JogAmp IRC Chat Logs

#jogamp @ irc.freenode.net - 20131208 05:05:56 (UTC)

20131208 05:05:56 -jogamp- Previous @ http://jogamp.org/log/irc/jogamp_20131207050556.html
20131208 05:05:56 -jogamp- This channel is logged @ http://jogamp.org/log/irc/jogamp_20131208050556.html
20131208 08:51:06 * monsieur_max (~maxime@anon) has joined #jogamp
20131208 10:20:52 * xranby-64 (~familjen@anon) has joined #jogamp
20131208 11:27:20 * Eclesia (~eclesia@anon) has joined #jogamp
20131208 11:27:24 <Eclesia> hi
20131208 11:58:45 <Eclesia> I have a strange bug when loading on image
20131208 11:59:22 <Eclesia> Image is a basic RGB 3*UByte of size 274*306
20131208 11:59:40 <Eclesia> when loading it I have the error :
20131208 11:59:49 <Eclesia> Exception in thread "Thread-2-FPSAWTAnimator-Timer0" java.lang.IndexOutOfBoundsException: Required 252142 remaining bytes in buffer, only had 251532
20131208 12:00:31 <Eclesia> 274*306*3 = 251532
20131208 12:00:45 <Eclesia> so why is he looking for 252142 bytes ?
20131208 15:19:02 <Eclesia> no idea ?
20131208 15:41:34 <Eclesia> sgothel: :(
20131208 15:42:37 <rmk0> hm, i've seen that happen before with non power of two textures
20131208 15:42:51 <rmk0> i assumed it was my fault and didn't bother with it...
20131208 15:43:19 * rmk0 attempts to reproduce
20131208 15:45:41 <rmk0> reproduced
20131208 15:46:22 <Eclesia> so it's not just me ^^
20131208 15:46:46 <rmk0> http://waste.io7m.com/2013/12/08/TestTSize.java
20131208 15:46:59 <rmk0> "Required 252142 remaining bytes in buffer, only had 251532"
20131208 15:47:52 <Eclesia> exactly, like you said, it is somekind of power of 2 or 8 somewhere in the opengl spec
20131208 15:48:07 <Eclesia> but I couldn't find anything so far
20131208 15:48:18 <rmk0> it used to be a hard requirement that texture widths and heights had to be powers of 2
20131208 15:48:38 <rmk0> they added non-po2 textures as an extension, and now i think it's allowed in core
20131208 15:49:14 <rmk0> yeah... is part of opengl 2.1
20131208 15:49:43 <rmk0> oh, 2.0, actually
20131208 15:50:03 * Eclesia works in GL4 so it should be working
20131208 15:52:09 <rmk0> wonder where 252142 comes from...
20131208 15:52:16 <Eclesia> If I change the texture size on the fly, all the UV will be wrong, I don't have any 'clean' workaround for this problem :/
20131208 15:54:09 <rmk0> ah, i think it's GL_UNPACK_ALIGNMENT
20131208 15:55:21 <rmk0> gl.glPixelStorei(GL.GL_UNPACK_ALIGNMENT, 1);
20131208 15:55:46 <rmk0> try calling that before uploading texture data
20131208 15:56:19 * Eclesia testing
20131208 15:56:52 <rmk0> the default is to have rows of pixel data start on word-aligned boundaries
20131208 15:56:58 <rmk0> which makes that expected buffer size that much larger
20131208 15:57:29 <rmk0> with power of two textures, i imagine it "always works", mostly by coincidence
20131208 15:57:58 <Eclesia> nice it works
20131208 15:58:15 <rmk0> opengl is pretty much russian roulette
20131208 15:58:44 <Eclesia> there must be a correct way to calculate this value
20131208 15:59:57 <rmk0> i think it's more that you have to guarantee that your texture data has correctly aligned rows
20131208 16:00:20 <rmk0> i'm not sure anyone actually does that in practice... is no problem to use byte-aligned rows
20131208 16:00:51 <rmk0> if your rows are word-aligned, then your texture will automatically be the expected size in bytes, obviously
20131208 16:01:41 <Eclesia> word size is 2 or 4 in opengl ?
20131208 16:02:31 <Eclesia> ha, find an explication here : http://stackoverflow.com/questions/15052463/given-the-pitch-of-an-image-how-to-calculate-the-gl-unpack-alignment last post
20131208 16:02:58 <rmk0> 4
20131208 16:09:48 <Eclesia> rmk0> thanks for the help
20131208 16:10:35 <rmk0> no problem
20131208 16:14:59 <Eclesia> great week, I have more image readers then the jvm + imageio together ^^
20131208 16:15:07 <Eclesia> and plenty of bugfix
20131208 16:21:53 <sgothel> GLBuffers <- to calculate correct buffer size
20131208 16:22:03 <sgothel> we use same to validate ..
