November 28, 2014, 16:08 (GMT) |
Merge branch 'master' into mesh-transfer-data |
November 27, 2014, 21:31 (GMT) |
Minor cleanups. |
November 27, 2014, 21:01 (GMT) |
Merge branch 'master' into mesh-transfer-data |
November 27, 2014, 15:22 (GMT) |
Islands 'TubeCase' WIP: Aaaannnnnnnnnnndd... Working! Now, tons of cleanup and reorg of code ahead, current one is just too much ugly. Also, have to recheck stupid raycast case, would look like vcos space transform is wrong again. |
November 27, 2014, 09:31 (GMT) |
Merge branch 'master' into mesh-transfer-data |
November 25, 2014, 20:09 (GMT) |
Merge branch 'master' into mesh-transfer-data |
November 25, 2014, 19:05 (GMT) |
Islands 'TubeCase' WIP: More fixes, everything is mostly hooked up now. Quick tests shows everything seems to work as expected so far. And, divine surprise, performances seem pretty good! Note we limit AStar to a small number of steps (4 currently), reasoning is, if we map highpoly (dst) onto lowpoly (src), this should be enough in most cases (smaller dst polys onto larger ones are unlikely to span over more than two or three src polys). And in reverse case, taking care of islands does not make much sense anyway, result shall be crappy anyway. Remaining TODOs: * Finish cutting edges handling (right now nothing is done yet to solve issues!). * Cleanup/reorganization (AStar part has been designed as generic as possible, will go to BLI), names, etc. * Double, triple and quadruple checks of mem handling (*very* easy to get a mem leak in such code :/ ). |
November 25, 2014, 19:04 (GMT) |
Merge branch 'master' into mesh-transfer-data |
November 25, 2014, 11:14 (GMT) |
Islands 'TubeCase' WIP: create AStar graph from islands. First step of actual implementation. |
November 25, 2014, 08:56 (GMT) |
Merge branch 'master' into mesh-transfer-data |
November 25, 2014, 08:54 (GMT) |
Islands 'TubeCase' WIP: implementation of AStar path algo. Theoritical code for now, not linked to anything yet. |
November 23, 2014, 19:52 (GMT) |
Merge branch 'master' into mesh-transfer-data |
November 23, 2014, 19:39 (GMT) |
Island 'Tube Case': WIP - store 'innercuts' in islands. This is the easiest part of the work - find and store edges that are borders (seams for UVs e.g.), and yet 'inside' islands, i.e. used by two or more polgons of the island. Now, we have to detect dest polys that get mapped 'accross' such edges - this is the real difficult part... Note: not sure yet whether we'll need also all 'border' edges of an island, easy to add later if needed anyway. |
November 23, 2014, 14:41 (GMT) |
Merge branch 'master' into mesh-transfer-data |
November 23, 2014, 14:37 (GMT) |
Remap: Loop raycasting: fallback to nearest check in case raycasting fails. Reason is, it's usually really bad to have most loops of a dest face to map correctly to a source island, and one or two mapping to nothing because raycasting went outside of the island's geometry. Also, in raycast/nearest queries, systematically do a copy of given co/no, we do nor loose much, and inplace modifications here was getting way too much confusing (easy to generate bugs this way, e.g. with 'refinement' process of raycasting, co/no ended up getting transformed several times!). |
November 22, 2014, 17:10 (GMT) |
Merge branch 'master' into mesh-transfer-data |
November 21, 2014, 22:56 (GMT) |
Gah, without the debug print please :| |
November 21, 2014, 22:55 (GMT) |
Minor comment fix. |
November 21, 2014, 21:38 (GMT) |
cleanup: use max data-transfer define instead of checking all flags |
November 21, 2014, 21:32 (GMT) |
picky: rename |
|