#jogamp @ irc.freenode.net - 20140316 05:06:09 (UTC)


20140316 05:06:09 -jogamp- Previous @ http://jogamp.org/log/irc/jogamp_20140315050609.html
20140316 05:06:09 -jogamp- This channel is logged @ http://jogamp.org/log/irc/jogamp_20140316050609.html
20140316 07:51:04 * monsieur_max (~maxime@anon) has joined #jogamp
20140316 09:46:05 * [Mike] (~Mike]@anon) Quit ()
20140316 09:52:58 * hija (~hija@anon) has joined #jogamp
20140316 13:03:14 * hija_ (~hija@anon) has joined #jogamp
20140316 13:04:43 * hija (~hija@anon) Quit (Ping timeout: 264 seconds)
20140316 13:04:43 * hija_ is now known as hija
20140316 13:09:18 * hija_ (~hija@anon) has joined #jogamp
20140316 13:10:38 * hija (~hija@anon) Quit (Ping timeout: 240 seconds)
20140316 13:10:38 * hija_ is now known as hija
20140316 15:45:51 * bbbruce (~bx@anon) Quit (Read error: Connection reset by peer)
20140316 15:46:04 * bbbruce (~bx@anon) has joined #jogamp
20140316 18:21:40 * hija (~hija@anon) Quit (Quit: hija)
20140316 18:58:46 * hija (~hija@anon) has joined #jogamp
20140316 20:38:25 <sgothel> 2.2-train -> master .. welcome ..
20140316 20:54:21 <sgothel> http://forum.jogamp.org/JOGL-Post-2-1-5-2-2-train-gt-master-API-Changes-and-the-like-td4031927.html
20140316 20:54:34 <sgothel> .. and I got my little math sessions :)
20140316 20:54:54 <sgothel> zero or not zero .. that is the question :)
20140316 20:56:18 <sgothel> @Mark / @All: So there is this notion .. of preferring all objects to be organized spatial by some sorting algo (e.g. Octree) instead of going the old route and using a DAG scenegraph ? Or can both (presentations) live together in harmony ?
20140316 21:03:33 <rmk0> sgothel: not sure... for this renderer, i insert static (ie, not moving) objects into an octree once, and then (re)-insert moving objects every frame. when it comes to rendering, i take whatever's currently visible from the octree and batch objects by material (divided into opaque and translucent batches) to reduce the number of state changes
20140316 21:03:51 <rmk0> so it's not purely one or the other
20140316 21:04:20 <rmk0> all the opaques get rendered first in an arbitrary order (using the depth buffer to control visibility), and then the translucents get rendered back to front
20140316 21:04:27 <sgothel> so .. an octree queue for rendering ..
20140316 21:04:38 <sgothel> DAG -> OctQueue -> GL
20140316 21:04:43 <rmk0> an octree for visibility, no queue
20140316 21:04:59 <sgothel> I mean .. like a processing path ..
20140316 21:05:15 <sgothel> *pass
20140316 21:05:25 <rmk0> well the path is basically Octree -> Material batching -> GL
20140316 21:06:08 <sgothel> me looking for adequate structure for UI ..
20140316 21:06:49 * void256 (~void@anon) has joined #jogamp
20140316 21:06:55 <sgothel> i.e. should not be too big ..
20140316 21:07:19 <sgothel> I guess it's time to read NiftyGUI .. good that he just showed up :)
20140316 21:08:33 <sgothel> Ofc .. the UI structure shall be small .. since it is intended to be integrated in a users graph
20140316 21:08:55 <sgothel> on the other hand .. it should be efficient and simple .. hmm
20140316 21:09:03 <sgothel> *convenient
20140316 21:10:16 <sgothel> @All: Help w/ rotation in the controller (of demo w/ UIShape) ..
20140316 21:10:23 <sgothel> ?
20140316 21:10:40 <sgothel> http://jogamp.org/git/?p=jogl.git;a=commit;h=5e728baa72517865d602580b920d9bdfdfb26c65
20140316 21:11:01 <sgothel> http://jogamp.org/git/?p=jogl.git;a=blob;f=src/test/com/jogamp/opengl/test/junit/graph/demos/ui/SceneUIController.java;h=fbdc72a1fe83f381d082ee4177ee4136628a8d07;hb=5e728baa72517865d602580b920d9bdfdfb26c65#l127
20140316 21:11:11 <sgothel> the rotation is very odd ..
20140316 21:11:28 <sgothel> I know .. one should rotate an object while it is at 0,0,0 ..
20140316 21:11:47 <sgothel> so here .. the origin is somewhere else .. and hence wrong
20140316 21:12:44 <sgothel> I tried the location with a centered shape at origin .. but when I moved it back to it's place after rotation - result was not so good
20140316 21:13:03 <sgothel> I guess some simple beginner mistake :)
20140316 21:15:31 <sgothel> @Mark ? @Jens ?
