Various fixes to UV2 unwrapping and the GPU lightmapper. Listed here for
context in case of git blame/bisect:
* Fix UV2 unwrapping on import, also cleaned up the unwrap cache code.
* Fix saving of RGBA images in EXR format.
* Fixes to the GPU lightmapper:
- Added padding between atlas elements, avoids bleeding.
- Remove old SDF generation code.
- Fix baked attenuation for Omni/Spot lights.
- Fix baking of material properties onto UV2 (wireframe was
wrongly used before).
- Disable statically baked lights for objects that have a
lightmap texture to avoid applying the same light twice.
- Fix lightmap pairing in RendererSceneCull.
- Fix UV2 array generated from `RenderingServer::mesh_surface_get_arrays()`.
- Port autoexposure fix for OIDN from 3.x.
- Save debug textures as EXR when using floating point format.
We've been using standard C library functions `memcpy`/`memset` for these since
2016 with 67f65f6639.
There was still the possibility for third-party platform ports to override the
definitions with a custom header, but this doesn't seem useful anymore.
Avoid unnecessary allocation of temporary buffers for each mip, and creates
only one Image with the compressed data.
Also renames variable and reorders code for clarity.
Clarify that squish is now only used for decompression.
Documented which formats can be decompressed in Image.
We do our own image loading, threading, and memory management in Godot already,
so the only components we need from etcpak (at least as of now) are the
`Compress*` methods defined in `ProcessDxtc.cpp` and `ProcessRGB.cpp`.
So we don't need to compile or vendor the rest.
For MinGW this is tricky to do as a two-step process like it was implemented,
as `std:🧵:native_handle()` is implementation-defined and depending on
the MinGW distribution, it may or may not be a pthread handle.
With mingw-gcc as packaged in Linux distros with pthread support it worked
fine, but with llvm-mingw it was problematic.
Setting the name in the thread directly as done for Apple platforms is simpler
and works fine.
Co-authored-by: Hein-Pieter van Braam-Stewart <hp@tmm.cx>
- `etc` module was renamed to `etcpak` and modified to use the new library.
- PKM importer is removed in the process, it's obsolete.
- Old library `etc2comp` is removed.
- S3TC compression no longer done via `squish` (but decompression still is).
- Slight modifications to etcpak sources for MinGW compatibility,
to fix LLVM `-Wc++11-narrowing` errors, and to allow using vendored or
system libpng.
Co-authored-by: Rémi Verschelde <rverschelde@gmail.com>
This is a tricky one, it used to work, but it was wrong, because in such
a scenario instead of passing NULL as required by the API, it would pass
a buffer containing the `\0` terminator.
This stopped working on a specific miniupnpc version, when they fixed
some network endianess issue on Windows, to which we made a workaround,
which in turn would probably result in failures when the interface is
specified.
This commit address the issue properly, by checking the specified
interface string size, and correctly passing NULL instead of the empty
string when necessary.
Also reverts the commit that introduced the bogus workaround:
e85330231c
One of those PR when the explaination is much longer then code changes
:).
Also include public domain assets in `COPYRIGHT.txt` with Unlicence text or
dual-licensing scheme.
And document commit hashes for most thirdparty code in `thirdparty/README.md`
for clarity, and in case there's no tag matching the included version numbers.
Happy new year to the wonderful Godot community!
2020 has been a tough year for most of us personally, but a good year for
Godot development nonetheless with a huge amount of work done towards Godot
4.0 and great improvements backported to the long-lived 3.2 branch.
We've had close to 400 contributors to engine code this year, authoring near
7,000 commit! (And that's only for the `master` branch and for the engine code,
there's a lot more when counting docs, demos and other first-party repos.)
Here's to a great year 2021 for all Godot users 🎆
Updated glslang and Vulkan headers/loader following the instructions found in thirdparty/README.
glslang was updated to the 'known good' matching Vulkan SDK version 1.2.162.0. Vulkan headers and loader were updated to the commit tagged with sdk-1.2.162.0.
'vk_mem_alloc.h' and 'vk_mem_alloc.c' are unchanged since there hasn't been a new tagged release since 2.3.0.
Here's the Vulkan release notes for this update:
https://vulkan.lunarg.com/doc/sdk/1.2.162.0/windows/release_notes.html
Reverted and removed the unnecessary fix-mingw-snprintf patch for glslang as well as the mention of it in thirdparty/README.md.
-Happens on import by default for all models
-Just works (tm)
-Biasing can be later adjusted per node or per viewport (as well as globally)
-Disabled AABB.get_support test because its broken
Since we clone the environments to build thirdparty code, we don't get an
explicit dependency on the build objects produced by that environment.
