Add Assimp support.

[tanukibouwer]

This PR adds an AssimpParser to Trenchbroom that will load any entity model in a format that Assimp supports if another parser didn't get to import it first.
assimp is required as supported model format in the GameConfig.
This does not include animation, so all meshes are static.
Assimp supports mdl, so that means this includes Half-Life model support.
md5mesh and LWO are also included, but are untested.
Solves #3449.
Helps out #1140.
Solves #719, hopefully.
This code's pretty big and complicated compared to what I usually write so apologies if it's messy.
A fix for #3906 would pair well with this PR as importing 'modern' models will still take a while to do due to building the spatial tree.

[kduske]

Thanks, this looks promising! It currently doesn't build and it needs a bit of work to get it into a mergeable state (style changes and such). Are you up for making such changes yourself? In that case I'd do a proper review and leave comments in the PR. I can also take over and redo the PR myself if you don't to address lots of comments.

Oh, and, this depends on #3883, correct?

[tanukibouwer]

Thanks, this looks promising! It currently doesn't build and it needs a bit of work to get it into a mergeable state (style changes and such). Are you up for making such changes yourself? In that case I'd do a proper review and leave comments in the PR. I can also take over and redo the PR myself if you don't to address lots of comments.

Yeah, I'll be looking at this today and see if I can get this fixed.

Oh, and, this depends on #3883, correct?

Nope, my mistake. I only just noticed that commit was accidentally included.
This is my third time ever making a PR so I'm struggling a lot on the Git side, but I'll try to get it right!

[kduske]

No worries, git is hard. You're doing a great job and I can help sort out any git history issues before we merge.

[RobertBeckebans]

Nice! 👍

[tanukibouwer]

Alright, I think I've fixed the compile issue.

No worries, git is hard. You're doing a great job and I can help sort out any git history issues before we merge.

Thanks, that'd be great.

[HeadClot]

Bit of a question would this address #2628 and #2103 ? Even it it addresses it for future work I would be happy. Either way good work!

[tanukibouwer]

Bit of a question would this address #2628 and #2103 ? Even it it addresses it for future work I would be happy. Either way good work!

#2628 gets solved by this.
#2103 does not get solved by this, but Assimp does have the ability to export to glTF, so in the future we could create a general Assimp exporter that can export to any of Assimp's formats, making use of #3883's addition of export options.
And thanks!

[HeadClot]

Bit of a question would this address #2628 and #2103 ? Even it it addresses it for future work I would be happy. Either way good work!

#2628 gets solved by this. #2103 does not get solved by this, but Assimp does have the ability to export to glTF, so in the future we could create a general Assimp exporter that can export to any of Assimp's formats, making use of #3883's addition of export options. And thanks!

Thank you for the insight into your PR. Looking forward to seeing it merged into main :)

[WatIsDeze]

If in the meantime, IQM support could be added, we're golden. Because for some reason Assimp is just not into it.

Then again, for our project we might post our own PR soon enough if you guys are too occupied. No worries.

https://github.com/culacant/raylib/blob/1604fa9e475cbfaa1a1fede9d147435548bdf5ef/src/iqml.c

[kduske]

If in the meantime, IQM support could be added, we're golden. Because for some reason Assimp is just not into it.

Then again, for our project we might post our own PR soon enough if you guys are too occupied. No worries.

https://github.com/culacant/raylib/blob/1604fa9e475cbfaa1a1fede9d147435548bdf5ef/src/iqml.c

I'm not sure what the relevance to this PR is?

[HeadClot]

If in the meantime, IQM support could be added, we're golden. Because for some reason Assimp is just not into it.

Then again, for our project we might post our own PR soon enough if you guys are too occupied. No worries.

https://github.com/culacant/raylib/blob/1604fa9e475cbfaa1a1fede9d147435548bdf5ef/src/iqml.c

My advice would be to make this a separate PR to add .IQM support. My 2 cents though.

[tanukibouwer]

Compile tests should be working now.
Windows test was failing due to me using a minus operator where I shouldn't.
The Linux build was failing due to issues in Assimp's unit tests but those should now be disabled.

