MikuMikuStudio

Fork

SDK:
- change update center locations

git-svn-id: http://jmonkeyengine.googlecode.com/svn/trunk@10665 75d07b2b-3a1a-0410-a2c5-0572b91ccdca

SDK:
- Update bundled blender version to 2.67b

git-svn-id: http://jmonkeyengine.googlecode.com/svn/trunk@10664 75d07b2b-3a1a-0410-a2c5-0572b91ccdca

- Fixed the way handedness/parity is computed in the tangent binormal generator

git-svn-id: http://jmonkeyengine.googlecode.com/svn/trunk@10663 75d07b2b-3a1a-0410-a2c5-0572b91ccdca

removed the mesh collision shape from HelloTerrainCollision.java

git-svn-id: http://jmonkeyengine.googlecode.com/svn/trunk@10662 75d07b2b-3a1a-0410-a2c5-0572b91ccdca

fixed issue when saving a mesh in hw skinning mode was not restoring the buffers.
restored the code to enforce BoneIndex and BoneWeight to be in the heap for software anim because it was failing on android, until I find a proper solution

git-svn-id: http://jmonkeyengine.googlecode.com/svn/trunk@10659 75d07b2b-3a1a-0410-a2c5-0572b91ccdca

Hadrware skinning buffers are now not saved along the other buffers in a j3o file.
They are created when loading the model.
This was causing issue when loading j3o files saved before the change

git-svn-id: http://jmonkeyengine.googlecode.com/svn/trunk@10658 75d07b2b-3a1a-0410-a2c5-0572b91ccdca

Hardware Skinning now uses its own bone index and bone weight buffers. The vertex buffers are initialized empty when the model is loaded and placed in the cache.
They are populated only if hardware skinning is used with the model.

BoneIndex and BoneWeight buffers are now always CpuOnly and only used for Software Skinning.
Some enhancement could be done to save memory by not generating the bindPose buffers if hardware skinning is used as it doesn't need them.

git-svn-id: http://jmonkeyengine.googlecode.com/svn/trunk@10657 75d07b2b-3a1a-0410-a2c5-0572b91ccdca

Nifty-1.3.3

git-svn-id: http://jmonkeyengine.googlecode.com/svn/trunk@10656 75d07b2b-3a1a-0410-a2c5-0572b91ccdca

* Display LWJGL version when context is created (why was this part removed?)

git-svn-id: http://jmonkeyengine.googlecode.com/svn/trunk@10655 75d07b2b-3a1a-0410-a2c5-0572b91ccdca

* MeshLoader: submesh names were not loaded properly due to incorrect check

git-svn-id: http://jmonkeyengine.googlecode.com/svn/trunk@10654 75d07b2b-3a1a-0410-a2c5-0572b91ccdca

fixed nifty dependencies

git-svn-id: http://jmonkeyengine.googlecode.com/svn/trunk@10653 75d07b2b-3a1a-0410-a2c5-0572b91ccdca

Removed logging of unset uniform for Android to match Desktop and avoid flooding the log

git-svn-id: http://jmonkeyengine.googlecode.com/svn/trunk@10652 75d07b2b-3a1a-0410-a2c5-0572b91ccdca

Set log level to FINEST for unset uniform

git-svn-id: http://jmonkeyengine.googlecode.com/svn/trunk@10651 75d07b2b-3a1a-0410-a2c5-0572b91ccdca

Improve error message when shader uniform is not set

git-svn-id: http://jmonkeyengine.googlecode.com/svn/trunk@10650 75d07b2b-3a1a-0410-a2c5-0572b91ccdca

Matrix3f has now a normalize(Matrix3f store) and a normalizeLocal() method

git-svn-id: http://jmonkeyengine.googlecode.com/svn/trunk@10649 75d07b2b-3a1a-0410-a2c5-0572b91ccdca

SpotLigth: the cosine of the inner and outer angle of a spotlight are packed in the same float to send it to the shader (to save some varryings). This imples that their precision is no more than 0.001. This was producing wrong lighting for very close angles because the cosine were equals.
There is now a check that ensure that the outer cos is lower than the inner cos.

