#jogamp @ irc.freenode.net - 20141015 05:06:23 (UTC)

20141015 05:06:23 -jogamp- Previous @ http://jogamp.org/log/irc/jogamp_20141014025423.html
20141015 05:06:23 -jogamp- This channel is logged @ http://jogamp.org/log/irc/jogamp_20141015050623.html
20141015 05:23:21 <sgothel> *thanks for the best wishes*
20141015 05:23:55 <sgothel> regarding visibility and publicity - well, JOGL/JogAmp is still used in universities, and major science projects/products
20141015 05:24:11 <sgothel> so we indeed are not short of public attention
20141015 05:24:57 <sgothel> then .. I 'only' use email for efficiency .. and for having all messages available, so I don't care about a 'sexy web frontend'
20141015 05:25:06 <sgothel> many major projects simply use mailman
20141015 05:25:34 <sgothel> IMHO it is more important to have all modules run stable and very well as advertised
20141015 05:26:00 <sgothel> so JOGL should be good w/ 2.2.4 (major bugs and regressions due to OS updates are addressed)
20141015 05:26:10 <sgothel> JOCL would need some _major_ love
20141015 05:26:39 <sgothel> the right bug entries (questions) are addressed for JOCL (modularization, CL profiles ..)
20141015 05:27:00 <sgothel> Wade and I have to find time to do the work .. others are ofc welcome!
20141015 05:27:53 <sgothel> Currently I am cleaning our EGL layer to prepare it working well with desktop OpenGL, 'surface-less context', and later for Wayland
20141015 05:28:12 <sgothel> that should push usability of JOGL a bit further
20141015 05:29:59 <sgothel> @zuzub: I am a grandfather already, and this is Qun's and my 1st child, a son, who already is the 'little uncle' :)
20141015 05:31:04 <sgothel> @Julien: Yes, I also don't consider Mir .. but I won't give it bad wishes .. just an answer to no question motivated by company politics IMHO
20141015 05:31:52 <sgothel> And Wayland is also no real replacement for heavy X11 usage .. but a lightweight alternative IMHO
20141015 06:09:58 * hija (~hija@anon) has joined #jogamp
20141015 06:14:49 * hija (~hija@anon) Quit (Ping timeout: 260 seconds)
20141015 06:42:47 * monsieur_max (~maxime@anon) has joined #jogamp
20141015 06:57:52 * hija (~hija@anon) has joined #jogamp
20141015 07:02:22 * hija (~hija@anon) Quit (Ping timeout: 240 seconds)
20141015 07:13:46 * jvanek (jvanek@anon) has joined #jogamp
20141015 07:15:06 * hija (~hija@anon) has joined #jogamp
20141015 07:30:59 * eclesia (~husky@anon) has joined #jogamp
20141015 07:31:07 <eclesia> good morning
20141015 07:34:36 <zubzub> 07:31 <+sgothel> And Wayland is also no real replacement for heavy X11 usage .. but a lightweight alternative IMHO
20141015 07:34:40 <zubzub> it is actually
20141015 07:34:44 <zubzub> but not quite yet ;)
20141015 07:34:54 <zubzub> the xdg_shell protocol needs to mature first
20141015 07:35:01 <zubzub> which will take several years
20141015 07:35:18 <zubzub> given that all attention was on the mobile space now
20141015 07:39:03 * hija (~hija@anon) Quit (Quit: hija)
20141015 07:43:35 <zubzub> sgothel: does jogl have egl bindings?
