mirror of
https://github.com/OpenSpace/OpenSpace.git
synced 2026-01-11 05:59:57 -06:00
* Simplest possible PBO implementation. * Add PBO class * TileLoadJob owns raw tile data * Working on a soluton to cache textures and reuse them * PBO and cached textures working for one texture type. Color textures. * Threadpool for tile requests uses LRU cache as queue * Remove framesUntilRequestFlush * Clean up * Clean up * Use prioritizing concurrent job manager * Use TileTextureInitData to initialize RawTileDataReader. * Class TextureContainer owns the textures to use for tiles. * Using TileTextureInitData to determine if new caches need to be created. * Remove WriteDataDescription * Remove TileDataLayout * Rendering many different layer types again * TileProviderByLevel gives layergroup id to tile providers * Comment away use of PBO * Erase unfinished requests to make room for new ones * Enable choice of PBO or not. * Enable resetting of asynctiledataprovider * Add the ability to use PBO and also load to CPU * Update ghoul * Solve culling issue. * Texture pointer of Tile is now a raw pointer. Currently break single image tile provider and text tile provider. * Add gpudata * Move fetching of shader preprocessing data to LayerManager * No comparisons to determine shader recompilation. * Show the tile cache size in the GUI * Clean up and comment. * Solve bug where float is interpreted as NaN * Enable ability to blend between layers again * Fix single image provider * Fix windows build error * Fix OSX compile issue. * Some clean up * Showing correct texture data size * Enable use of text tile providers again. No backgroupd image path however. * Change cache size from GUI * Clean up * Solve osx compilation error. * Update ghoul * Make it possible to switch between PBO and not during runtime. * Enable resetting of tile datasets * change function module in moduleengine to identify module by name * MemoryAwareTileCache is no longer a singleton * Update ownership of properties for globe browsing * Logging info about resetting tile reader. * Logging info * Fix requested changes * Fix some compile warnings. * Fix compilation warnings * Add ability to blend values with blend parameter. Also define settings through lua dict. * Fix some comments on pull request. * Change formatting * Change formatting * Change formatting * Fix pull request comments. * Those are details * Make Mercury great again. * Make Earth great again. * Solve conflict * Test to sometimes use valueblending and sometimes not * Not always use value blending * Update ghoul * Change from auto to explicit type. * Update test for LRU Cache * Include algorithm.
107 lines
5.1 KiB
C++
107 lines
5.1 KiB
C++
/*****************************************************************************************
|
|
* *
|
|
* OpenSpace *
|
|
* *
|
|
* Copyright (c) 2014-2017 *
|
|
* *
|
|
* 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 __OPENSPACE_MODULE_GLOBEBROWSING___CHUNK_RENDERER___H__
|
|
#define __OPENSPACE_MODULE_GLOBEBROWSING___CHUNK_RENDERER___H__
|
|
|
|
#include <memory>
|
|
|
|
namespace ghoul { namespace opengl {
|
|
class ProgramObject;
|
|
} }
|
|
|
|
namespace openspace {
|
|
|
|
struct RenderData;
|
|
|
|
namespace globebrowsing {
|
|
|
|
class Chunk;
|
|
class Grid;
|
|
class GPULayerManager;
|
|
class LayerManager;
|
|
class LayerShaderManager;
|
|
class RenderableGlobe;
|
|
|
|
class ChunkRenderer {
|
|
public:
|
|
ChunkRenderer(std::shared_ptr<Grid> grid,
|
|
std::shared_ptr<LayerManager> layerManager);
|
|
|
|
/**
|
|
* Chooses to render a chunk either locally or globally depending on the chunklevel
|
|
* of the <code>Chunk</code>.
|
|
*/
|
|
void renderChunk(const Chunk& chunk, const RenderData& data);
|
|
void update();
|
|
|
|
void recompileShaders(const RenderableGlobe& globe);
|
|
|
|
private:
|
|
/**
|
|
* Chunks can be rendered either globally or locally. Global rendering is performed
|
|
* in the model space of the globe. With global rendering, the vertex positions
|
|
* of a chunk are calculated in the vertex shader by transforming the geodetic
|
|
* coordinates of the chunk to model space coordinates. We can only achieve floating
|
|
* point precision by doing this which means that the camera too close to a global
|
|
* tile will lead to jagging. We only render global chunks for lower chunk levels.
|
|
*/
|
|
void renderChunkGlobally(const Chunk& chunk, const RenderData& data);
|
|
|
|
/**
|
|
* Local rendering of chunks are done using linear interpolation in camera space.
|
|
* All four corner points of the chunk are calculated in double precision on the
|
|
* CPU and transformed to camera space with double precision matrix transforms.
|
|
* These positions can then be cast to floats and uploaded to the vertex shader.
|
|
* The vertex shader rendering performs linear interpolation between the four
|
|
* corner points to get the resulting chunk. This means that there will be an error
|
|
* due to the curvature of the globe. The smaller the patch is (with higher chunk
|
|
* levels) the better the approximation becomes. This is why we only render local
|
|
* chunks for higher chunk levels.
|
|
*/
|
|
void renderChunkLocally(const Chunk& chunk, const RenderData& data);
|
|
|
|
ghoul::opengl::ProgramObject* getActivatedProgramWithTileData(
|
|
std::shared_ptr<LayerShaderManager> layeredShaderManager,
|
|
std::shared_ptr<GPULayerManager> gpuLayerManager,
|
|
const Chunk& chunk);
|
|
|
|
// shared pointer to a grid which can be the same for all rendered chunks.
|
|
std::shared_ptr<Grid> _grid;
|
|
std::shared_ptr<LayerManager> _layerManager;
|
|
|
|
// Two different shader programs. One for global and one for local rendering.
|
|
std::shared_ptr<LayerShaderManager> _globalLayerShaderManager;
|
|
std::shared_ptr<LayerShaderManager> _localLayerShaderManager;
|
|
|
|
// Layered texture uniforms are chached in the uniform ID handles.
|
|
std::shared_ptr<GPULayerManager> _globalGpuLayerManager;
|
|
std::shared_ptr<GPULayerManager> _localGpuLayerManager;
|
|
};
|
|
|
|
} // namespace globebrowsing
|
|
} // namespace openspace
|
|
|
|
#endif // __OPENSPACE_MODULE_GLOBEBROWSING___CHUNK_RENDERER___H__
|