git-svn-id: http://jmonkeyengine.googlecode.com/svn/trunk@10648 75d07b2b-3a1a-0410-a2c5-0572b91ccdca

BatchNode now uses absolute world transforsm to update the underlying batch mesh, instead of computing the offset from next frame.
The real geometry mesh data is now used as a bind pose.
This avoids errors accumulations over time in some case and is less expensive.
See this post for the original issue http://hub.jmonkeyengine.org/forum/topic/batchnode-bug/

git-svn-id: http://jmonkeyengine.googlecode.com/svn/trunk@10647 75d07b2b-3a1a-0410-a2c5-0572b91ccdca

first set of tests for the Geometry. There's 18 tests in total, covering roughly 40% of the total number of methods in the geometry class.

To get all the tests to run, I needed to add EasyMock and some related jar files. This enables me to mock classes other the the one actually being tested to get a better test isolation.

git-svn-id: http://jmonkeyengine.googlecode.com/svn/trunk@10646 75d07b2b-3a1a-0410-a2c5-0572b91ccdca

Fixed the build path for the unit tests. Yesterday, I created a problem there. Which should be fixed now.

git-svn-id: http://jmonkeyengine.googlecode.com/svn/trunk@10645 75d07b2b-3a1a-0410-a2c5-0572b91ccdca

Fixed the junit tests. They now work through the ant script as well as when run directly.
The fix required adding all the build/* folders to the "compile tests" libraries listing as well as moving the junit.jar in that listing to the top of the list.

git-svn-id: http://jmonkeyengine.googlecode.com/svn/trunk@10644 75d07b2b-3a1a-0410-a2c5-0572b91ccdca

Fixed a bug where attempting to collide with an empty mesh (0 vertices) would throw an exception rather than returning 0 results.


git-svn-id: http://jmonkeyengine.googlecode.com/svn/trunk@10643 75d07b2b-3a1a-0410-a2c5-0572b91ccdca

GeometryBatchFactory, fixed wrong model bounds updating

git-svn-id: http://jmonkeyengine.googlecode.com/svn/trunk@10642 75d07b2b-3a1a-0410-a2c5-0572b91ccdca

LodGenerator : fixed typo

git-svn-id: http://jmonkeyengine.googlecode.com/svn/trunk@10641 75d07b2b-3a1a-0410-a2c5-0572b91ccdca

LodGenerator : updated licence

git-svn-id: http://jmonkeyengine.googlecode.com/svn/trunk@10640 75d07b2b-3a1a-0410-a2c5-0572b91ccdca

LodGenerator used squared length of en edge as parameter for its collapse cost instead of the length. It gives better results

git-svn-id: http://jmonkeyengine.googlecode.com/svn/trunk@10639 75d07b2b-3a1a-0410-a2c5-0572b91ccdca

Implemented a Lod Generator based on Ogre progressive mesh

git-svn-id: http://jmonkeyengine.googlecode.com/svn/trunk@10638 75d07b2b-3a1a-0410-a2c5-0572b91ccdca

* Fixed assertion error bug in NativeObjectManager.deleteAllObjects()
* Fixed ID collision bug in NativeObjectManager by introducing NativeObject.getUniqueId()

git-svn-id: http://jmonkeyengine.googlecode.com/svn/trunk@10637 75d07b2b-3a1a-0410-a2c5-0572b91ccdca

Android: fix missing period in javadoc

git-svn-id: http://jmonkeyengine.googlecode.com/svn/trunk@10636 75d07b2b-3a1a-0410-a2c5-0572b91ccdca

SDK: Update global library settings for Android OpenAL Soft

git-svn-id: http://jmonkeyengine.googlecode.com/svn/trunk@10635 75d07b2b-3a1a-0410-a2c5-0572b91ccdca

SDK: Update location of Android OpenAL Soft jar file

git-svn-id: http://jmonkeyengine.googlecode.com/svn/trunk@10634 75d07b2b-3a1a-0410-a2c5-0572b91ccdca