illation/etengine

Realtime 3D Game-Engine with a focus on space sim. Written in C++ 14

etengine's Issues

Use Eigen::Vector instead of self made Vector.h

I'd like to know the reason for writing your own vector class. Is it just because of the extra dependency? Or are there other reasons?

Support MSYS2 MinGW-w64

Your project uses the cmake build system so it's not difficult to add support for MSYS2 MinGW-w64.

Please support MSYS2 MINGW64

When I saw the submodule /ETEngine/Engine/third_party/gtk/GTK-for-Windows-Runtime-Environment-Installer I immediately know your project doesn't support MSYS2 MINGW64. Because on MSYS2 MINGW64, this submodule is absolutely not needed as we have packages for GTK3 and GTK-for-Windows-Runtime-Environment-Installer is indeed made using MSYS2's GTK3 packages.

Outdated Cmake commands

Nothing big but -H. -Bbuild is apparently deprecated since 3.13. Its -S . - B build now.
Since the -H is apparently not documented it took me a while to figure out what the intend was.
https://stackoverflow.com/questions/11143062/getting-cmake-to-build-out-of-source-without-wrapping-scripts

Not urgent or anything seeing how you are still winding up to a release.
Just documenting.

Can someone explain how to build with Visual Studio 2022 ?

Changing this :

git clone https://github.com/Illation/ETEngine
cd ETEngine/Projects/Demo
cmake -G "Visual Studio 15 2017 Win64" -S . -B build
cmake --build build --target all --config Develop
cmake --build build --target install
cmake --build build --target cook-installed-resources-EtEngineDemo

Into this :

git clone https://github.com/Illation/ETEngine
cd ETEngine/Projects/Demo
cmake -G "Visual Studio 17.4.3 2022 Win64" -S . -B build
cmake --build build --target all --config Develop
cmake --build build --target install
cmake --build build --target cook-installed-resources-EtEngineDemo

Does not seems to be enought

decomposed transform component rotation is in wrong coordinate space

glm decomposition does not have a left handed version, thus the resulting world rotation is mangled as if its input had been a right handed rotation

this problem has been overlooked for a large part of the engines development, resulting in code that is adapted to work with this bug

the correct solution would be to not use decomposition here at all, but because of the adapted code the correct solution causes incorrect behaviour

Remove unnecessary null pointer checks

An extra null pointer check is not needed in functions like the following.

• RawAsset::UnloadInternal
• Scene::SetStarfield
• Singleton::DestroyInstance

lhs not defined in Vector function

ETEngine/Engine/source/EtMath/Vector.inl

Line 365 in 827b8aa

result[i] = lhs[i] + scalar;

/home/nero/repos/ETEngine/Engine/source/EtMath/Vector.inl:365: error: ‘lhs’ was not declared in this scope; did you mean ‘rhs’?
  365 |   result[i] = lhs[i] + scalar;
      |               ^~~
      |               rhs

Discord Server

Many of the engine projects on github launch discord servers lately.
How about one for ETEngine too?

error: ‘fabs’ was not declared in this scope; did you mean ‘iabs’

Hey, I am getting this error during compilation in Engine/third_party/bc7enc/bc7enc/rgbcx.h.

What's the cause and how do I fix that?

Appveyor build fails to generate libraries

For some reason vcpkg fails to install 7zip during the setup phase, and therefore cannot download gtkmm and its dependencies.
https://ci.appveyor.com/project/Illation/etengine

That said even if that worked it might fail anyway, due to the time it takes to generate the library files during a first time build. Appveyor has a time limit of 1 hour, which may be exceeded by the time the 50 or so dependencies are built.

If it did work, it still isn't ideal, as nobody wants to wait that long to validate a source build and run unit tests.

The best solution would probably be to have a release with all the libraries prebuilt, which appveyor can simply download and then proceed straight towards the engine build instead.

Add DirectX 12/Vulkan backend

Description

Currently the engine only support OpenGL backend and while it's sufficient I think a second API like DirectX 12 or Vulkan could be a nice addition. It can improve performance and can also make rendering more efficient. Moreover adding these APIs will also open up the possibility to use features such as Ray-Tracing and NVIDIA's DLSS.

DirectX 12 vs Vulkan (Metal?)

The two major candidate for an alternative backend is DirectX 12 and/or Vulkan (I think Metal should be at a lower priority for now). Since @Illation is already planning to add DirectX 12 I don't think there is a question in between the two, but I am slightly inclined towards Vulkan mainly because it is cross platform and I love using Linux. Also, in this discussion people tend to prefer Vulkan since it has better documentation and is cross platform (can be biased tough).

Final Thoughts

Adding a second backend can really increase the potential of the engine. I personally prefer Vulkan but even DirectX 12 will be a good addition compared to the current backend. I am willing to contribute to the project and would love to hear everyone's opinion regarding this. One of the major challenge is to add the second API while not breaking OpenGL meanwhile using the second API to it's fullest potential, if anyone has experience regarding this I would love to hear your opinion on how to tackle this problem.

Thanks

Font buffer size is being calculated incorrectly

I believe the following lines should not take into account the number of elements in the TextVertex struct, but rather simply store the current number of tVerts.

ETEngine/source/Engine/GraphicsHelper/TextRenderer.cpp

Line 168 in 13f203b

pFont->m_BufferStart = (int32)tVerts.size() *(sizeof(TextVertex) / sizeof(float));

ETEngine/source/Engine/GraphicsHelper/TextRenderer.cpp

Line 209 in 13f203b

pFont->m_BufferSize = (int32)(tVerts.size()) * (sizeof(TextVertex) / sizeof(float)) - pFont->m_BufferStart;

E.g, the correct code would be

pFont->m_BufferStart = (int32)tVerts.size();

and

pFont->m_BufferSize = ((int32)tVerts.size()) - pFont->m_BufferStart;