20131208 16:22:33 <Eclesia> hi sgothel
20131208 16:23:11 <Eclesia> thanks but a bit late :D
20131208 16:23:19 <sgothel> it's never too late :)
20131208 16:24:00 <Eclesia> have you played with jpeg2000 decoding ?
20131208 16:24:01 <sgothel> so now you know how we do it ..and validate it .. a backtrace of GLBuffers.sizeof(..) gives a bit more insight
20131208 16:24:50 <sgothel> nope, not yet .. wait, I guess around 2000 I ran a few tests to heat up my room in winter, now we have those cryptocurrencies for that :)
20131208 16:29:17 <Eclesia> it's the last format from java ImageIo lib I don't have yet. I think It will take a few weeks to do it
20131208 16:31:10 <Eclesia> freedom is near ^^
20131208 16:36:06 * hija (~hija@anon) has joined #jogamp
20131208 17:04:11 <hharrison> sgothel: how are things? Manage to keep your feet dry this weekend?
20131208 17:43:05 <sgothel> Hi Harvey, yes - hehe - we watches the sea on to of our levies .. now it's almost calm again
20131208 17:44:53 <sgothel> Still 'warm' in BC I assume ?
20131208 18:33:57 <hharrison> Warming a bit today
20131208 18:47:52 <hharrison> to be fair it was only down to -6C or so
20131208 18:48:23 <sgothel> wow, we are still around [0 .. +4]C
20131208 18:49:02 <hharrison> that's more like what we are usually at...or maybe a bit warmer
20131208 18:52:38 <hharrison> My sister in law was in -38 most of the week
20131208 18:53:06 <sgothel> over the 'hill' or just a few miles away from the waterfront ? :)
20131208 18:53:33 <sgothel> right .. your water is cold ..
20131208 18:56:21 <hharrison> Indeed, over the 'hill'
20131208 18:57:12 <hharrison> We've been working hard for a big conference this week, the big 'unveiling' of jaamsim to the simulation community
20131208 18:57:31 <sgothel> great .. you record it ?
20131208 18:57:42 <sgothel> i.e. public stuff ?
20131208 18:58:17 <hharrison> I'm not at this one, doing one in April
20131208 18:58:29 <hharrison> Papers are public, not sure about recorded sessions
20131208 18:58:57 <hharrison> Wlll hopefully have a nice public demo of a terrian rendering sometime in the new year
20131208 18:59:13 <sgothel> you know .. it's easy, find a photoshop .. drive a few miles .. don't forget the tripod :)
20131208 18:59:14 <hharrison> w/ public data so people can try it
20131208 18:59:23 <sgothel> sweet!
20131208 19:00:04 <hharrison> We have some 3d-shells of buildings in downtown vancouver, for siggraph 2014, we should use that as a jogl-promoter
20131208 19:00:33 <sgothel> perfect .. something is happening ahead of time :)
20131208 19:00:34 <hharrison> 'come find the jogamp guys' -> here
20131208 19:01:03 <sgothel> http://ekisto.sq.ro/ :)
20131208 19:01:16 <sgothel> I am behind the yahoo building .. damn it :)
20131208 19:01:30 <sgothel> couldn't do a 'JogAmp' search .. too bad
20131208 19:02:25 <sgothel> oh .. github.2012 data only ..
20131208 19:02:52 <hharrison> Out with the dog for a bit, but now that the conference is happening, I'll be more active around here
20131208 19:03:11 <sgothel> sweet .. wuff
20131208 19:03:48 <hharrison> back soon
20131208 19:04:25 <xranby-64> cool tuff! it appears my tower got clear view :)
20131208 19:04:29 <xranby-64> stuff
20131208 19:04:37 <xranby-64> oh no
20131208 19:05:01 <xranby-64> its behind demoscenepassivist hehe
20131208 19:06:30 <xranby-64> it appears we got a jogamp suburb heind that yahoo building :)
20131208 19:06:54 <sgothel> wonder where we are today :)
20131208 19:07:01 * Eclesia hear 'public' . where ? where ? where ? xD
20131208 19:07:14 <Eclesia> hi xranby-64
20131208 19:07:14 <sgothel> (08:01:03 PM) sgothel: http://ekisto.sq.ro/ :)
20131208 19:08:44 <xranby-64> Eclesia: hi!
