Additionally, shadows on objects, that do not cast shadows themselves, can be enabled from app_settings\ogre_scene.xml, by setting objectshadows="true". By default they're off because they suffer from backprojecting issues (would need proper depth-shadowmaps).
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
By looking at the configuration, it seems that we're using a single 512^2 shadow map with LiSPSM projection for the directional sky? To get the proper shadows, can it be done that the avatar geometry is rendered in the shadow generation pass but not in the color pass when in FPS mode? This would give the proper shadows while avoiding visual artifacts related to rendering the avatar when the camera is inside it.
Also, it looked like the SM resolution was limited to 512^2 due to not using a separate larger hardware z-buffer that could accommodate the rendering of e.g. 2048^2 shadow map. This observation was based on static code analysis, so not sure if it's 100% accurate.
For better shadowing, we could create a separate big hardware depth texture for the shadow pass, and focus the shadow map creation only to relevant objects.
All this relies heavily on how we work with Ogre though, as the geometry render paths are completely managed by Ogre objects. Perhaps open this one again when we get into the Ogre rendering system and add materials/custom shader support.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
By looking at the configuration, it seems that we're using a single 512^2 shadow map with LiSPSM projection for the directional sky? To get the proper shadows, can it be done that the avatar geometry is rendered in the shadow generation pass but not in the color pass when in FPS mode? This would give the proper shadows while avoiding visual artifacts related to rendering the avatar when the camera is inside it.
Also, it looked like the SM resolution was limited to 512^2 due to not using a separate larger hardware z-buffer that could accommodate the rendering of e.g. 2048^2 shadow map. This observation was based on static code analysis, so not sure if it's 100% accurate.
For better shadowing, we could create a separate big hardware depth texture for the shadow pass, and focus the shadow map creation only to relevant objects.
All this relies heavily on how we work with Ogre though, as the geometry render paths are completely managed by Ogre objects. Perhaps open this one again when we get into the Ogre rendering system and add materials/custom shader support.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Assuming you are talking about the avatar's shadow, in FPS mode the avatar mesh is hidden and therefore there is no shadow to cast to the ground.
Feel free to reopen if you are referring to some other shadows.
Additionally, shadows on objects, that do not cast shadows themselves, can be enabled from app_settings\ogre_scene.xml, by setting objectshadows="true". By default they're off because they suffer from backprojecting issues (would need proper depth-shadowmaps).
By looking at the configuration, it seems that we're using a single 512^2 shadow map with LiSPSM projection for the directional sky? To get the proper shadows, can it be done that the avatar geometry is rendered in the shadow generation pass but not in the color pass when in FPS mode? This would give the proper shadows while avoiding visual artifacts related to rendering the avatar when the camera is inside it.
Also, it looked like the SM resolution was limited to 512^2 due to not using a separate larger hardware z-buffer that could accommodate the rendering of e.g. 2048^2 shadow map. This observation was based on static code analysis, so not sure if it's 100% accurate.
For better shadowing, we could create a separate big hardware depth texture for the shadow pass, and focus the shadow map creation only to relevant objects.
All this relies heavily on how we work with Ogre though, as the geometry render paths are completely managed by Ogre objects. Perhaps open this one again when we get into the Ogre rendering system and add materials/custom shader support.
By looking at the configuration, it seems that we're using a single 512^2 shadow map with LiSPSM projection for the directional sky? To get the proper shadows, can it be done that the avatar geometry is rendered in the shadow generation pass but not in the color pass when in FPS mode? This would give the proper shadows while avoiding visual artifacts related to rendering the avatar when the camera is inside it.
Also, it looked like the SM resolution was limited to 512^2 due to not using a separate larger hardware z-buffer that could accommodate the rendering of e.g. 2048^2 shadow map. This observation was based on static code analysis, so not sure if it's 100% accurate.
For better shadowing, we could create a separate big hardware depth texture for the shadow pass, and focus the shadow map creation only to relevant objects.
All this relies heavily on how we work with Ogre though, as the geometry render paths are completely managed by Ogre objects. Perhaps open this one again when we get into the Ogre rendering system and add materials/custom shader support.