Ray-traced VC and Vulkan Transition #379
Conversation
|
Hi @jarmonik, regarding your "do not merge" comment, what we often do for NASSP dev, is leave PRs as "draft", until we're ready to merge them. Then we mark them as ready for review. That way they can't accidentally get merged early, but we can still use them as a placeholder for an ongoing project. |
- Passed the data to a client level - Addressed issue #379 (implemented but not tested) - Cleared up filename and pathname confusion in D3D9Surface.cpp - Cleared up visualization of shadow map (Debug)
-Activated pre-bake process
…hader and are operational. - Fixed bugs from IProcess.cpp - Baked light sources can be adjusted from debug controls.
|
This project is ready enough for shader development to start. Waiting feedback.. |
…edVC # Conflicts: # OVP/D3D9Client/D3D9Util.h
|
@jarmonik, what is needed for a test of this feature? |
|
There is currently nothing to test. The shader would need to be finished first. Felix24 knows how to build ray-traced textures and how to write shaders so he knows both sides of the equation. I don't know who else could do that. I suppose I could try to write some kind of shader but I have close to zero experience with 3D modeling and ray-tracing, so, I am not going to make any textures. It's a holiday season so it might take some time until things move forward. Meanwhile I try to improve the sun shadow quality. |
…BakedVC # Conflicts: # OVP/D3D9Client/D3D9Util.h
Fed shadow cascades to shaders Added a debug visualization of cascade coverage
…ine tuning required.
…e level selection.
…edVC # Conflicts: # OVP/D3D9Client/DebugControls.cpp # OVP/D3D9Client/Scene.cpp
|
The C++ side should be now in a working condition. The shaders still need some work. Feedback required regarding fully baked configuration. Ambient light probe experiment was successful, although, without HDR the small difference between dim local light and sun light causes some issues. With HDR upgrade, performance optimizations and quality improvements, this can work. With this upgrade it would be possible to block Earth glow coming in a closed payload bay. It might be good idea to wait for the release of Orbiter 2024 before merge and then start updating the stock vessels since no more DX7 compatibility is required. |
|
@jarmonik I have been testing with internal dynamic shadows using code from this commit: a193053 The result in NASSP is quite good in my opinion: https://www.dropbox.com/s/emczp6g854riys4/Screenshot%202025-01-07%2020.36.06.png?raw=1 The ambient lighting may be a little too low now but before it was much too high in my opinion as the sun seemed to illuminate the entire interior face directly opposite to it, even if there is no window in between to shine through, whereas now it is just through an actual window. Maybe eventually some other way can be found to restore some of the lost ambient light in a more realistic way if that makes sense. Anyway I have built my own version of the Orbiter 2024 D3D9 client that has it enabled but would it be feasible to have this feature pushed to OpenOrbiter, maybe with an tick box to enable/disable in the options? |
…edVC # Conflicts: # OVP/D3D9Client/D3D9Client.cpp # OVP/D3D9Client/D3D9Client.h # OVP/D3D9Client/D3D9Client.rc # OVP/D3D9Client/Scene.cpp # OVP/D3D9Client/Scene.h # OVP/D3D9Client/VideoTab.cpp # OVP/D3D9Client/gcCore.cpp # OVP/D3D9Client/resource.h # Orbitersdk/include/OrbiterAPI.h # Src/Orbiter/Orbiter.cpp # Src/Orbiter/OrbiterAPI.cpp
|
Is it possible to enable the sunlit VC shadows by default, instead of through the debug menu? |
In a future it will be automatically enabled if the VC is configured to support shadows. Right now it's still in early state of development. |
|
Would it be possible to have the state of the checkbox for enabling the shadows be saved, so that once it is enabled manually that preference is saved for that vessel and mesh? |
… cleanup pending.
|
Looks like GitHub fails to fetch sub-module/project https://github.com/Microsoft/DirectXMath How is that supposed to be done properly ? It's all inline module. No module specific binaries/libraries are being build build. |
Possibly a question for @Xyon |
"cmake magic", probably. It doesn't look like that's the most recent build issue, is the package pull still a problem? |
Right now the problem is that Vulkan SDK doesn't found. I would expect that major SDK's being installed by default on server side and shared for every project (in which case it would be search path issue) but if that's not the case then Vulkan SDK would need to be installed on GitHub and made available for the project. I have no idea how is that supposed to work. There's no rush with this. I can do development and it build's fine locally. But dependency problems should be solved before merge to main. There's a lot of work to do before that, |
Implementation of ray-traced VC textures. Issue #378
https://www.orbiter-forum.com/threads/raytraced-vc-textures-with-panel-lights.40347/
Moving to Vulkan renderer. DX9 discontinued. Code modifications to D3D9Client in main branch should be suspended. Likely going to be lost during final merge.