- Add warning in tests when stabilizing screenshots fails
(Note - it still fails sometimes, just less now with trying to get a good screenshot)
git-svn-id: svn://svn.code.sf.net/p/irrlicht/code/trunk@5150 dfc29bdd-3216-0410-991c-e03cc46cb475
- Set the platform SDK for VS2010 and VS 2012 throughout to Windows7.1SDK
- Fix the fix for compiling on GCC5
git-svn-id: svn://svn.code.sf.net/p/irrlicht/code/trunk@5141 dfc29bdd-3216-0410-991c-e03cc46cb475
- Fix CSoftwareTexture2 calculation for OriginalSize of ITexture. It had returned the changed texture size instead of the original one.
git-svn-id: svn://svn.code.sf.net/p/irrlicht/code/trunk@5138 dfc29bdd-3216-0410-991c-e03cc46cb475
Also clarified animation documentation to describe current behavior more exactly.
CSkinnedMesh had returned the last key instead of the number of keys.
Thx to whoever mentioned to me once that our example dwarf is not playing his full animation in the meshviewer.
git-svn-id: svn://svn.code.sf.net/p/irrlicht/code/trunk@5118 dfc29bdd-3216-0410-991c-e03cc46cb475
- Fix loading of .X and .B3D models with non-normalized quaternion rotations
git-svn-id: svn://svn.code.sf.net/p/irrlicht/code/trunk@5102 dfc29bdd-3216-0410-991c-e03cc46cb475
.x meshloader regards now AnimTicksPerSecond (thx @qian for a test-model and bugreport).
CAnimatedMeshSceneNode::setMesh had commented-out the setAnimationSpeed line in version r3526 which was about a joint-cache fix for skinned meshes. But there was no comment about why that line had to be removed or commented out, so my guess is that this was only a test (I hope). And it caused animation-speed values for the meshes to be ignored unless user specified it explicitly while animation range was still changed in setMesh.
git-svn-id: svn://svn.code.sf.net/p/irrlicht/code/trunk@5097 dfc29bdd-3216-0410-991c-e03cc46cb475
That dreadFun function looks rather suspiciously like some test which doesn't belong into Irrlicht anyway...
git-svn-id: svn://svn.code.sf.net/p/irrlicht/code/trunk@5096 dfc29bdd-3216-0410-991c-e03cc46cb475
- Interface getMeshType moved from IAnimatedMesh up to IMesh.
- Static b3d mesh-writer written by Hendu, support for animated meshes added by JLouisB, testing and bugfixes by CuteAlien.
git-svn-id: svn://svn.code.sf.net/p/irrlicht/code/trunk@5095 dfc29bdd-3216-0410-991c-e03cc46cb475
You can still access the frustum boundingbox through the frustum itself.
Also CSceneCollisionManager collision functions ignore now empty collision boxes.
This means cameras no longer show up in the node-collision of the SceneCollisionManager.
Tests have been adapted correspondingly.
Note that this behavior was once before changed in Irrlicht (rev. 345). But it was one of many changes and there was no documentation or log-message about why anyone would like
the cameras to show up in node-collisions based on their frustum boundingbox. Without any hints about the why I decided to change it back as it's just confusing.
git-svn-id: svn://svn.code.sf.net/p/irrlicht/code/trunk@5094 dfc29bdd-3216-0410-991c-e03cc46cb475
(checked ranges of available values - casts look fine there).
git-svn-id: svn://svn.code.sf.net/p/irrlicht/code/trunk@5093 dfc29bdd-3216-0410-991c-e03cc46cb475
Move ECOLOR_FORMAT enums added in r4984 to the end (adding in them in the middle can mess with serialization).
git-svn-id: svn://svn.code.sf.net/p/irrlicht/code/trunk@5092 dfc29bdd-3216-0410-991c-e03cc46cb475
This was caused by quick calls to XUnmapWindow followed by XMapWindow which some WM's can't handle.
Adding a long sleep call (~200ms) in between would have been a workaround, but looked ugly as the window is hidden in the meantime.
I found no reference in X11 documentation that unmapping is necessary for this, so I removed those calls now.
Changing the resize flags still works in my tests on Debian and Ubuntu. If it causes problems we have to switch to the ugly hack
or get certain systems to repair their window managers.
git-svn-id: svn://svn.code.sf.net/p/irrlicht/code/trunk@5091 dfc29bdd-3216-0410-991c-e03cc46cb475
- new media for OpenGL-drawVPL_h. This test is about finding out if things changed, and this result has been like it is now at least since 1.8
- required match can be reduce for this test, as minor aliasing artifacts always are different between cards and don't matter here.
git-svn-id: svn://svn.code.sf.net/p/irrlicht/code/trunk@5090 dfc29bdd-3216-0410-991c-e03cc46cb475
The defaults had been changed in Irrlicht 1.6 (svn r2925) as line-smoothing did sometimes lead to software emulation under OpenGL.
git-svn-id: svn://svn.code.sf.net/p/irrlicht/code/trunk@5089 dfc29bdd-3216-0410-991c-e03cc46cb475
- viePortText has improved according to Thomas
- flyCircleAnimator... again brightness differences (maybe due to latest material fixes). If that comes up again will have to investigate further.
git-svn-id: svn://svn.code.sf.net/p/irrlicht/code/trunk@5088 dfc29bdd-3216-0410-991c-e03cc46cb475
Otherwise materials like EMT_TRANSPARENT_ALPHA_CHANNEL no longer worked.
Not sure if this is now causing other problems, but it no longer seems to break the known cases (tests about that pass again). If there are new problems we need new tests to show 'em.
git-svn-id: svn://svn.code.sf.net/p/irrlicht/code/trunk@5086 dfc29bdd-3216-0410-991c-e03cc46cb475
(new result looks better, media just didn't get updated when it was changed).
git-svn-id: svn://svn.code.sf.net/p/irrlicht/code/trunk@5084 dfc29bdd-3216-0410-991c-e03cc46cb475
Was using the wrong variable (RelativeRotation instead of RelativeScale) as default value.
This was also not working in older Irrlicht versions, but can't be fixed there as those had no default parameters for getAttribute yet.
This allows for example to load .irr files created with new CopperCube while setting a parent-node (some attributes will still be ignored as .irr files always expect the scene itself as parent, that would need some special SceneNode type to work around).
git-svn-id: svn://svn.code.sf.net/p/irrlicht/code/trunk@5081 dfc29bdd-3216-0410-991c-e03cc46cb475
cutealien
nadro