20131208 19:09:25 <Eclesia> somkind of social graph viewer ?
20131208 19:09:40 <xranby-64> yes, you may choose the github dataset
20131208 19:09:47 <sgothel> http://processq.tumblr.com/post/69098066993/ekisto-design
20131208 19:10:33 <Eclesia> funny
20131208 19:12:22 <Eclesia> no bitbucket :(
20131208 19:12:32 <Eclesia> nice anyway
20131208 19:14:38 <Eclesia> ouch, I'm near thinkerpop because I worked on orientdb. shame on me.
20131208 19:16:41 <Eclesia> finished tiff bw, grayscale,rgb reader :)
20131208 19:17:24 <xranby-64> Eclesia: dicom?
20131208 19:17:40 <xranby-64> the xray image format used at hospitals
20131208 19:17:48 <Eclesia> just the maven module. no reader yet
20131208 19:18:42 <Eclesia> made SGI,RAW,PBM,PGM,PPM and partial tiff this week. that's a nice improvement in a single week :)
20131208 19:19:23 <Eclesia> and WBMP, but no tests for this one, couldn't find a single file of this format
20131208 19:20:18 <Eclesia> the WAP spec in completly obsolete with the new generation of mobile devices
20131208 19:20:24 <xranby-64> Eclesia: i have some discs filled with dicom images since my last hospital lunge collapse visit
20131208 19:21:20 <Eclesia> I have,'t look at the spec yet. I discovered this formatonly recently
20131208 19:22:00 <sgothel> http://amide.sourceforge.net/
20131208 19:22:03 <Eclesia> I want to explore jpeg 2000 first ^^
20131208 19:22:22 <sgothel> (talking about 3d voxels, volumes, RMI/CT ..)
20131208 19:22:30 <hharrison> xranby: there are a few dicom-viewer projects that used java3d
20131208 19:22:43 <sgothel> @Eclesia: Consider using OpenCL/OpenGL for jpeg2000 ?
20131208 19:22:55 <sgothel> WebP ?
20131208 19:22:59 <hharrison> see the recent post in the java3d forum on jogamp
20131208 19:23:14 <xranby-64> sgothel: hharrisonthanks.. i will check how they work in practise
20131208 19:23:22 * xranby-64 (~familjen@anon) has left #jogamp
20131208 19:23:38 <hharrison> http://forum.jogamp.org/YaDiV-A-Software-for-Medical-Visualization-using-Java3D-td4030802.html
20131208 19:24:03 <sgothel> nice
20131208 19:24:06 <hharrison> I think Image4J or the Fiji project also have dicom viewers
20131208 19:24:24 <hharrison> It was because of his bugreport I issued -pre9
20131208 19:25:22 <Eclesia> sgothel: pure java first. I only had a quick look at OpenCL, and if I fear I will have same creepy exception I had with opengl at the beginning, right now I don't have the motivation to do it in opencl.
20131208 19:26:11 <Eclesia> sgothel: webp will wait until I do WebM. that's far away ^^
20131208 19:29:11 <sgothel> @Harvey .. time to setup the Java3D web page and link all the great stuff :)
20131208 19:29:45 <sgothel> (or we do a sub-category on the jogl www page, but IMHO misleading)
20131208 19:42:48 * xranby-64 (~familjen@anon) has joined #jogamp
20131208 19:45:05 <hharrison> Yeah, definitely time
20131208 19:57:19 <Eclesia> bye all ++
20131208 19:57:30 * Eclesia (~eclesia@anon) Quit (Quit: Leaving.)
20131208 20:29:53 <xranby-64> http://icedtea.classpath.org/bugzilla/show_bug.cgi?id=1592 icedtea-web now support mixing signed and unsigned jars
20131208 20:31:03 <xranby-64> hence it will be easier for people to mix unsigned applets with the signed jogamp jars using icedtea-web 1.5!
20131208 20:52:56 <sgothel> hurray .. so the jogl-demos will work again
20131208 21:11:30 <xranby-64> sgothel: yes the jogl-demos now work again
20131208 21:11:40 * odin_ (~Odin@anon) Quit (Ping timeout: 260 seconds)
20131208 21:12:24 <xranby-64> http://jogamp.org/deployment/jogamp-current/jogl-demos/jogl-newt-applet-runner-angelesgl2es1.html runs using the new icedtea-web 1.5
20131208 21:15:35 <xranby-64> the jnlp based jogl-demos already work with icedtea-web 1.4
20131208 21:15:47 <xranby-64> using webstart
20131208 21:15:57 <xranby-64> now applets work as well
20131208 21:25:23 <xranby-64> hmm the city backdrop of the jogl-demos Vertex Program Refract is rendered upside down http://jogamp.org/deployment/jogamp-current/jogl-demos/VertexProgRefract.jnlp
20131208 21:25:46 <xranby-64> possible caused by one of the last years image loader optimizations?
