* Squash development commits for PR
* Fix failing build on armeabi-v7a via android NDK
* Update with blendshape support
* Migrate to auto-cloning and patching tinyusdz (instead of manually copying files)
* Update to latest rendermesh-refactor branch commit
* Remove tracked file
* Update to use recent commit to "dev" branch
"rendermesh-refactor" was merged to "dev" around 9 May 2024 but merge
was not obvious from commit messages
* Add UNUSED() macro
(cherry picked from commit d89fe8f034c353cc5cc5b3ac78cd8845e006de38)
* Update tinyusdz branch
* Prevent per-ABI (x86, x86_64 etc) clone on android
* Add verbose logging cmake option
* Fix macro and patch
* Address compiler warnings
* Address compiler warnings
* Address compiler warnings
* Attempt prevent re-clone/re-patch once downloaded by any ABI build
* Disable tinyusdz clone/build by default
assimp github PR auto-CI checks clone/build the tinyusdz code, and reject PR
due to compiler warnings in the 3rd party external tinyusdz project
---------
Co-authored-by: Steve M <praktique-tellypresence@yahoo.com>
* Allow empty slots in aiMesh::mTextureCoords.
1.Explicitly say in documentation that empty slots are allowed (it was unclear).
2.Change GetNumUVChannels() implementation to allow empty slots.
3.Revert fraction of 2da2835b29 where empty slots are detected and error logged.
* Fix#5632 by reverting fraction of d6aacefa1e where Collada texcoords are renumbered to avoid empty slots.
---------
Co-authored-by: Kim Kulling <kimkulling@users.noreply.github.com>
The latest Cineware SDK explicitly asserts macOS support in its
documentation, but Assimp's C4D importer only works with Windows MSVC
targets. Let's improve its portability by refactoring importer code to
not depend on MSVC-only data types and quirks, and add support for linking
against the universal macOS static libraries provided in the Cineware
SDK.
Note that the C4D importer still cannot support Linux platforms because
Maxon does not provide the necessary precompiled Cineware libraries for
that platform. Windows MinGW targets are also out of the question as the
MinGW toolchain uses compiled libraries in a different format.
Co-authored-by: Kim Kulling <kimkulling@users.noreply.github.com>
- std::min/max were not defined in StackAllocator.inl; Also added explicit template arguments to break macro expansion if Windows.h is included prior and NOMINMAX macro is not present.
- Made static_assert statements compatible with C++11 in ProcessHelper.cpp.
- Removed unused string_view include in ObjFileParser.cpp.
* ColladaParser: check values length
fixes: #4286
* Refactor calculation of size for data
---------
Co-authored-by: Kim Kulling <kimkulling@users.noreply.github.com>
* Fix identity matrix check
Adds an extra epsilon value to check the matrix4x4 identity. The method is also used to export to GLTF/GLTF2 format to check node transformation matrices. The epsilon value can be set using AI_CONFIG_CHECK_IDENTITY_MATRIX_EPSILON with the default value set to 10e-3f for backward compatibility of legacy code.
* Fix type of float values in the unit test
* Update matrix4x4.inl
Fix typo
* Update matrix4x4.inl
Remove dead code.
* Add isIdentity-Test
* Update AssimpAPITest_aiMatrix4x4.cpp
---------
Co-authored-by: Kim Kulling <kimkulling@users.noreply.github.com>
When loading a mesh exported from assimp into Blender, it warns that it has an incorrect class.
While debugging, I traced it back to this being `Blendshape` where `Geometry` was expected. This
is likely because this node describes a `Geometry`, which is used as a blendshape. I'm not sure
if any other DCC tools or places to import it expect `Blendshape` instead (i.e. was this code
ever tested?), but it fixes its use in Blender.
Co-authored-by: Kim Kulling <kimkulling@users.noreply.github.com>
* Update XFileImporter.cpp
Comment out boneIdx conditional which caused massive breakage
* Update XFileImporter.cpp
Fix typo
* Update XFileImporter.cpp
Dummy whitespace change to attempt to re-trigger failing CI tests
---------
Co-authored-by: Steve M <praktique-tellypresence@yahoo.com>
Co-authored-by: Kim Kulling <kimkulling@users.noreply.github.com>
The previous table was in an incorrect order, leading to index references in DXF producing the wrong colours when converted.
Also added other entries to bring the total number of ACI colours up to the number that can be used in DXF files
Changed order of operations for insert positioning and scaling. Need to position the inserts before scaling it, otherwise the position ends up up being position*scale
Currently, when the coordIndex attribute of an IndexedLineSet contains a
polyline with > 2 indices, X3DGeoHelper::coordIdx_str2faces_arr() will
incorrectly determine the primitive type to be aiPrimitiveType_TRIANGLE or
aiPrimitiveType_POLYGON instead of aiPrimitiveType_LINE.
To fix this, this commit adds functions to explicitly handle an IndexedLineSet.
Fixes#3101
Steve M
Stepan Hrbek
Kim Kulling
Matthias Möller
Julian Knodt
Garux
ycn2022
michaelsctts
Julian Uy
陈瑞锋
Alejandro González
RoboSchmied
Kristjan Skutta
Laura Hermanns
sSsA01
Adam Mizerski
fvbj
Florian Born
Galen Xiao
seanth
Robert Reif
Andre Schulz
copycd
tangxin