Re: [osg-users] Copying rendered image before and after post-processing
Hi Sebastian, Thanks for your answer. Eventually I went with a another solution (copy a scaled-down texture), as copying with alternating PBOs always resulted in the texture after post-processing. Unfortunately I had no more time for this. It could be related to old hardware though (GeForce 8800). Thanks, Shahar Sebastian Messerschmidt wrote: Hi Shahar, I guess your scene is being rendered to an FBO already. So simply create a camera taking the resulting FBO as input, a new texture as output an attach a simple glsl program to copy. Also I guess osgPPU has some copy processor, but I'm not sure. cheers Sebastian Hi, I've got an OSG application with a single viewer and camera. The camera is set to render to FBO - Code: camera-attach(osg::Camera::COLOR_BUFFER0, m_texture2D); camera-setRenderTargetImplementation(osg::Camera::FRAME_BUFFER_OBJECT); Rendering consists of the following main steps: 1) Render scene 2) Post-processing using osgPPU 3) In final draw callback, copy the rendered image to the application using two alternating PBOs Now I'd like to add one more step - 1.5) Copy the rendered image to application before post-processing The copied image will only be used in the application (e.g. to save it), and will not be displayed nor copied back to the graphics card (like the osgprerender example). In addition, I don't mind getting it with a delay of a few frames. How would you recommend implementing this step efficiently, while maintaining the current performance level of the application (runs at 60Hz). I guess I could use two separate cameras to render the scene (one without osgPPU), but I wonder if there is a better way. Thanks, Shahar -- Read this topic online here: http://forum.openscenegraph.org/viewtopic.php?p=55940#55940 ___ osg-users mailing list http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org ___ osg-users mailing list http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org -- Post generated by Mail2Forum -- Read this topic online here: http://forum.openscenegraph.org/viewtopic.php?p=56024#56024 ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
[osg-users] Copying rendered image before and after post-processing
Hi, I've got an OSG application with a single viewer and camera. The camera is set to render to FBO - Code: camera-attach(osg::Camera::COLOR_BUFFER0, m_texture2D); camera-setRenderTargetImplementation(osg::Camera::FRAME_BUFFER_OBJECT); Rendering consists of the following main steps: 1) Render scene 2) Post-processing using osgPPU 3) In final draw callback, copy the rendered image to the application using two alternating PBOs Now I'd like to add one more step - 1.5) Copy the rendered image to application before post-processing The copied image will only be used in the application (e.g. to save it), and will not be displayed nor copied back to the graphics card (like the osgprerender example). In addition, I don't mind getting it with a delay of a few frames. How would you recommend implementing this step efficiently, while maintaining the current performance level of the application (runs at 60Hz). I guess I could use two separate cameras to render the scene (one without osgPPU), but I wonder if there is a better way. Thanks, Shahar -- Read this topic online here: http://forum.openscenegraph.org/viewtopic.php?p=55940#55940 ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
Re: [osg-users] Passing location-based data to a fragment shader
Hi Sebastian,
Thanks for the detailed explanation and the hints regarding texture/world
coordinates. I've tried binding at the tile level as you suggested, but
apparently the additional data source is not fast enough to be used at
real-time.
At this point, I'm going to have the original terrain model regenerated with
two textures for each tile, instead of constructing the second texture on the
fly. As far as I understand, binding a second texture should be possible with
the OSG and TXP formats.
So thanks again for attempting to help with issue.
Best,
Shahar
Sebastian Messerschmidt wrote:
You don't really need to bind at a level this deep. Usually if the
terrain tile is loaded and represent a LOD you want to work with, you
can simple add your data texture(s) as uniform to the stateset of the tile.
Given you have a texture with the data-image called texture you can do
it this way in your callback:
osg::StateSet* state_set = node-getOrCreateStateSet();
osg::Uniform* sampler = new osg::Uniform(MyDataSamper, 2
/*texture unit*/);
state_set-setTextureAttribute(unit, texture ,osg::StateAttribute::ON);
state_set-addUniform(sampler);
With this you bind the texture to a sampler for your shader, at the
texture unit 2. Depending on how many textures are already bound you
might need to modify the unit.
In the fragment shader you can now access the texture via:
uniform sampler2D MyDataSampler
..
..
