Advanced Atmospheres

[Schneegans]

This PR adds the possibility to add alternative atmospheric models to CosmoScout VR. As an example, it adds the excellent scattering model by Eric Bruneton.

For reviewing these changes I would suggest to use the GitHub diff only for the files not in plugins/csp-atmospheres. All files in plugins/csp-atmospheres were nearly rewritten so that it may be easier to look at them as a new plugin.

The csp-atmospheres plugin received a major refactoring. Here are the main consequences:

• The attributes enableWater, waterLevel, enableClouds, and cloudAltitude are now stored per planet. Enabling water on Earth will not flood Mars anymore.
• waterLevel and cloudAltitude are now given in meters, not relative to the atmosphere height anymore.
• cloudAltitude can now be configured at runtime in the user interface.
• In non-HDR mode, the atmosphere now performs filmic tonemapping which results in much less over-exposure around the sun during sunset or sunrise.
• The shaders are now loaded from files and not compiled into the binary.
• No support for light shafts anymore. While not impossible, it would be pretty difficult to implement this properly with Eric Bruneton's model given that we use cascaded shadow maps. Furthermore, this feature used to have a terrible performance anyways.
• No graphical tests of the atmosphere anymore. With the new architecture, it's not possible anymore to render an atmosphere without CosmoScout's core classes. We need a new concept for graphical tests anyways...

Other changes:

• pApproximateSceneBrightness is now inverted. Before it actually contained something like "approximate scene darkness" :wink:
• The default exposure and glare values as well as the glare-slider mapping have been tweaked for a better appearance of the atmospheres in HDR mode.
• The triangle ordering in the ToneMappingNode has been fixed. It now does not need to disable culling.
• The default star rendering mode has been changed to eSmoothDisc which looks better in HDR mode (IMHO).

This should be discussed:

• Should HDR mode be enabled per default? It looks better, but also brings some performance issues.

[coveralls]

Pull Request Test Coverage Report for Build 4447331638

• 0 of 773 (0.0%) changed or added relevant lines in 10 files are covered.
• 125 unchanged lines in 11 files lost coverage.
• Overall coverage decreased (-2.0%) to 1.141%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
plugins/csp-atmospheres/src/models/bruneton/Model.hpp	0	1	0.0%
plugins/csp-atmospheres/src/models/cosmoscout/Model.hpp	0	1	0.0%
plugins/csp-stars/src/Stars.cpp	0	1	0.0%
plugins/csp-atmospheres/src/Atmosphere.hpp	0	2	0.0%
plugins/csp-atmospheres/src/models/bruneton/internal/Model.hpp	0	2	0.0%
plugins/csp-atmospheres/src/models/cosmoscout/Model.cpp	0	48	0.0%
plugins/csp-atmospheres/src/models/bruneton/Model.cpp	0	56	0.0%
plugins/csp-atmospheres/src/Plugin.cpp	0	76	0.0%
plugins/csp-atmospheres/src/Atmosphere.cpp	0	169	0.0%
plugins/csp-atmospheres/src/models/bruneton/internal/Model.cpp	0	417	0.0%
<!--	Total:	0	773	0.0%	-->

Files with Coverage Reduction	New Missed Lines	%
src/cs-graphics/SetupGLNode.hpp	1	0%
src/cs-utils/TestImageCompare.hpp	1	0%
plugins/csp-atmospheres/src/Plugin.cpp	2	0%
src/cs-utils/FrameStats.hpp	2	0%
plugins/csp-atmospheres/src/Atmosphere.cpp	3	0%
src/cs-utils/DefaultProperty.hpp	3	0%
src/cs-graphics/SetupGLNode.cpp	6	0%
src/cs-utils/utils.cpp	12	23.68%
src/cs-utils/Property.hpp	15	0%
src/cs-utils/TestImageCompare.cpp	31	0%
<!--	Total:	125		-->

Totals
Change from base Build 4366064776:	-2.0%
Covered Lines:	193
Relevant Lines:	16915

---

💛 - Coveralls

[JonasGilg]

When the sun is over the horizon, but behind some terrain these artifacts come up.

This also happens in non HDR mode, although less intense:

[JonasGilg]

On Mars there seems to be a very hard cutoff with HDR enabled:

[Schneegans]

On Mars there seems to be a very hard cutoff with HDR enabled:

This is due to the lack of multi-scattering in the Cosmoscout model. This should be the same as in the main branch.

[Schneegans]

When the sun is over the horizon, but behind some terrain these artifacts come up.

This is a limitation of the Bruneton model. It is the result of floating point precision issues. Maybe we can improve this in a custom model in the future!