Loop normals: first step needs testing!

Hi guys,

I (hopefully) mostly finished first step in loop normals work. Now, all objects have the option to enable loop normals (with an optional angle threshold above which edges are always sharp). Once enabled, final derived meshes (structs representing objects to draw or render) gets a normal CD layer for their loops (and a tessellated version for its tfaces as well). Those loop normals allow to represent flat faces and sharp edges. As a temp UI, I added loop normals and angle options in the “Item” panel, “Properties” of the 3DView.

Code can be checked out from my svn branch, https://svn.blender.org/svnroot/bf-blender/branches/vgroup_modifiers

EDIT We are now using GIT and Phabricator, so this work is now hosted in a Differential revision instead of a branch: http://developer.blender.org/D11…

These data are used in OpenGL preview (only when “VBO” option is enabled, for now, I prefer to modify viewport code as few as possible currently!), and for tangent space compute as well (note that for this one, I assumed it is only used over final dm, this may not be always true, have to check this yet). These are also the same as exported in OBJ and FBX.

Only remaining TODO for this first step is to export TSpace in FBX (btw, if someone could provide me one or two simple ASCII FBX files with TSpace data, would be nice 😉 ) – and of course, to test!

I hope the core algo to compute loop (split) normals is now OK (I fixed two bugs here recently ), but esp. tspace/loopnormals combinations are to be checked carefully (baking)…

EDIT CoDEmanX just uploaded a Windows build of the WIP branch: http://www.graphicall.org/925 Many thanks to him! 😀