main()
{
vec2 tex_coord = gl_TexCoord[0].st;
vec4 data = texture2D(MyDataSampler, tex_coord);
}
--
Read this topic online here:
http://forum.openscenegraph.org/viewtopic.php?p=55008#55008
___
osg-users mailing list
osg-users@lists.openscenegraph.org
http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
Re: [osg-users] Passing location-based data to a fragment shader
Hi Nick, Yes, this is part of what I'm trying to do (I'd like to do the same with osg models). I'd be glad if you could elaborate on the different approach, even though I would probably go with pre-generating the texture along with the archive. Thanks, Shahar Trajce Nikolov NICK wrote: Hi Shahar, I read this post and found it interesting. Just out of curiosity : Each texel in the new texture, should be set based on the world location of the corresponding texel in the current geometry texture Are you trying to overlay a texture on top of txp archive and it should be georeferenced? If so, there might be a different approach Nick -- Read this topic online here: http://forum.openscenegraph.org/viewtopic.php?p=55009#55009 ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
Re: [osg-users] Passing location-based data to a fragment shader
Sebastian Messerschmidt wrote: Hi Shahar Kosti: As you don't provide any exact numbers I can only guess. But usually I'd bind one or more textures per tile of paged set. Depending on your hardware bind of 8192x8192 textures is no problem. Can you be a bit more specific on how big the set is? Also for the hidden textures - I don't really get what you mean. My best guess is, that you mean not displayed. Therefore write a custom fragment shader and bind the data-textures along with you diffuse or whatever textures to the stateset of the tile. As the fragment shader is fully programmable, it is your choice what is done with the data. cheers Sebastian Hi Sebastian, Sorry for my late reply, it took me some time to get the data. One TXP I'm currently looking at, represents a 3000 km² area with texture resolution of 5m/pixel. Some areas have better resolution, up to 10 cm/pixel. The additional metadata has similar resolution. So obviously wouldn't fit on a single texture. Regarding your suggestions, could you elaborate on the bind one or more textures per tile of paged set approach? My current approach (which doesn't work), is to intercept paged nodes using a custom ReadFileCallback, which runs a NodeVisitor. The visitor finds all Geode objects and then the underlying geometries. I can bind a new texture to these geometries, but I'm not sure how to set the values correctly. Each texel in the new texture, should be set based on the world location of the corresponding texel in the current geometry texture. I'm not sure how to do that using the geometry vertex and texture coordinate indices. I hope this is clear enough. Thanks for the help, Shahar -- Read this topic online here: http://forum.openscenegraph.org/viewtopic.php?p=54810#54810 ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
[osg-users] Passing location-based data to a fragment shader
Hi all, I'm trying to combine two data sources in a shader: * Terrain model - a large object in OSG or TXP formats. * Metadata grid - a data structure that provides additional information for 2D coordinates in the terrain (at a certain resolution). I'd like to use some of the additional metadata values, in a fragment shader that runs on the terrain object. In other words, modify each pixel based on an additional value from the metadata grid. I tried to pass the values as vertex attributes: compute the world location of each vertex, and set the vertex attribute from the metadata grid. That didn't work very well, as the additional data resolution is very high, almost the same as the terrain textures, and the vertices are too far apart from each other. How would you recommend to pass the required data from the application to the shader? Thanks, Shahar -- Read this topic online here: http://forum.openscenegraph.org/viewtopic.php?p=54673#54673 ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
Re: [osg-users] Passing location-based data to a fragment shader
Hi Sebastian, Thanks for your response. What if the data does not fit in a single 2D texture? in the worst case scenario, the terrain format would be TXP (i.e. paged) and the additional data resolution would be close to the textures resolution. Is there any way to have a set of hidden textures, and access them in the fragment shader? Shahar Sebastian Messerschmidt wrote: Hi Shahar, If the data is 2D and simply projected to x,y you could use a 2D texture and sample it per fragment based on generated texture coordinates -- Read this topic online here: http://forum.openscenegraph.org/viewtopic.php?p=54683#54683 ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
[osg-users] [osgPPU] Perfomance issues with PBO for input textures
Hi, I'm new to osgPPU and OSG in general, so please bear with me :) I'm trying to use the built-in mechanism in osgPPU::Unit for copying the rendered textures to the CPU. I'm using a simple tester based on one of the osgPPU, and my initialization code is structured like this: Code: // Somewhere in the middle of the Processor graph: osgPPU::Unit* pUnit = new osgPPU::UnitInOut(); pUnit-setUsePBOForInputTexture(0, true); parentUnit-addChild(pUnit); With this code the frame rate drops significantlly, from 60 to 10-30 (dependning on the viewport size). If I comment out the glGetTexImage call inside Unit.cpp, everything works as before (but nothing gets copied, of course). Considering I don't mind getting the texture with a delay of a few frames, is there any way to improve the performance of the current implementation? I must say I've read about PBO (http://www.songho.ca/opengl/gl_pbo.html), but I'm not sure whether this is the correct approach here, or how to integrate it with the existing code base. Thanks, Shahar -- Read this topic online here: http://forum.openscenegraph.org/viewtopic.php?p=45844#45844 ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
[osg-users] Integrating OSG with Qt: QOSGWidget memory consumption
Hi, I'm new to OSG and trying to create a Qt-based application with OSG windows inside. To do this, I'm following the osgviewerQT example using the multi-threaded CompositeViewer mode (--MTCompositeViewer). When I start the example with a model it seems to be loaded 6 times, which is also the number of views. This makes the memory consumption much larger than the other modes in that example. Why is the model loaded multiple times? Is it possible to load it once for all views? OSG version - 2.9.9, Qt version - 4.5.1 Command line parameters: osgviewerQT.exe -- QOSGWidget --MTCompositeViewer my_model.ive Thank you! -- Read this topic online here: http://forum.openscenegraph.org/viewtopic.php?p=33181#33181 ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
Re: [osg-users] Integrating OSG with Qt: QOSGWidget memory consumption
Eventually I went with a modified version of the --CompositeViewer part of the example. I'm still not sure what is wrong with the other approach - the number of GraphicsContext and View objects is the same. -- Read this topic online here: http://forum.openscenegraph.org/viewtopic.php?p=33445#33445 ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