[WatIsDeze]

If in the meantime, IQM support could be added, we're golden. Because for some reason Assimp is just not into it.
Then again, for our project we might post our own PR soon enough if you guys are too occupied. No worries.
https://github.com/culacant/raylib/blob/1604fa9e475cbfaa1a1fede9d147435548bdf5ef/src/iqml.c

My advice would be to make this a separate PR to add .IQM support. My 2 cents though.

We will. We've already forked TB to do what we got to, of course our schedule is big too but it won't be too long from now. Then it at last has .iqm support :P

[tanukibouwer]

@kduske The compile tests should work now, could you allow them to run, please?

[tanukibouwer]

For some reason the Windows compile seems to be failing to link Assimp. I haven't been able to figure out why from the logs, so I'm gonna try compiling it on Windows myself, but installing QT is going to take ages so this might take a long while.

[ericwa]

Found the link error on Windows: https://stackoverflow.com/questions/64086365/unresolved-external-when-just-adding-an-incomplete-type-in-class-in-header

Changing the following in AssimpParser.h fixes it:

-class aiNode;
-class aiScene;
-class aiMesh;
+struct aiNode;
+struct aiScene;
+struct aiMesh;

[tanukibouwer]

Found the link error on Windows: https://stackoverflow.com/questions/64086365/unresolved-external-when-just-adding-an-incomplete-type-in-class-in-header

Changing the following in AssimpParser.h fixes it:

-class aiNode;
-class aiScene;
-class aiMesh;
+struct aiNode;
+struct aiScene;
+struct aiMesh;

Thank you so much for the info, I still haven't managed to get TB to compile on Windows so I haven't even been able to try anything myself.
We now no longer seem to get any linker issues, but now we're getting an error from building the manual..?

...
  Generating gen-manual
  Generating gen-manual/images
  Building Custom Rule D:/a/TrenchBroom/TrenchBroom/app/CMakeLists.txt
  Generating gen-manual/_reset.css, gen-manual/_responsive.css, gen-manual/_specific.css, gen-manual/_variables.css, gen-manual/default.css, gen-manual/shortcuts_helper.js
  Generating gen-manual/index.html
  fatal: No names found, cannot describe anything.
  -- Using version description "unstable-refs_pull_3910_merge-6ab6b1f792aeac002ea412ffb7a33e5eba6a2c94" from environment variables TB_PULL_REQUEST_HEAD_SHA and GITHUB_REF
  Generating gen-manual/shortcuts.js
C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\MSBuild\Microsoft\VC\v160\Microsoft.CppCommon.targets(241,5): error MSB8066: Custom build for 'D:\a\TrenchBroom\TrenchBroom\cmakebuild\CMakeFiles\a502fc3a8f4d61d0f9672e1a591f155d\gen-manual.rule;D:\a\TrenchBroom\TrenchBroom\cmakebuild\CMakeFiles\8635effa45ca0cfd450b9c2810790814\_reset.css.rule;D:\a\TrenchBroom\TrenchBroom\cmakebuild\CMakeFiles\8635effa45ca0cfd450b9c2810790814\images.rule;D:\a\TrenchBroom\TrenchBroom\cmakebuild\CMakeFiles\684b38fc4b6a1527dc5058b0de531cc8\AxisRestriction.png.rule;D:\a\TrenchBroom\TrenchBroom\cmakebuild\CMakeFiles\8635effa45ca0cfd450b9c2810790814\index.html.rule;D:\a\TrenchBroom\TrenchBroom\cmakebuild\CMakeFiles\8635effa45ca0cfd450b9c2810790814\shortcuts.js.rule;D:\a\TrenchBroom\TrenchBroom\cmakebuild\CMakeFiles\2e180fc0465b9396928512ecc0196cf6\GenerateManual.rule;D:\a\TrenchBroom\TrenchBroom\app\CMakeLists.txt' exited with code -1073741515. [D:\a\TrenchBroom\TrenchBroom\cmakebuild\app\GenerateManual.vcxproj]
  Building Custom Rule D:/a/TrenchBroom/TrenchBroom/lib/assimp/tools/assimp_cmd/CMakeLists.txt
  CompareDump.cpp
  ImageExtractor.cpp
  Main.cpp
...