20140316 21:16:58 <sgothel> .. guess I have to continue my lessons on proper mat-ops ..
20140316 21:17:48 <sgothel> maybe it was looking bad due to the perspective .. and too close to nearZ
20140316 21:18:19 <rmk0> sgothel: feel like those are in the wrong order
20140316 21:19:00 <rmk0> i'm not familiar with the PMVMatrix interface, but i'd try doing the translation, then the scale, then the rotation
20140316 21:19:21 <rmk0> because they'll concatenate and effectively happen in the opposite order (if the interface works the way i think it does)
20140316 21:19:44 <sgothel> I guess I had that before .. must have been introdruced while playing .. oops
20140316 21:19:50 <sgothel> yes .. normal matrix
20140316 21:20:19 <sgothel> right .. the reverse order thingy .. oh oh
20140316 21:20:38 <rmk0> even after three years, i still make that mistake
20140316 21:20:57 <sgothel> so then we have:
20140316 21:20:58 * monsieur_max (~maxime@anon) Quit (Write error: Broken pipe)
20140316 21:21:51 <sgothel> P * Mv1 * Mv2 ( = T * S * R )
20140316 21:22:10 <sgothel> and since its done reverse .. rotate is from .. err .. origin
20140316 21:22:17 <rmk0> right
20140316 21:22:37 <sgothel> so I may make a translation 'before' (coding after) the rotate so it gets centered ?
20140316 21:22:46 <sgothel> i.e. rotate around it's center
20140316 21:22:53 <rmk0> yep
20140316 21:23:19 <sgothel> now it makes sense .. and the pop/pull of the matrices .. wasn't quite so clear after the 15th hour at work :)
20140316 21:23:28 <rmk0> assuming the ui elements aren't currently drawn from the center
20140316 21:23:37 <rmk0> which... i assume they aren't
20140316 21:23:46 <rmk0> is the origin for each element in one of the corners?
20140316 21:23:58 <sgothel> sadly yes .. and I guess that has to change
20140316 21:24:12 <rmk0> i don't think that's unusual for a ui library
20140316 21:24:23 <sgothel> it's for the OutlineShape in general
20140316 21:24:29 <sgothel> lower-left
20140316 21:24:35 <rmk0> right
20140316 21:24:42 <rmk0> i mean ... yes ... not "right"
20140316 21:25:03 <rmk0> *ahem*
20140316 21:25:26 <sgothel> I fixed the whole graph shape org .. so now you can translate an already triangulated shape (good for text .. and all that)
20140316 21:25:46 <sgothel> then the shape goes w/ 2d transform to the region -> VBO
20140316 21:26:01 <sgothel> (shape is 2d only)
20140316 21:26:25 <sgothel> (i.e. the curves .. OutlineShape that is)
20140316 21:26:40 <sgothel> we have like 1000fps for changing text right now :)
20140316 21:26:48 <rmk0> heh
20140316 21:26:56 <sgothel> was like 100fps before
20140316 21:27:49 <sgothel> just b/c the curve's tex-coords are unrelated to the vertices .. otherwise .. impossible
20140316 21:27:55 <sgothel> (detail of the math)
20140316 21:28:10 <sgothel> so .. wouldn't work w/ loop-blinn :)
20140316 21:28:41 <sgothel> great .. will try fixing rotation
20140316 21:29:02 <sgothel> it's actually fun todo graphics programming again .. after all the janitor work :)
20140316 21:29:10 <rmk0> can imagine
20140316 21:29:27 <rmk0> don't know how you can work on jogl so much without burning out
20140316 21:29:39 <rmk0> is a great library, but a lot of it isn't exactly "interesting" work
20140316 21:29:43 <rmk0> just fighting drivers forever
20140316 21:29:53 <sgothel> disciplin .. and my wife would beat me otherwise :)
20140316 21:30:03 <rmk0> hehe
20140316 21:30:08 <sgothel> ah .. I like the details of system programming
20140316 21:30:44 <sgothel> and making things platform agnostic .. sort of a freedom drive
20140316 21:30:49 <rmk0> yep
20140316 21:31:14 <sgothel> what still bugs me here .. is the copying of vectors .. grrr
20140316 21:31:53 <sgothel> guess .. then it's time for a UIShape w/ playing movies .. to select etc :)
20140316 21:34:20 <sgothel> re janitor work: when my wife did her phd .. it's all the same, they all work 3-4 years on the one subject .. not really too exciting all the time, sure not. So it's not just like to have the one idea here - but it must be executed - otherwise worthless.