20141015 07:44:31 <zubzub> as in getEglDisplay
20141015 07:45:55 * doev (~doev@anon) has joined #jogamp
20141015 07:51:51 * hija (~hija@anon) has joined #jogamp
20141015 07:56:40 * doev2 (~doev@anon) has joined #jogamp
20141015 07:57:52 * doev (~doev@anon) Quit (Ping timeout: 240 seconds)
20141015 08:05:51 * hija (~hija@anon) Quit (Quit: hija)
20141015 08:23:49 <zubzub> basically the functions defined here:
20141015 08:23:51 <zubzub> https://www.khronos.org/files/egl-1-4-quick-reference-card.pdf
20141015 08:29:00 * gouessej (5ee4b442@anon) has joined #jogamp
20141015 08:30:39 <gouessej> Hi
20141015 08:30:51 <gouessej> monsieur_max: WebGL isn't the panacea
20141015 08:31:09 <eclesia> gouessej: I don't think that's a valid english word :D
20141015 08:31:39 <gouessej> monsieur_max: Mozilla's decision just causes planned obsolescence
20141015 08:31:43 <gouessej> eclesia: https://translate.google.com/#fr/en/panac%C3%A9e
20141015 08:32:05 <gouessej> eclesia: http://www.wordreference.com/fren/panac%C3%A9e
20141015 08:32:36 <gouessej> eclesia: http://www.larousse.fr/dictionnaires/francais-anglais/panac%C3%A9e/57201
20141015 08:32:50 <gouessej> eclesia: I think it is enough for now :)
20141015 08:33:12 * eclesia was wrong, looks strange anyway
20141015 08:34:07 <monsieur_max> gouessej: planned obsolescence is not a real problem in web technologies where everything changes totally every 5 years, my point of view :)
20141015 08:34:12 <gouessej> I started learning English at almost 6 years ol but thank you for trying to fix my English anyway, I'm not foolproof
20141015 08:35:44 <gouessej> monsieur_max: Most WebGL implementations use offscreen rendering and work a lot worse and less reliably than JOGL on a lot of machines
20141015 08:36:33 <gouessej> monsieur_max: If I need a super powerful machine just to play with a clone of Doom, I will consider that WebGL drives my machine artificially obsolete
20141015 08:37:12 <monsieur_max> well, yet people seems to be able to make games with it, and it still better than anything else on browsers
20141015 08:38:10 <gouessej> monsieur_max: WebGL is far better with applets of course but no, there are tons of laptops and even powerful desktop computers that get poor performance with it even now
20141015 08:38:29 <zubzub> but web is the future!
20141015 08:38:39 <gouessej> monsieur_max: Even OpenJSCAD is slow on some machines with an NVIDIA Quadro
20141015 08:38:43 <zubzub> *write local running mobile app*
20141015 08:38:44 <monsieur_max> i'm not comparing it to jogl in anyway , i'm pro-desktop, not pro-web , but i still think that webgl is a great step forward
20141015 08:39:36 <gouessej> monsieur_max: Yes but it's slower than JOGL and there are still lots of performance problems on laptops with Intel chips and tons of drivers are still blacklisted
20141015 08:40:56 <gouessej> monsieur_max: As long as I can't play with Doom on a pentium MMX with WebGL, I will consider it encourages planned obsolescence
20141015 08:41:12 <monsieur_max> ok ...
20141015 08:43:13 <gouessej> monsieur_max: Would you buy a supercomputer just to perform a few additions mostly because it is trendy? WebGL is very well integrated, things have been improved, it is a major technology and it is enough when you don't care about tons of machines with no extremely good support of offscreen rendering
20141015 08:43:48 <gouessej> monsieur_max: but I'm fed up with the mantra "HTML5 or die"
20141015 08:44:43 <bbbruce> maybe this particular implementation of Doom in WebGL is just not very good?
20141015 08:45:01 <gouessej> no it uses Emscripten
20141015 08:45:13 <gouessej> it's based on PRBoom
20141015 08:45:34 <gouessej> and anyway I have the same problem of performance with other WebGL games and applications too
20141015 08:45:47 <gouessej> the port of Cube 2 too
20141015 08:45:57 <gouessej> less than 5 FPS lol
20141015 08:46:11 <gouessej> even without shaders, bump mapping
20141015 08:46:14 <gouessej> ....
