| 09:08.56 | CIA-48 | BRL-CAD: 03abhi2011 * r47014 10/brlcad/trunk/src/libged/simulate/ (simphysics.cpp simulate.c): |
| 09:08.56 | CIA-48 | BRL-CAD: checked to see if some subtle positioning issues could be solved by not |
| 09:08.56 | CIA-48 | BRL-CAD: calculating the BB each iteration, but that created large differences between |
| 09:08.56 | CIA-48 | BRL-CAD: the correct and rendered position, so sticking with the recalculation of BB each |
| 09:08.56 | CIA-48 | BRL-CAD: iteration approach and focusing on adding rt |
| 09:14.39 | abhi2011 | I was wondering if I can add a fix to the primitive editor in mged to take the in focus object by default and show its attributes, instead of having to type in the name each time |
| 10:36.18 | *** join/#brlcad jordisayol (~jordisayo@unaffiliated/jordisayol) |
| 11:06.00 | abhi2011 | hmm there seems to be no options for an overlap function in struct application_bundle |
| 11:10.53 | abhi2011 | hmm I ll just shoot the rays one by one at the moment using the single ray shooting functions, translating their position a bit across and up/down to shoot a grid |
| 11:23.54 | abhi2011 | hmm so to get the exact bounds of the overlap I ll have to shoot at least a 100 rays in each of the x,y, z direction |
| 11:23.59 | abhi2011 | probably more |
| 11:24.06 | abhi2011 | if the scene is large |
| 11:24.58 | abhi2011 | I ll probably get the bounding box of the scene and shoot rays at 0.1 mm intervals in each of the 3 directions in a 3d grid |
| 13:19.22 | ``Erik | that'd be an awful lot of rays... the "marching cubes" tesselator does that kinda thing and can take gobs of cpu time on even simple stuff |
| 13:20.22 | ``Erik | um, try "g-stl -8 -a 10 /path/to/db/m35.g component" for an idea, -8 uses marching cubes and -a is grid spacing in mm |
| 13:21.29 | ``Erik | (and it only shoots one direction until intersections are found) |
| 13:23.02 | ``Erik | conversly, raytrace the three views at -s100000 or so, which'd ignore the book keeping and matching parts *shrug* |
| 14:34.11 | abhi2011 | umm ok, I actually need to use librt functions to do this, so I cant do it in the command line |
| 14:35.15 | abhi2011 | if 100 is too high will probably have to settle for smaller scenes as of now with max about 20 rays in each direction |
| 14:39.19 | ``Erik | just noting ways to see the amount of computation necessary, not saying how to do your certain task :) |
| 14:44.44 | abhi2011 | :) |
| 14:55.37 | abhi2011 | hmm I need to undo a affine transformation and I have the matrix that did the transformation |
| 14:55.42 | abhi2011 | so I need to invert the matrix |
| 14:55.56 | abhi2011 | no direct macro for that I guess |
| 14:56.01 | abhi2011 | in vmath.h |
| 15:47.08 | abhi2011 | hmm ok just transposing the upper left 3 by 3 matrix and inverting the translation vector seems to work for inverting an affine transf. |
| 16:05.15 | abhi2011 | phew!...solved a particularly annoying positioning issue : http://postimage.org/image/113v0wsmc/full/ |
| 16:06.17 | abhi2011 | since bullet returns world transformations, it's important to undo the object transformation in mged, before applying rt_matrix_transform() |
| 16:06.35 | abhi2011 | I was inverting the translation from the previous iteration |
| 16:06.39 | abhi2011 | but not the rotation |
| 16:07.02 | abhi2011 | leading to subtle differences, like objects outside their BB over large iterations |
| 16:07.59 | CIA-48 | BRL-CAD: 03abhi2011 * r47015 10/brlcad/trunk/src/libged/simulate/ (simcollisionalgo.cpp simulate.c simulate.h): Corrected positioning code and removed BB calculation in every iteration |
| 16:09.07 | abhi2011 | Also another change is that I do not calculate the BB every iteration, just at the first iteration, to give me something(a cuboid) to create as a collision shape in the bullet world |
| 16:11.37 | abhi2011 | after that bullet manages the aabb quiet well and its best not to recalculate the AABB and insert larger cubes into the physics world when the simulation has begun |
| 16:12.19 | abhi2011 | that tends to lock the objects into their rotated position leading to no manifolds being generated etc.....well all set to shoot rays now |
| 16:17.01 | CIA-48 | BRL-CAD: 03abhi2011 * r47016 10/brlcad/trunk/src/libged/simulate/ (simcollisionalgo.cpp simulate.c): Corrected some // comments in C code to prevent warnings |
| 17:27.32 | CIA-48 | BRL-CAD: 03tbrowder2 * r47017 10/brlcad/trunk/doc/docbook/resources/brlcad/book-covers-fo-template.xsl: mod to allow unique colors file per each en book |
| 17:28.16 | CIA-48 | BRL-CAD: 03tbrowder2 * r47018 10/brlcad/trunk/doc/docbook/resources/brlcad/brlcad-fo-stylesheet-covers.xsl: remove old multi-book file not usable for parallel make |
| 17:29.18 | CIA-48 | BRL-CAD: 03tbrowder2 * r47019 10/brlcad/trunk/doc/docbook/resources/brlcad/ (4 files): add unique entry stylesheet per en book |
| 17:30.40 | CIA-48 | BRL-CAD: 03tbrowder2 * r47020 10/brlcad/trunk/doc/docbook/BRLCAD_DOC.pm: mods for generating and inserting unique color file per en book |
| 17:32.28 | CIA-48 | BRL-CAD: 03tbrowder2 * r47021 10/brlcad/trunk/doc/docbook/create-book-covers.pl: mods for generating and using unique cover and color file per en book |
| 17:34.09 | CIA-48 | BRL-CAD: 03tbrowder2 * r47022 10/brlcad/trunk/doc/docbook/Makefile.am: mods to allow en book pdf generation with no parallel make restrictions; some convenience definition macros used |
| 18:28.20 | abhi2011 | hmm trying to find a simple way to draw a line for the rays, like nirt does |
| 18:28.44 | abhi2011 | draw.c seems to do it by inventing a solid....a bit confusing what its doing though |
| 18:29.03 | abhi2011 | _ged_invent_solid() in draw.c |
| 20:26.36 | *** join/#brlcad merzo (~merzo@237-14-132-95.pool.ukrtel.net) |
| 22:23.42 | *** join/#brlcad b0ef (~b0ef@78.58.34.95.customer.cdi.no) |
| 22:23.42 | *** join/#brlcad starseeker (~starseeke@BZ.BZFLAG.BZ) |
| 22:23.42 | *** join/#brlcad dtidrow_desk (~dtidrow@c-68-84-167-135.hsd1.mi.comcast.net) |
| 23:55.32 | CIA-48 | BRL-CAD: 03abhi2011 * r47023 10/brlcad/trunk/src/libged/simulate/ (simcollisionalgo.cpp simphysics.cpp simulate.c simulate.h): Added manifold drawing code |