User:Anuragmurty

About me

• Name: Anurag Murty
• E-mail address: anuragmurty@gmail.com
• IRC username: anuragmurty

I am an MTech student of Computational Science at IISc Bangalore

About The GSoc Project

Brief Description of my project on the GSoC page

Development Log

First task is compilation of BRL-CAD

BRL-CAD succesfully built from source

Learning to use the Software / Explore the code

Have Completed the tutorial on BRL-CAD available on the site(Basically mged commands and USING the software)

Raytracing Commands related to my project work

having learnt to use rt , rtweight from a USERs perspective, going through the respective codes and understanding the modules now

Wrote a small patch for bitv.c in src/libbu.Waiting for feedback:)

Going through the gqa program as I have been asked to learn to use and understand it too

patch for ctype.c also submitted. waiting for reviews. had some problem with the previous patch too. Turned out that it wasnt in the unified diff format as required.

[ PS: Some problem with IRC. Unable to join the brlcad channel. Should be fixed in a while hopefully.]

      21st to 23rd -- done understanding code for rtexample.c. discussed with mentor(d_rossberg) about how to extend rt_shootray for getting the voxels. 

parameters to be used : size of voxels,rays to be shot per voxel, minimum volume to be considered as threshold for considering. have started with implementation of a first , very basic program to get voxels.(one ray per voxel for now, simple shapes also).

first patch accepted.. test_bitv.c!! second patch i submitted test_ctype has to be modified according to hacking guidelines...

working on rt_shootray for 3 views, which will be submitted today, but in a little while.

Written rtexample_3view.c that uses the midpoints of the faces of the bounding rectangular parallelopiped to shoot three rays in three perpendicular directions and gives data of the partitions encountered(the hit code is exactly taken from rtexample.c). the same is being submitted as a patch.

Indentation changes made to test_ctype. Also, a typo related patch was submitted yesterday. Working on shooting a grid of rays at object to be raytraced(towards the negative x-axis). Should be able to submit that as a patch tonight or maybe tomorrow.

i will be going through codes for libanalyze and libged commands as directed by brlcad. Also, I will not be available on IRC tonight(as i will be travelling)

30/5/2012 : Got commit access today. Lot of time spent trying to rebuild brlcad. Some problems with MGED dependencies. Everything deleted and rebuilt. Working fine now. Though still unable to locate what went wrong. Presently completing the code for shooting a grid of rays and putting it in src/conv so I can make my first commit.

31/5/2012 made my first commit. A simple, safe g-voxel program which did not actually get voxels. Just shot a grid of rays.

1/6/2012 i wrote a simple voxelize program. right now, it just prints if the voxel in the path of each ray is in or out. Right now, the parameters are hardcoded(parameters like threshold a voxel is to be filled to consider as in, size of voxel in each dimension , others). I think i also will have to consider some other method apart from only using the bounding box as in some cases(example, a 3-D plus sign) too many rays are shot which might be a waste of time.

5/6/2012 have shifted the evaluation of voxel to main and writing code for multiple rays per voxel. though multiple rays through a voxel is buggy(minor, most probably). Will commit as soon that is done

6/6/2012 there was problem with voxel calculations when hitDistOut was on a voxel boundary. Also, I was computing the ending ray data slightly wrong. this was also rectified.

multiple rays are being shot now-- through same point because i have to ask my mentor whether those rays have to be random and how to proceed on volume calculations after that.

I have two variables for now- ray num threshold, threshold- depending on how the IN/OUT of a voxel has to be determined. one of these two will be discarded.

7/6/2012 studying gqa.c

8/6/2012 Made lots of changes to previous code of g-voxel.c. Also, now running "./g-voxel [.g file] [region-name]" in bin/ of the build directory outputs a file "voxels.txt" that gives a frame by frame representation of the desired region.

12/6/2012 g-voxel.c modified so it allows multiple rays per voxel. for input raysPerVoxel what I do is shoot a uniform raysPerVoxel*raysPerVoxel rays, so probably this variable name is a misnomer. again the output is voxels.txt which gives the output text file frame by frame.

14/6/2012 same as last time, floats replaced by fastf_t. some unnecessary statements variables and statements removed(mistakes pointed out by d_rossberg).. a major error was allocating space for ALL voxels. so now as soon as a raytrace is done, the results are printed out to file voxels.txt. thus at any point space for only numVoxels[0] is assigned, major savings.

I need to change the structure for rayInfo so it includes region names also. this means using pp->pt_regionp->reg_name from struct partition but i am presently getting bugs in the result. so working on this so that I canc hange the output format so it lists data region wise.

18/6/2012 -20/6/2012 structure for rayInfo modified. rayInfo is structure for storing info each time a raytrace is called. It contains a linked list of regions inside each voxel. voxels.txt now has data in the format:

1/0 (coordinates of voxel) (region-name) (fill fraction)

where 1/0 tell you if the voxel is in or not(according to threshold).

also to store names in a local variable of main(), i have used bu_vls_strgrab function.

21/6/2012 corrected a bug i found while doing some tests on g-voxel.c with some .g file inputs

26/6/2012 first step towards moving the voxelize work into a library. made a function that is called from main in g-voxel.c and takes in as input userparameters for a voxelize and outputs a file with the necessary voxel information. used a structure for input parameters and additional inputs might be added later. TODO: implement a callback function for writing result into the file.

28/6/2012 implemented callback function for printing results of g-voxel.c. moving it to libanalyze is next todo. slight problem being faced : i am unable to commit using svn commit. svn: Commit failed (details follow): svn: OPTIONS of 'https://brlcad.svn.sourceforge.net/svnroot/brlcad/brlcad/trunk/src/conv': Could not create SSL connection through proxy server: Could not authenticate to proxy server: rejected Digest challenge (https://brlcad.svn.sourceforge.net) trying to fix for a long time now

29/6/2012 callback committed, some more modifications

2/7/2012 library- libanalyze changed for accommodating voxelize functions, in voxels.c in libanalyze , all the functions are defined. also, g-voxel.c modified and uses functions from libanalyze