20141015 08:46:57 <gouessej> Most of what I wrote is still true now: http://gouessej.wordpress.com/2012/06/22/webgl-une-technologie-prometteuse-qui-doit-encore-faire-ses-preuves-webgl-a-promising-technology-that-has-yet-to-prove-itself/
20141015 08:48:34 <gouessej> Did Ars Technica claim that Notch was crazy for coding Minecraft in Java? Now it promotes WebGL lol
20141015 08:52:41 * eclesia consider obsolete OpenGL if it's not at least 3.3
20141015 08:53:15 <bbbruce> eclesia: that's what I do
20141015 08:56:28 * hija (~hija@anon) has joined #jogamp
20141015 09:00:44 <gouessej> I consider obsolete OpenGL it isn't supported by JOGL, OpenGL 1.0?
20141015 09:00:52 <bbbruce> why bother?
20141015 09:01:42 <gouessej> Just try to explain to an end user that he needs a super modern machine to play with a game that looks like it had been released in 1997
20141015 09:03:01 <bbbruce> practically all nvidia, amd and most intel hd stuff made in the last 4 years supports gl 3.3
20141015 09:03:38 <bbbruce> on windows and osx at least
20141015 09:40:18 <gouessej> There are still people under Windows XP but it shouldn't prevent you from playing with the latest OpenGL features especially if you only want to support machines sold since 2010
20141015 09:40:39 * gouessej (5ee4b442@anon) Quit (Quit: Page closed)
20141015 11:12:34 * monsieur_max1 (~maxime@anon) has joined #jogamp
20141015 11:16:30 * monsieur_max1 (~maxime@anon) Quit (Client Quit)
20141015 11:35:47 <eclesia> rmk0: hi, may I ask a question on shadow mapping (http://http.developer.nvidia.com/GPUGems3/gpugems3_ch08.html) , I have converted the direcx code in glsl http://pastebin.com/aiVLBbzG . To be sure I'm doing it right :
20141015 11:36:15 <eclesia> 1 - when building the shadow map : use the computeMoments function to calculate values
20141015 11:36:23 <eclesia> 2 - blur the shadowmap
20141015 11:36:58 <eclesia> 3 - use function shadowContribution in the scene rendering fragement shader
20141015 11:37:03 <eclesia> is that right ?
20141015 11:38:13 * xranby (~xranby@anon) Quit (Ping timeout: 260 seconds)
20141015 11:53:02 * xranby (~xranby@anon) has joined #jogamp
20141015 11:56:06 * gouessej (5ee4b442@anon) has joined #jogamp
20141015 11:56:21 <gouessej> Does one of us have a reddit account?
20141015 11:57:18 <gouessej> This guy seems to have misread my tutorial: http://www.reddit.com/r/IntelliJIDEA/comments/2jaxmd/how_to_get_java3d_to_work_in_intellij_on_os_x/
20141015 11:58:07 <gouessej> I clearly indicated in the "troubleshooting" section what he has to do in this case: http://gouessej.wordpress.com/2012/08/01/java-3d-est-de-retour-java-3d-is-back/#troubleshooting
20141015 11:58:39 <gouessej> He has forgotten to uninstall Java3D
20141015 11:58:48 <zubzub> create an account
20141015 11:58:52 <zubzub> answer
20141015 11:59:51 <gouessej> I prefer avoiding to create an account on reddit, I'm already fed up answering on StackOverflow. I asked people to use our forum in the tutorial :s
20141015 12:00:15 <eclesia> so don't answer, let him find the solution on his own
20141015 12:07:35 <gouessej> but maybe one of you could contact him on reddit if you already have an account
20141015 12:10:58 <zubzub> I dont have an account
20141015 12:11:22 <zubzub> also when people don't read the doc, I usually don't bother to answer
20141015 12:18:26 <gouessej> I've just updated this section with his error message
20141015 12:40:47 * gouessej (5ee4b442@anon) Quit (Quit: Page closed)
20141015 12:49:32 * xranby (~xranby@anon) Quit (Quit: Leaving.)
20141015 13:14:17 * xranby (~xranby@anon) has joined #jogamp
20141015 13:15:10 <rmk0> .o.