20140316 21:34:29 <rmk0> yep
20140316 22:43:51 * magaio (~magaio@anon) Quit (*.net *.split)
20140316 22:44:45 * magaio (~magaio@anon) has joined #jogamp
20140316 22:50:29 * phao (~phao@anon) has joined #jogamp
20140316 23:12:09 <sgothel> kernel:[3555019.425507] Dazed and confused, but trying to continue
20140316 23:12:12 <sgothel> wow
20140316 23:12:45 * sgothel (~sgothel@anon) Quit (Quit: Leaving.)
20140316 23:17:10 * sgothel (~sgothel@anon) has joined #jogamp
20140316 23:17:10 * sgothel (~sgothel@anon) Quit (Changing host)
20140316 23:17:10 * sgothel (~sgothel@anon) has joined #jogamp
20140316 23:17:10 * ChanServ sets mode +v sgothel
20140317 00:07:33 * void256 (~void@anon) Quit (Quit: ChatZilla 0.9.90.1 [Firefox 27.0.1/20140212131424])
20140317 00:18:05 <sgothel> contemplating weather to create the sub-project for graph-ui ..
20140317 00:18:48 <sgothel> i.e. remove graph* from jogl, keeping jogl smaller .. and allow changing stuff faster in graph-ui
20140317 00:18:58 <sgothel> then the name .. jogl-graph -> jogg ? :)
20140317 00:19:39 <sgothel> anybody experience w/ svg-salamander ?
20140317 00:19:44 <sgothel> looks small
20140317 00:21:07 <sgothel> debian-linux jessie upgrade .. holy-moly .. alot, plus systemd finally ..
20140317 00:24:58 * phao (~phao@anon) Quit (Ping timeout: 240 seconds)
20140317 00:38:26 * sgothel (~sgothel@anon) Quit (Quit: Leaving.)
20140317 00:45:53 * sgothel (~sgothel@anon) has joined #jogamp
20140317 00:45:53 * sgothel (~sgothel@anon) Quit (Changing host)
20140317 00:45:53 * sgothel (~sgothel@anon) has joined #jogamp
20140317 00:45:53 * ChanServ sets mode +v sgothel
20140317 00:48:32 <sgothel> @Xerxes: nice helping w/ the audio thingy .. note MovieCube can do audio only as well afaik, i.e. via '-vid' option .. or was it SimpleMovie .. ?
20140317 00:49:21 <sgothel> .. btw .. debian-jessie machine is still working after dist-upgrade :)
20140317 00:56:16 * sgothel (~sgothel@anon) Quit (Read error: No route to host)
20140317 01:03:49 * magaio (~magaio@anon) Quit (Read error: Connection reset by peer)
20140317 01:04:08 * magaio (~magaio@anon) has joined #jogamp
20140317 01:15:25 * sgothel (~sgothel@anon) has joined #jogamp
20140317 01:15:25 * ChanServ sets mode +v sgothel
20140317 01:17:45 <sgothel> 'excellent' NV driver freezes kernel when req. odd FBO
20140317 01:17:53 <sgothel> (latest NV driver)
20140317 01:18:24 <sgothel> amd simply says: GL Error 0x505 while creating RenderAttachment[type DEPTH, format 0x81a6, samples 0, 16001x6400, name 0x0, obj 0x716d9094]
20140317 01:18:24 <sgothel> at com.jogamp.opengl.FBObject$RenderAttachment.initialize(FBObject.java:398)
20140317 01:18:55 <sgothel> in short: pick a wrong matrix -> NV linux kernel crash .. OMG
20140317 01:19:19 <sgothel> NVIDIA-Linux-x86_64-334.21.run <- the one
20140317 01:19:43 <sgothel> NVIDIA-Linux-x86_64-331.38.run <- didn't freeze but produces these weird 'confused' kernel messages
20140317 01:25:27 <sgothel> glGetError() returned the following error codes after a call to glRenderbufferStorage(<int> 0x8D41, <int> 0x81A6, <int> 0x1A79, <int> 0xCA0): GL_OUT_OF_MEMORY ( 1285 0x505),
20140317 01:25:28 <sgothel> at javax.media.opengl.DebugGL4bc.writeGLError(DebugGL4bc.java:29501)
20140317 01:25:28 <sgothel> at javax.media.opengl.DebugGL4bc.glRenderbufferStorage(DebugGL4bc.java:18012)
20140317 01:25:52 <sgothel> glRenderbufferStorage failed because there was not enough available memory to complete the operation (GL_OUT_OF_MEMORY)
20140317 01:25:58 <sgothel> AMD catalyst ++ :)
20140317 01:38:32 <sgothel> Q: How to properly test against such out of memory ? I could not find a way .. yes, sure, we do check GL_MAX_RENDERBUFFER_SIZE etc ..
20140317 05:04:08 * hija (~hija@anon) Quit (Quit: hija)
20140317 05:06:10 -jogamp- Continue @ http://jogamp.org/log/irc/jogamp_20140317050610.html