20131208 21:42:39 <xranby-64> sgothel: whoha.. i managed to produce a native crash using the "Water" jogl-demo part of JRefract
20131208 21:42:46 <xranby-64> steps to reproduce
20131208 21:42:50 <xranby-64> 1. /usr/local/bin/javaws http://jogamp.org/deployment/jogamp-next/jogl-demos/JRefractNoOGL.jnlp
20131208 21:43:19 <xranby-64> 2. Actions -> Water
20131208 21:44:29 <xranby-64> (on my hardware it open up a window and display a blue screen (something not working with the mesa drivers, kernel emits "radeon: The kernel rejected CS, see dmesg for more information.")
20131208 21:44:43 <xranby-64> 3. press the swing subwindow close button -> crash
20131208 21:47:42 <xranby-64> logs: https://gist.github.com/xranby/347087345b46a02691dd
20131208 21:52:04 <hharrison> xranby: I as well have the radeon cs problem
20131208 21:53:51 <xranby-64> i am not able to reproduce the crash using the same demo in standalone form javaws http://jogamp.org/deployment/jogamp-current/jogl-demos/ProceduralTexturePhysics.jnlp prints the same radeon cs warnings and displays a blue window.. press close button -> no crash
20131208 21:55:14 <xranby-64> thus the crash may be GLJPanel related
20131208 21:55:59 <xranby-64> the JRefractNoOGL.jnlp is using a GLJPanel that i am able to reproduce the crash with
20131208 21:56:03 <hharrison> I've seen it a lot of reports when people are using the KWin window manager
20131208 21:56:57 <hharrison> Mine crashes with even simple calls to glBindTexture
20131208 21:57:25 <xranby-64> ok. i should test with mesa 10 that got released this week
20131208 21:57:51 <xranby-64> and see if that fixes any of these mesa gallium amd issues
20131208 22:23:14 * hharrison (~chatzilla@anon) Quit (Quit: ChatZilla 0.9.90.1 [Firefox 25.0/20131030081700])
20131208 22:31:14 <rmk0> hope the latest mesa sorts out their ES2/3 contexts
20131208 22:31:23 <rmk0> might hassle them about it
20131208 22:31:44 <rmk0> i see what you (anyone) means about them returning an ES3 context when an ES2 is requested
20131208 22:32:04 <rmk0> even worse, their ES3 context wasn't anything like correct/compatible last time i checked
20131208 22:32:08 <rmk0> no GLSL ES 300
20131208 22:33:35 * hharrison (~chatzilla@anon) has joined #jogamp
20131208 22:49:01 <xranby-64> http://www.mesa3d.org/relnotes/10.0.html - mesa 10 adds OpenGL 3.3 API
20131208 22:49:24 <xranby-64> Mesa 9.2.x series only implement OpenGL 3.1
20131208 22:50:52 <xranby-64> oh this is cool http://www.mesa3d.org/relnotes/9.2.4.html mesa 9.2.4 adds opensource adreno drivers
20131208 22:51:03 <xranby-64> based on the freedreno
20131208 22:51:06 <xranby-64> project
20131208 22:51:43 <xranby-64> this means that my sharp pcz1 from 2008 may finally get an opengl es driver :)
20131208 22:52:54 <xranby-64> 2009
20131208 22:53:19 <xranby-64> the sharp pcz1 did not ship with any opengl driver
20131208 22:53:27 <xranby-64> while the chip had support for it
20131208 22:53:47 * void256 (~void@anon) has joined #jogamp
20131208 22:59:24 * monsieur_max (~maxime@anon) has left #jogamp
20131209 00:21:29 * void256 (~void@anon) Quit (Quit: ChatZilla 0.9.90.1 [Firefox 25.0.1/20131112160018])
20131209 01:34:35 * hija (~hija@anon) Quit (Quit: hija)
20131209 02:50:57 * hija (~hija@anon) has joined #jogamp
20131209 05:05:56 -jogamp- Continue @ http://jogamp.org/log/irc/jogamp_20131209050556.html