20141015 13:15:14 * rmk0 prepares forum post
20141015 13:15:18 <rmk0> didn't get a chance yesterday
20141015 13:15:48 <rmk0> eclesia: nope, you use ComputeMoments() during the actual application of the shadows. when you're creating the shadow map, just store (depth, depth * depth)
20141015 13:16:21 <eclesia> rmk0: ok, but when ?
20141015 13:16:34 <rmk0> when?
20141015 13:16:38 <rmk0> when when?
20141015 13:17:02 <eclesia> when is used computeMoment ? the other methods doesnt seem to use it
20141015 13:17:36 <rmk0> ... let me check what i did
20141015 13:17:41 <rmk0> i've not looked at this in detail for months
20141015 13:29:26 <rmk0> ah, looks like ComputeMoments is optional, and i found i didn't need it
20141015 13:29:40 <rmk0> they mention that it uses partial derivatives and that they may be unstable on some hardware
20141015 13:30:41 <rmk0> i just clamp the minimum variance as they suggest
20141015 13:31:06 <eclesia> hm... I guess I'm still doing something wrong then. :/
20141015 13:31:46 <rmk0> take it step by step... when you're building the shadow map, you only need to render (depth, depth * depth) into an RG texture
20141015 13:32:01 <eclesia> already done and checked
20141015 13:32:04 <rmk0> right
20141015 13:33:01 <rmk0> during rendering of a shadow-projecting light, you obviously want to calculate a shadow factor so that you can attentuate the light
20141015 13:33:09 <rmk0> where 0.0 means "completely in shadow" and 1.0 means "not in shadow at all"
20141015 13:34:18 <rmk0> http://waste.io7m.com/2014/10/15/factor.txt
20141015 13:35:01 <eclesia> that's right, before using this approach I had shadow depth as float and the fragment depth in light space, as simple < was working as expected
20141015 13:35:04 <rmk0> there i'm passing in a clip-space position p (the surface being shaded), and a shadow map t_shadow_variance
20141015 13:37:49 <eclesia> I'll check this after work, lunch break is finished ^^
20141015 13:37:54 <rmk0> hehe
20141015 13:38:34 <rmk0> in my experience, the problem (if you're having stuff just not work) is almost always down to a mismatch between coordinate systems
20141015 13:38:47 <rmk0> like having post-projective depth in one value but eye-space depth in another, and so on
20141015 13:39:20 <rmk0> we badly need shading languages that make values in different coordinate systems type-incompatible
20141015 13:52:11 <eclesia> not sure that would be possible
20141015 13:58:30 * doev2 (~doev@anon) Quit (Quit: Verlassend)
20141015 13:59:18 <rmk0> i already know how to implement it
20141015 13:59:24 <rmk0> just don't have the time to write yet another compiler
20141015 13:59:32 <eclesia> lol
20141015 13:59:52 <eclesia> are you sure you don't want to rewrite the world like I do ?
20141015 14:00:07 <rmk0> only the ill-typed bits! \o/
20141015 14:01:31 <rmk0> tracking that sort of information is pretty easy in modern type systems
20141015 14:01:45 <rmk0> the main problem is that most people's idea of a type system is basically C
20141015 14:05:37 <zubzub> what's a type system?
20141015 14:06:44 <rmk0> the rules in a language that state which expressions/functions/statements are "valid" and which aren't
20141015 14:07:42 <rmk0> http://mvn.io7m.com/io7m-jparasol/specification/p1s7.xhtml#st200_p1s7
20141015 14:07:44 <rmk0> etc
20141015 14:17:00 <rmk0> could have, like:
20141015 14:17:08 <rmk0> type vector3f [c] = ...
20141015 14:17:16 <rmk0> type vector4f [c] = ...
20141015 14:17:24 <rmk0> type matrix_4x4f [c, d] = ...
