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
/*
   Copyright (c) 2014 DIVIDE-Studio
   Copyright (c) 2009 Ionut Cava

   This file is part of DIVIDE Framework.

   Permission is hereby granted, free of charge, to any person obtaining a copy of this software
   and associated documentation files (the "Software"), to deal in the Software without restriction,
   including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense,
   and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so,
   subject to the following conditions:

   The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

   THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED,
   INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
   IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
   WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE
   OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

 */

#ifndef _GL_VERTEX_ARRAY_H_
#define _GL_VERTEX_ARRAY_H_

#include "Hardware/Video/Buffers/VertexBuffer/Headers/VertexBuffer.h"
#include "Hardware/Video/OpenGL/Headers/glResources.h"
///Always bind a shader, even a dummy one when rendering geometry. No more fixed matrix API means no more VBs or VAs
///One VAO contains: one VB for data, one IB for indices and uploads to the shader vertex attribs for:
///- Vertex Data  bound to location Divide::VERTEX_POSITION_LOCATION
///- Colors       bound to location Divide::VERTEX_COLOR_LOCATION
///- Normals      bound to location Divide::VERTEX_NORMAL_LOCATION
///- TexCoords    bound to location Divide::VERTEX_TEXCOORD_LOCATION
///- Tangents     bound to location Divide::VERTEX_TANGENT_LOCATION
///- BiTangents   bound to location Divide::VERTEX_BITANGENT_LOCATION
///- Bone weights bound to location Divide::VERTEX_BONE_WEIGHT_LOCATION
///- Bone indices bound to location Divide::VERTEX_BONE_INDICE_LOCATION

class glVertexArray : public VertexBuffer {
public:
    glVertexArray();
    ~glVertexArray();

    bool Create(bool staticDraw = true);
    void Destroy();

    virtual bool SetActive();

    ///Never call Refresh() just queue it and the data will update before drawing
    inline bool queueRefresh() {_refreshQueued = true; return true;}

protected:
    friend class GFXDevice;
    void Draw(const vectorImpl<GenericDrawCommand>& commands, bool skipBind = false);

protected:
    /// If we have a shader, we create a VAO, if not, we use simple VB + IB. If that fails, use VA
    bool Refresh();
    /// Internally create the VB
    bool CreateInternal();
    /// Enable full VAO based VB (all pointers are tracked by VAO's)
    void Upload_VB_Attributes();
    /// Integrity checks
    void checkStatus();
    /// Internal draw command for a single command
    void Draw(const GenericDrawCommand& command, bool skipBind = false);

protected:
    GLenum _formatInternal;
    GLuint _IBid;
    GLuint _VBid;
    GLuint _VAOid;
    GLuint _indirectDrawBuffer;
    GLuint _usage;
    bool _animationData;     ///< Used to bind an extra set of vertex attributes for bone indices and bone weights
    bool _refreshQueued;     ///< A refresh call might be called before "Create()". This should help with that
    vectorImpl<vec4<GLhalf> >  _normalsSmall;
    vectorImpl<vec4<GLshort> > _tangentSmall;
    vectorImpl<vec4<GLshort> > _bitangentSmall;

    GLsizei _prevSize[VertexAttribute_PLACEHOLDER];
    GLsizei _prevSizeIndices;
};

#endif

Commits for Divide-Framework/trunk/Source Code/Hardware/Video/OpenGL/Buffers/VertexBuffer/Headers/glVertexArray.h

Diff revisions: vs.
Revision Author Commited Message
286 Diff Diff IonutCava picture IonutCava Fri 18 Jul, 2014 16:32:03 +0000

[Ionut] [[BR]]

    • very buggy commit ** [[BR]]

- New rendering pipeline: [[BR]]
— Bumped OpenGL version to 4.4 [[BR]]
— Command based rendering. GFXDevice dispatches rendering commands to their respective buffers [[BR]]
— State changing is exclusively controlled by the GFXDevice class [[BR]]
— Material properties and transform matrices are stored in shader storage buffers that are updated after culling and uploaded once per RenderPass. Objects use drawIDs to index data in shaders [[BR]]
— Removed all custom matrices and other, per shader, bind data [[BR]]
— Removed per node light data. Light culling and shader buffers will replace the old light management system [[BR]]
— Framebuffers now use the Texture class for attachments [[BR]]
— Sampler objects now work like RenderStateBlocks. GL_API handles them and stores them in a hash indexed map [[BR]]
— Removed “prepareMaterial” calls. Pre-draw material setup is now only limited to texture binding [[BR]]
— Moved immediate mode emulation and primitive rendering from API level to GFXDevice level [[BR]]
— Framebuffer rendering must be completed with a call to end() to ensure proper MSAA resolve (asserts, otherwise) [[BR]]
- Fixed hash code generation for RenderStateBlock (and now, glSamplerObject) using boost::hash_combine [[BR]]
- Temporarily disabled XML caching for materials [[BR]]

280 Diff Diff IonutCava picture IonutCava Sat 28 Jun, 2014 19:11:37 +0000

[Ionut] [[BR]]
- Improve ClipPlane management (bunch equations into the matrix shader buffer) [[BR]]
- Immediate mode emulation primitives set their own state based on specified hash instead of relying on state functions to be defined [[BR]]
- updateStates() function is now called right before the actual draw call to facilitate future batching work [[BR]]
- added tracked bool values (similar to those in ShaderInfo) to SGN class to fix selection bug. [[BR]]
— will probably create a “stateTracker” class to allow multiple classes to use tracked variables without all of the code duplication [[BR]]

