf_pos - rbn_pos));
    right_normal.normalize();
    vec3<F32> top_normal(Cross(ltn_pos - rtn_pos, rtf_pos - rtn_pos));
    top_normal.normalize();
    vec3<F32> bottom_normal(Cross(rbf_pos - rbn_pos, lbn_pos - rbn_pos));
    bottom_normal.normalize();
    vec3<F32> near_normal(Cross(ltn_pos - lbn_pos, rbn_pos - lbn_pos));
    near_normal.normalize();
    vec3<F32> far_normal(Cross(rtf_pos - rbf_pos, lbf_pos - rbf_pos));
    far_normal.normalize();

    planesOut[0].set(left_normal,   -Dot(left_normal, lbn_pos));   // Left
    planesOut[1].set(right_normal,  -Dot(right_normal, rbn_pos));  // Right
    planesOut[2].set(near_normal,   -Dot(near_normal, lbn_pos));   // Near
    planesOut[3].set(far_normal,    -Dot(far_normal, lbf_pos));    // Far
    planesOut[4].set(top_normal,    -Dot(top_normal, ltn_pos));    // Top
    planesOut[5].set(bottom_normal, -Dot(bottom_normal, lbn_pos)); // Bottom
}
};

Commits for Divide-Framework/trunk/Source Code/Platform/Video/GFXDevice.cpp

Diff revisions: vs.
Revision Author Commited Message
781 Diff Diff IonutCava picture IonutCava Wed 12 Oct, 2016 16:03:46 +0000

[IonutCava]
- ShaderComputeQueue <-> Material shader request bug fix (element stuck in QUEUED state)
- More Camera class cleanups
- Add a method of removing EnvironmentProbes from their Pool

780 Diff Diff IonutCava picture IonutCava Tue 11 Oct, 2016 16:15:24 +0000

[IonutCava]
- AtomicCounter in glUniformBuffer is now based on a glGenericBuffer implementation
- GFXDevice <-> Camera relation cleaned up a bit
- Added a readback system for glGenericBuffer
- Fixed a bug with Z_PRE_PASS stage culling

779 Diff Diff IonutCava picture IonutCava Mon 10 Oct, 2016 15:31:45 +0000

[IonutCava]
- Frustum calculation bug fix

778 Diff Diff IonutCava picture IonutCava Mon 10 Oct, 2016 15:18:32 +0000

[IonutCava]
- Rendering abstraction cleanup

777 Diff Diff IonutCava picture IonutCava Fri 07 Oct, 2016 16:14:48 +0000

[IonutCava]
- Continue to implement CommandBuffer / RenderPass / RenderSubPass system.
— Rendering artefacts are still present.

776 Diff Diff IonutCava picture IonutCava Thu 06 Oct, 2016 15:57:56 +0000

[IonutCava]
- Improve the CommandBuffer system to include the notion of a RenderPass /RenderSubPass with output render target info stored in them (not used yet)

772 Diff Diff IonutCava picture IonutCava Thu 29 Sep, 2016 15:50:57 +0000

[IonutCava]
- Better GPU Vendor / Renderer detection

771 Diff Diff IonutCava picture IonutCava Tue 27 Sep, 2016 15:39:49 +0000

[IonutCava]
- Modify GFXDevice::flushRenderQueues so that it creates command buffers instead of directly calling rendering calls:
— Command buffers contain buffer, shader, texture and geometry info
— Command buffer state is now parsed and set at the API level instead of GFXDevice level
— This will allow easier porting to low level APIs later on
- Indirect draw is now a GenericDrawCommand render option instead of an explicit buffer draw parameter

770 Diff Diff IonutCava picture IonutCava Tue 20 Sep, 2016 16:12:27 +0000

[IonutCava]
- Prepare velocity map calculation (later used for temporal AA and motion blur):
— Add a previous depth buffer render target that blits the main depth buffer at the end of the frame
— Add a new render target to the screen FBO: velocity map
- Store normals as RG instead of RGB and pack/unpack in shaders. (compute >>>> bandwith)
- Rename Forward+ renderer to something more descriptive

(Requires deletion of all parsed & cached data)

769 IonutCava picture IonutCava Fri 16 Sep, 2016 15:18:09 +0000

[IonutCava]
- Clean up camera class:
— Remove “reflection” specific code and add a generic “reflect” method to allow the calling code to handle camera placements and reflections. This got rid of the “reflectionPlane” member in the RenderParams struct
— Remove some member variables and replace them with on-demand computed variables on the stack.
— Add a “destroyCamera” call to the CameraManager to allow better resource cleanup when needed
- Prepare a rendering system overhaul:
— Add dual screen render targets: current frame and previous frame. Used for future features such as TXAA and Motion Blur.
— Prepare code for an implementation similar to the Doom (2016) system: http://www.adriancourreges.com/blog/2016/09/09/doom-2016-graphics-study/ (the code was already going in that direction)