20141015 14:17:53 <rmk0> mult4 : forall c d, matrix_4x4f [c, d] → vector_4f [c] → vector_4f [d]
20141015 14:18:24 <rmk0> that's a function that takes a matrix that represents a transform from coordinate system c to coordinate system d, and takes a vector in coordinate system c, and produces a vector in coordinate system d as a result
20141015 14:18:48 <rmk0> languages like haskell, ml, etc, can infer all of that automatically
20141015 14:19:03 <rmk0> let v_eye = mult4 view_matrix v_world
20141015 14:19:15 <rmk0> v_eye : vector_4f [eye_space]
20141015 14:19:16 <rmk0> etc
20141015 14:19:20 <eclesia> nice
20141015 14:19:31 <rmk0> i wanted to have this stuff in parasol, but i didn't have enough time to formalize it all
20141015 14:19:37 <rmk0> will definitely do it for whatever replaces parasol
20141015 14:20:02 <eclesia> parasol is your tool ?
20141015 14:20:06 <rmk0> yeah
20141015 14:20:10 <eclesia> :o
20141015 14:20:14 <rmk0> got ... frustrated ... with glsl very early on
20141015 14:22:34 <zubzub> parasol?
20141015 14:23:02 <rmk0> http://mvn.io7m.com/io7m-jparasol
20141015 14:25:41 <rmk0> http://mvn.io7m.com/io7m-jparasol/documentation/p2s2.xhtml#st200_p2s2
20141015 14:25:51 <rmk0> "i hate this language"
20141015 14:26:11 <rmk0> that's also out of date... i think there are 17 versions now
20141015 14:26:33 <zubzub> yeah that was the first thing that struck me
20141015 14:26:38 <zubzub> when looking into glsl
20141015 14:40:04 <rmk0> there are some concessions made to ES2
20141015 14:40:14 <rmk0> like... when i started work on parasol, ES3 didn't even exist
20141015 14:40:25 <rmk0> so i avoided putting things into the language that couldn't possibly work on ES2
20141015 14:40:51 <rmk0> there is some stuff in there now that won't (multiple fragment shader outputs, etc)
20141015 14:41:14 <rmk0> but in a couple of years when ES2 is dead and gone, a language that replaces parasol will be quite different
20141015 14:41:31 <rmk0> support for uniform buffer objects is high on the list
20141015 15:01:25 * jvanek (jvanek@anon) Quit (Quit: Leaving)
20141015 15:04:38 * eclesia going home and testing shadows ^^
20141015 15:04:41 * eclesia (~husky@anon) has left #jogamp
20141015 15:06:45 * gouessej (5ee4b442@anon) has joined #jogamp
20141015 15:07:52 <gouessej> The FUD campaign goes on ... on Reddit: http://www.reddit.com/r/gamedev/comments/149fi5/help_needed_with_shooting_game_in_java_jogl/c7b4krf
20141015 15:08:51 <gouessej> "because of the larger selection of game-specific platforms available for it"
20141015 15:22:40 * doev (~doev@anon) has joined #jogamp
20141015 15:24:21 <rmk0> http://forum.jogamp.org/Forum-improvement-td4033326.html#a4033363
20141015 15:25:38 <gouessej> rmk0: I'm going to read it...
20141015 15:28:38 <gouessej> rmk0: Great conclusion
20141015 15:29:08 <gouessej> rmk0: When I read that, I see the difference between a native speaker and me
20141015 15:29:11 <gouessej> :s
20141015 15:29:36 <rmk0> hehe
20141015 15:33:01 <rmk0> manually processing the nabble database is a lot of work
20141015 15:33:54 <rmk0> and i don't think the result would preserve any of the URIs for any of the different forum packages
20141015 15:34:07 <rmk0> is... just not something anyone seems to value
20141015 15:34:45 <gouessej> I want to preserve the URIs
20141015 15:34:58 <gouessej> this is what I said since the very beginning
20141015 15:35:02 <rmk0> urhur
20141015 15:35:14 <gouessej> I wrote that I don't want to break the permanent links
20141015 15:35:19 <gouessej> sorry, it wasn't clear
20141015 15:35:24 <rmk0> no, it was clear
20141015 15:35:32 <gouessej> I often post links to our forum on StackOverflow
20141015 15:35:43 <rmk0> i was saying that none of the other forum packages seem to place value on preserving link integrity
20141015 15:35:58 <gouessej> Sorry, it's my fault
20141015 15:36:04 <gouessej> I should sleep a bit more
20141015 15:36:11 <rmk0> that seems to be something that's dying out on the internet in general... almost nothing survives more than a few years
20141015 15:36:47 <gouessej> Personally, I like games that will still work in several decades and links too :)
20141015 15:36:53 <rmk0> yeah
20141015 15:37:39 <gouessej> I don't want to register on Reddit :(
20141015 15:38:22 <gouessej> Xerxes is on Reddit!