279 Diff Diff IonutCava picture IonutCava Sat 28 Jun, 2014 16:20:46 +0000

[Ionut][[BR]]
- Unified draw command system into GenericDrawCommand [[BR]]
- Fixed double VB creation in Object3D [[BR]]
- Reworked triangle data computation (now Object3D’s responsibility) [[BR]]
- VertexArrays can now render using the primitive type specified in the draw command instead of the fixed type specified at creation [[BR]]
- Optimized terrain shader [[BR]]

276 Diff Diff IonutCava picture IonutCava Fri 13 Jun, 2014 18:23:32 +0000

[Ionut][[BR]]
- Terrain now inherits from Object3D (instead of directly from SceneNode) [[BR]]
- Restructured RenderInstance to support multiple draw commands [[BR]]
— Multiple draw commands per instance are separated to different LoD batches automatically on render [[BR]]

274 Diff Diff IonutCava picture IonutCava Fri 13 Jun, 2014 12:56:03 +0000

[Ionut] [[BR]]
- Simplified buffer based rendering (VertexBuffer & GenericVertexData are now exclusively command based) [[BR]]
- Update bone buffer to persistent mapped-SSBO system with double-buffering [[BR]]
- Simplified sky rendering and (finally) reduced the sun’s radius [[BR]]

270 Diff Diff IonutCava picture IonutCava Thu 05 Jun, 2014 19:08:07 +0000

[Ionut][[BR]]
- SubMeshes now use a single vertex buffer (owned by the parent mesh) with proper offsetting and querying [[BR]]
— Implementation not optimal yet [[BR]]
- Cleaned up VertexBuffer related classes [[BR]]

268 Diff Diff IonutCava picture IonutCava Wed 04 Jun, 2014 10:20:27 +0000

[IonutCava] [[BR]]
- Terrain and vegetation updates: [[BR]]
— Use standard phong lighting model [[BR]]
— Separate grass patches per chunk [[BR]]
— Reduce number of required texture bind points for terrain by adding both textures and normal maps in a single texture array [[BR]]
— Draw terrain with 2 drawcalls (using glMultiDrawElements twice, once for LoD == 0 and once for LoD > 0) [[BR]]
[[BR]]
- Buffer management updates: [[BR]]
— Unified buffer binding system [[BR]]
— Made Uniform buffer objects visible to all classes and moved lightUBO to LightManager, matricesUBO to GFXDevice etc [[BR]]
— Added initial support for Shader Buffer Objects [[BR]]
— Moved viewport management to GFXDevice class [[BR]]
[[BR]]
- Rendering improvements: [[BR]]
— Added Hi-Z occlusion culling to instanceCull shader (using Hi-Z depth buffer constructed in GFXDevice class) [[BR]]
— Resolved issues with shadow maps overriding texture bind locations [[[BR]]
— Improved shader and shader program loading [[BR]]
— Improved material transparency support [[BR]]
— Added buffer lock manager class from “Approaching Zero Driver Overhead in OpenGL (Presented by NVIDIA)” speech at GDC14 [[BR]]

265 Diff Diff k1ngp1n picture k1ngp1n Tue 27 May, 2014 19:47:56 +0000

[Ionut] [[BR]]
- reworked uniform buffer binding [[BR]]
- modified framebuffer mipmap update [[BR]]

264 Diff Diff k1ngp1n picture k1ngp1n Tue 27 May, 2014 13:04:38 +0000

[Ionut] [[BR]]
- New camera system: [[BR]]
— Stack based (push/pop active camera) [[BR]]
— Per camera frustum [[BR]]
— Per camera projection [[BR]]
— Shadow mapping now uses the new camera system (each light has a shadow camera) [[BR]]
[[BR]]
- New terrain, water and vegetation system: [[BR]]
— Direct import of terrain from L3DT both as 8-bit grayscale and 16-bit raw formats [[BR]]
— Multiple alpha maps for texture splatting support [[BR]]
— Per texture detail map support [[BR]]
— Transform feedback based visibility culling for vegetation [[BR]]
— Alpha-map based vegetation placing on terrain [[BR]]
— Water now uses a refraction map instead of alpha blending for better refraction effects [[BR]]
[[BR]]
- SceneGraph improvements: [[BR]]
— Better bounding box updates from child to parent [[BR]]
— Better onDraw / postDraw implementation [[BR]]
[[BR]]
- Performance improvements: [[BR]]
— Batch 2D rendering tasks together as much as possible[[BR]]
— Improve renderStateBlock hash generation for better sorting [[BR]]
— Move most state related gfx changes from API implementation to the GFXDevice class [[BR]]
— Move viewport stack to the GFXDevice class [[BR]]
— Added various performance counters (frame duration with round-robin query system, draw call count, etc) [[BR]]
— Fixed and improved clip plane management [[BR]]
— Delete client-side texture data after GPU upload [[BR]]
[[BR]]
- New features: [[BR]]
— Added callback support to the Task class so that a specific task may call a function in the main thread upon completion [[BR]]
— Added mouse-wheel zoom support to Orbit/ThirdPerson cameras [[BR]]
— Added 2D Texture Array support (e.g. used for terrain textures) [[BR]]

260 k1ngp1n picture k1ngp1n Sat 03 May, 2014 19:33:41 +0000

[Ionut] [[BR]]
- More work on CSM. Reverted back to VSM to allow easier blurring and filtering [[BR]]
- Rewrote FrameBuffer classes to fix MSAA and memory leaks that occurred on resize (double creation of textures) [[BR]]
— Removed FBType. FrameBuffer types depend only on attachment types and if it’s MSAA enabled or not [[BR]]
- Rewrote RenderState classes to be API agnostic [[BR]]
- Bug fixes in math classes [[BR]]
- Changed “resolution change” logic [[BR]]
- Simplified some options in config.xml [[BR]]
- Improved per-light shadow options [[BR]]