So when we update thirdparty code, Godot code using it is not necessarily
rebuilt (I think it is for changed headers, but not for changed .c/.cpp files),
which can lead to an invalid compilation output (linking old Godot .o files
with a newer, potentially ABI breaking version of thirdparty code).
This was only seen as really problematic with bullet updates (leading to
crashes when rebuilding Godot after a bullet update without cleaning .o files),
but it's safer to fix it everywhere, even if it's a LOT of hacky boilerplate.
-Reworked how meshes are treated by importer by using EditorSceneImporterMesh and EditorSceneImporterMeshNode. Instead of Mesh and MeshInstance, this allows more efficient processing of meshes before they are actually registered in the RenderingServer.
-Integrated MeshOptimizer
-Reworked internals of SurfaceTool to use arrays, making it more performant and easy to run optimizatons on.
This reverts commit f697e7879b.
Part of the update introduced a regression:
https://github.com/memononen/nanosvg/issues/188.
We could include a local revert of the problematic commit but let's just do a
full revert to our previous version, and I'll re-update once the regression
is fixed upstream.
Fixes#43641.
Actually sdk-1.2.154.1 for Vulkan-Loader.
glslang is updated to bacaef3237c515e40d1a24722be48c0a0b30f75f which is the
known-good version for Vulkan-ValidationLayers 1.2.154.0.
COPYRIGHT.txt was synced with the current version of the glslang LICENSE.txt,
and `glslang/register_types.cpp` now uses the upstream definition for its
default builtin resource instead of hardcoding it.
This code currently isn't compiled (and cannot compile).
We plan to re-add OpenGL ES-based renderer(s) in Godot 4.0 alongside Vulkan
(probably ES 3.0, possibly also a low-end ES 2.0), but the code will be quite
different so it's not relevant to keep this old Godot 3.2 code.
The `drivers/gles2` code from the `3.2` branch can be used as a reference for
a potential new implementation.
Implements exit codes into the engine so tests can return their statuses.
Ideally we don't do this, and we use FIXUP logic to 'begin' and 'end' the engine execution for tests specifically.
Since realistically we're initialising the engine here we don't want to do that, since String should not require an engine startup to test a single header.
This lowers the complexity of running the unit tests and even for
physics should be possible to implement such a fix.
Upgrade OIDN to 1.1.0, the latest stable version that doesn't need
the ISPC compiler.
Documented the changes made during the removal of TBB and added a patch
file for them.
As of Godot 3.0, HQ2X is no longer used to upscale the editor theme
and icons on hiDPI displays, which limited its effective uses.
HQ2X was also used to upscale the project theme when the "Use Hidpi"
project setting was enabled, but results were often less than ideal.
The new StyleBoxFlat and SVG support also make HQ2X less important
to have as a core feature.
This decreases binary sizes slightly (-150 KB on most platforms,
-212 KB on WebAssembly release).
This partially addresses #12419.
-Added LocalVector (needed it)
-Added stb_rect_pack (It's pretty cool, we could probably use it for other stuff too)
-Fixes and changes all around the place
-Added library for 128 bits fixed point (required for Delaunay3D)
The previous code would always use SSE2 intrinsics, which is not valid
on UWP ARM platforms (and likely not on some x86 platforms either).
The patch has been submitted upstream too:
https://github.com/richgel999/jpeg-compressor/pull/13
Upstream development restarted after 13 years. Changes:
2020-02-02: Version 0.5.0
Minor speed improvement on the decompressor.
Prevent memory violation when decompressing corrupted input.
2020-01-10: Version 0.4.0
Only code & infrastructure clean-up, no new functionality.
The comment mentioned a conflict with libwebsockets, but we actually
still get this conflict even now that we don't use libwebsockets.
Not sure what component is clashing but we should basically just keep
this patch.
Follow-up to #36823.
For some weird reason 'git apply' does not error out when it does nothing,
so I missed that I did not apply the patch properly in #36823...
This broke the UWP 32-bit x86 build.
Upstream removed the option in KhronosGroup/Vulkan-Loader#260, which
breaks our current use case.
This commit reverts KhronosGroup/Vulkan-Loader#260 is our vendored
loader.
We may need to re-evaluate how we link the loader, but until then,
reverting this PR fixes Windows support after the upgrade to a recent
SDK version in #36932.
- Improve the SCsub to allow unbundling and remove unnecessary code.
- Move files around to match upstream source.
- Re-sync with upstream commit 308db73d0b3c2d1870cd3e465eaa283692a4cf23
to ensure we don't have local modifications.
- Doesn't actually build against current version 5.0.1 due to the lack
of the new ArmaturePopulate API that Gordon authored. We'll have to
wait for a public release with that API (5.1?) to enable unbundling.
