#emc-devel | Logs for 2008-01-06

[01:58:53] <jmkasunich> is this a bug or am I stupid?
[01:59:24] <jmkasunich> according to http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?CoordinateSystems, the DRO should read the combined effect of G5x, G92, and G43
[02:09:43] <jmkasunich> immediately after homing, it does not - it ignores G54 (which is in effect)(
[02:10:10] <jmkasunich> if you issue G54 in MDI, the display immediately changes to the offset values, and remains that way from then on
[02:26:49] <jmkasunich> ok, I'
[02:27:02] <jmkasunich> I'm pretty convinced there is a bug that affects the DRO
[02:27:37] <jmkasunich> steps to reproduce:
[02:27:44] <jmkasunich> set non-zero G54 offsets
[02:27:46] <jmkasunich> shut down emc
[02:27:49] <jmkasunich> start up emc
[02:28:00] <jmkasunich> home axes
[02:28:14] <jmkasunich> DRO reads 0,0 (or whatever your home coords are)
[02:28:22] <jmkasunich> issue G0X0Z0 in MDI
[02:28:49] <jmkasunich> machine goes to 0,0 in the G54 coord system, but the display still reads in the G53 system (so it reads the offsets)
[02:28:58] <jmkasunich> issue G53G0X0Z0
[02:29:12] <jmkasunich> it goes back to the home position, and the DRO reads zero
[02:31:36] <jmkasunich> issue G54
[02:31:57] <jmkasunich> nothing happens
[02:32:00] <jmkasunich> issue G55
[02:32:10] <jmkasunich> nothing happens (because G55 offsets are zero)
[02:32:13] <jmkasunich> issue G54
[02:32:24] <jmkasunich> DRO changes to the current position in G54 coordinates
[02:32:37] <jmkasunich> from that point on the DRO is working correctly
[02:32:56] <jmkasunich> but prior to issing the G55, the DRO was displaying machine coords
[02:34:30] <jmkasunich> this is a lathe config, lemme try the mill config
[02:41:40] <jmkasunich> interesting - in mill config it works correctly - during and after homing the DRO indicates position in g54 coords
[02:43:23] <jmkasunich> gotta be an initial condition thing - homing a second time doesn't put the DRO back into G53, it displays G54 coords during and after homing
[02:44:00] <jmkasunich> bbl, gotta walk the dog
[02:45:00] <jmkasunich> this is 2.2.2 btw
[04:46:42] <cradek> jmkasunich: the difference is the mill configuration loads a program initially
[04:47:05] <cradek> the lathe config starts up without a program loaded. if you load a program it will set those offsets
[04:47:31] <cradek> I agree it's probably a bug somewhere but once I figured out the pattern I sort of forgot about it
[19:58:27] <skunkworks> jmkasunich: http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Lathe_Advanced_Features#Lathe_specific_radius_compensation
[19:58:48] <jmkasunich> thats the one
[19:59:46] <skunkworks> went to the timeguy.com > jeplers site > emc-axis site > that linked to the wiki ;)
[20:00:01] <Neil__> I'm planning on modifying EMC for a machine vision application I'm working on. Has anybody any recommendations/experience of any opensource toolkits in c/c++. So far I've found mimas - looking for any anecdotal info
[20:00:50] <Neil__> sorry just to be clear thats a machine vision toolkit in c/c++
[20:04:00] <SWPadnos> I have some links to various image processing libraries
[20:04:11] <SWPadnos> they aren't specifically for machine vision though
[20:04:17] <SWPadnos> I'll paste a few:
[20:04:39] <Neil__> yes please - that would be great
[20:05:00] <SWPadnos> I just want to see which ones are still active ;)
[20:05:10] <SWPadnos> also, have you heard of the Elphel digital camera?
[20:06:26] <Neil__> no - what's make it special?
[20:07:11] <SWPadnos> it has an FPGA in it, the CPU is an etrax chip that runs Linux, and all the code for both the FPGA and CPU is open source
[20:07:56] <SWPadnos> it's actually a sourceforge project (the code)
[20:08:03] <SWPadnos> they may publish schematics as well, I don't recall
[20:08:24] <SWPadnos> http://www.mathtools.net/C_C__/Image_Processing/
[20:08:33] <Neil__> just checked out their site - very cool indeed
[20:08:41] <SWPadnos> that's a list of some image processing libraries
[20:08:55] <Neil__> excellent - thank you
[20:09:03] <SWPadnos> oh -the other nice thing about the camera is that it has only one connection - ethernet :)
[20:09:25] <Neil__> hurrah - no usb!
[20:09:37] <SWPadnos> yeah - even power over ethernet
[20:10:30] <alex_joni> I only remember using OpenCV a while back
[20:11:05] <SWPadnos> ok - just found that one in my set of camera design links
[20:11:08] <alex_joni> a couple years ago, it was quite easy to setup/start using (not sure they have a linux version though)
[20:11:22] <SWPadnos> yes, they do
[20:11:26] <SWPadnos> http://sourceforge.net/projects/opencvlibrary/
[20:11:50] <SWPadnos> http://www.cs.cmu.edu/afs/cs/project/cil/ftp/html/v-source.html
[20:11:52] <alex_joni> back then it was an intel download :)
[20:11:58] <alex_joni> iirc
[20:12:44] <Neil__> thankyou - very glad I asked now - sometimes when you google around it's very hard to spot the wheat from the chaff
[20:12:50] <SWPadnos> yesh
[20:13:04] <SWPadnos> some of my links may be chaff
[20:13:11] <SWPadnos> http://www.uni-koblenz.de/FB4/Institutes/ICV/AGPriese
[20:13:19] <SWPadnos> http://www.vision.huji.ac.il/home.html
[20:13:36] <SWPadnos> that last one seems to have changed a bit, but there may still be some info available on that site
[20:14:14] <SWPadnos> and of course there's all the stuff about using GPUs for processing, which is exceedingly good for image processing
[20:14:20] <SWPadnos> http://www.fibus.org/fibusimg.htm
[20:14:37] <Neil__> alex - not sure if you saw on the user channel but I finally finished the ATC on the sabre 750. It works a treat - although it could do with being optimised. As soon as I find a simple opensource video editting app to cut the file down to a sensible size I'll post a link
[20:15:33] <Neil__> swp - thanks - lots to checkout
[20:16:23] <SWPadnos> heh
[20:17:13] <Neil__> heh?
[20:18:27] <alex_joni> Neil__: yup, saw that
[20:18:40] <alex_joni> btu you were buckie555 at the time :)
[20:19:33] <fenn> here's another one http://en.wikipedia.org/wiki/Scale-invariant_feature_transform
[20:20:26] <Neil__> alex_joni - that's right. I'm a man of many names at the moment. The joys of using an unreliable internet connection.
[20:21:04] <alex_joni> Neil__: check #emc
[20:21:33] <Neil__> thanks fenn
[20:23:11] <fenn> suppose i should point out the free sift library: http://user.cs.tu-berlin.de/~nowozin/libsift/
[20:23:33] <SWPadnos> "heh" - lots to check out. don't I know it :)
[20:23:58] <SWPadnos> it was fun finding all those links, and eliminating lots of sites that had non-free (or non-Free) libraries, etc.
[20:42:51] <Neil__> SWPadnos - very much appreciated
[20:49:35] <SWPadnos> oh - you're welcome
[20:49:57] <SWPadnos> at least they may be useful now (the project I was researching that stuff for got pretty much dropped)