[kduske]

Hey, don't worry about all of the conflicts -- I merged a branch that introduces clang-format. Once this branch is ready to merge, it's pretty easy to get it back to a mergeable state. Just let me know when you're ready.

[tanukibouwer]

I've finally managed to figure out the problem. I had forgotten to get the Assimp DLL copied at build on Windows.
@kduske Feel free to review this now.

[ericwa]

Nice! I downloaded the Win64 build from CI, and it's displaying models from Half-Life for me (I'm using the Steam edition).

I haven't tested macOS or Linux yet.

I'd propose we statically link assimp on Linux. If we don't, our official .rpm/.deb installers are going to have to install our custom .so system wide, which sounds like a recipe for problems (it's also packaged in distros, e.g. https://packages.ubuntu.com/source/focal/assimp ). Alternatively, maybe we can tell the assimp cmake to use a suffix like assimp-tb for the .so?

(Also, I think we're going the right direction with building our own copy, as opposed to using distro provided copies - as the HL MDL support has only been stabilized in 5.1.0 which is only in prerelease and won't be in distros for a while: see: https://github.com/assimp/assimp/releases )

[tanukibouwer]

I've now (hopefully!) fixed the axis problems, so models should now be displaying correctly.

Models were not displaying correctly.

Seems like I got some math and some assumptions about Assimp's workings wrong. In any case, it's all fixed now. In my own projects, everything in TB now lines up perfectly with the models in-game.
That one failing Ubuntu CI test now suddenly passes, no idea why. clang-format's still messed up, though.

In any case, this is ready for review again.

[jdolan-chwy]

This is exciting to see! And looks like it's quite close to being merged. Yay!

[fhomolka]

I saw some IQM discussion over here, and I'd like to point out that Garrux added IQM support to Assimp, so any IQM requirements would also get fixed by this.

https://github.com/assimp/assimp/blob/3cca102531466f5e5f0af0d90858a23ad83ce662/code/AssetLib/IQM/IQMImporter.h

[codecat]

Wanted to add another comment of support for merging this PR. 👍

Additionally, does this also successfully load textures from its .mtl files?

[tanukibouwer]

Wanted to add another comment of support for merging this PR. +1

Additionally, does this also successfully load textures from its .mtl files?

Thanks! It does indeed read .mtl files. It currently gives a warning saying that there's a default material that it can't find a texture for that I have yet to look into but it doesn't have any effect on the resulting import.

[Nomad7]

Assimp supports mdl, so that means this includes Half-Life model support.

I'm looking at using TB for TFC mapping and would love to see this feature implemented :)

[RobertBeckebans]

Maybe you can use this: RobertBeckebans@7bf9e3c

[tanukibouwer]

I have updated the Assimp branch to be even with the master branch. Assimp is now also built with vcpkg, using the work of @RobertBeckebans. One of the MacOS checks is failing for some reason, but everything else seems to work fine.

[tanukibouwer]

Once again, I have updated the Assimp branch to be up to date with 2022.1.

[kduske]

I think I was finally able to fix the build problem on macOS. Could you rebase onto master one more time? Then it should become green.

[tanukibouwer]

I think I was finally able to fix the build problem on macOS. Could you rebase onto master one more time? Then it should become green.

It seems the workflow 'release' routines aren't working. At least they all seem to build okay.

[kduske]

Hmm, I don't understand this failure at all. It should only attempt to upload release builds when you are pushing a tag. Did you push any tags?

[tanukibouwer]

Hmm, I don't understand this failure at all. It should only attempt to upload release builds when you are pushing a tag. Did you push any tags?

I did not, as far as I can tell.

[kduske]

Ok, I have no idea what's wrong with this PR, but it works in my own PR over here: #4048 -- so I will close this one and finalize the commits on the other PR. Hope that's okay with you?

[tanukibouwer]

Ok, I have no idea what's wrong with this PR, but it works in my own PR over here: #4048 -- so I will close this one and finalize the commits on the other PR. Hope that's okay with you?

Sure, though it seems GH says the checks have passed now.