20141015 15:38:34 <gouessej> xranby: http://www.reddit.com/r/IntelliJIDEA/comments/2jaxmd/how_to_get_java3d_to_work_in_intellij_on_os_x/
20141015 15:40:07 <gouessej> xranby: Please tell him to ask Santa Claus to bring him a new brain with tons of functioning neurons so that he can read the f***ing "troubleshooting" section: http://gouessej.wordpress.com/2012/08/01/java-3d-est-de-retour-java-3d-is-back/#troubleshooting
20141015 15:40:34 <gouessej> I should calm down
20141015 15:40:53 <rmk0> hehe
20141015 15:41:05 <gouessej> I won't go into Zahand's rocket
20141015 15:41:09 <gouessej> I don't want to die
20141015 15:42:45 * gouessej (5ee4b442@anon) Quit (Quit: Page closed)
20141015 16:07:52 * doev (~doev@anon) Quit (Ping timeout: 240 seconds)
20141015 16:14:13 * Eclesia (~eclesia@anon) has joined #jogamp
20141015 16:24:15 * jvanek (jvanek@anon) has joined #jogamp
20141015 16:40:35 * monsieur_max (~maxime@anon) Quit (Quit: Leaving.)
20141015 17:18:09 * monsieur_max (~maxime@anon) has joined #jogamp
20141015 17:20:10 * hija (~hija@anon) Quit (Quit: hija)
20141015 17:21:11 <Eclesia> rmk0: yeah, finally it starts to work :)
20141015 17:22:39 <Eclesia> theoric question : in tutorials about shadow, the use the z value of the fragment position in light space, shouldn't it be the length or xyz ?
20141015 17:22:45 <Eclesia> length of*
20141015 17:26:29 * jvanek (jvanek@anon) Quit (Quit: Leaving)
20141015 17:50:55 * monsieur_max (~maxime@anon) Quit (Quit: Leaving.)
20141015 18:26:40 * monsieur_max (~maxime@anon) has joined #jogamp
20141015 18:37:17 * Eclesia (~eclesia@anon) Quit (Ping timeout: 260 seconds)
20141015 18:40:02 * Eclesia (~eclesia@anon) has joined #jogamp
20141015 18:40:26 * Eclesia is getting better, crashes the gpu more and more often ^^
20141015 20:03:23 * hija (~hija@anon) has joined #jogamp
20141015 20:36:14 * Eclesia (~eclesia@anon) Quit (Quit: Leaving.)
20141015 21:12:46 <sgothel> @Mark: thx for the summary (forum)
20141015 21:29:10 <rmk0> sgothel: \o/ i just replied again... i'm not sure where you got the idea that an import actually worked
20141015 21:29:30 <rmk0> i thought the irc log implied the exact opposite, unless you've read something i've not and i have a gap in my memory
20141015 21:29:33 <rmk0> which... isn't unlikely
20141015 21:30:03 <rmk0> i seem to remember it being "at best, we can import data, with a lot of painful labour, and it won't preserve URIs at the end"
20141015 21:30:42 <rmk0> if Elect wants to try it... he/she's welcome to try
20141015 21:53:45 * monsieur_max (~maxime@anon) Quit (Quit: Leaving.)
20141015 22:31:37 <sgothel> yeah .. sorry, 'started importing' == looked at the sql dump :)
20141015 22:48:51 <rmk0> hehe
20141016 05:06:23 -jogamp- Continue @ http://jogamp.org/log/irc/jogamp_20141016050623.html