GEM-dev December 2003

gem-dev@lists.iem.at

9 participants
20 discussions

pix_texture2 does not respect upsidedown flag
by Henning Haaland Kulander 04 Dec '03

04 Dec '03

Hi! While working on the filmFFMPEG.cpp file I have had problems getting the image to be displayed the correct way. I tried playing with the upsidedown flag in imageStruct, but it wouldn't work. I was using 05.film.pd for testing. When replacing pix_texture2 with pix_texture, I finally got it working. The problem with pix_texture2 is that the flipping is hardcoded using #ifdefs checking for the apple platform. This #ifdef is also done in line 108 in GemPixUtil.h, and will set upsidedown on all imageStructs on apple. I created a patch for pix_texture2 which does the test in runtime as it is done in pix_texture. This fixes my problem, but can possibly cause other problems.. :) Is there a reason for hardcoding this? Regards, Henning Kulander

2 1

curve has extra inlet?
by B. Bogart 04 Dec '03

04 Dec '03

Hey all, I noticed today that the [curve] has an extra inlet. [curve 5] makes an object with 7 inlets. I'm still using Cyrille's version of Gem, but noticed the problem opening the same patch in 87 and CVS. This may have already been fixed? I still have gotten no replies about what version of FTGL and/or freetype2 I need for gem under linux. Thanks Ben

4 4

releasing today ?
by IOhannes zmoelnig 03 Dec '03

03 Dec '03

hi all having worked a bit on the video/film stuff under windows, i hope it works now. however i made a change to pix_filmDarwin too (now that i think of it, i am not really sure anymore, why i did it), and so i need another test-run on all the platforms it works fine under linux (at least on my system) the pix_film stuff should now work again under windows (it turned out to be a problem with the auto-registering of classes: [pix_film] was registered before (say) [pix_filmNT], which didn't do anything, so nothing was drawn... known problems: pix_filmQT crashes, and pix_filmNEW (which is default under windows) crashes even more often under certain circumstances (it is: when i load a quicktime-movie and afterwards an AVI everything is fine. when i load an avi, loading quicktimes afterwards crashes (but loading avis is still fine) [pix_kaleidoscope] crashes with YUV (but i don't have the energy to fix this), both under linux and windows. i have changed some bits in pix_videoDS (namely: the return of the Callback() is now set to NOERROR instead of S_FALSE. the old behaviour gave me exactly one frame, and stopped capturing afterwards.) i have added dummy "enumerate" and "dialog" to the [pix_video] class (for all systems). anything else ? ah, i have made a bit of a mess with #define's in the film/movie/video section. everything else should work. are we ready ? mfg.as.dr IOhannes

3 3

Re: FFMPEG in GEM
by IOhannes zmoelnig 03 Dec '03

03 Dec '03

Henning Haaland Kulander wrote: > Hi! > > The problem now is that the image is flipped vertically and horizontally > and blue colors are green and vice versa. Do you have any idea why this > is, and how it can be fixed? What format does GEM use to store the image? flipping upside-down canbe done with the "upsidedown"-flag in the imageStruct-structure. the problem is, that openGL handles the upside-down orientation contrary to all other image-applications (1st pixel is lower-left edge instead of upper-left) he other problem is BGR-colorspace (or BGRA) vs RGB(A). (pix_* processing in) Gem supports 3 different colorspaces: GL_RGBA, yuv422 (UYVY) and Greyscale (GL_LUMINANCE). the [pix_texture] just passes the current-colorspace to the renderer, thus supporting other formats like GL_BGR too. however, the imageStruct-class has functions fromBGR(data) (and similar: like fromGrey()) to convert data into "native"-Gem-colourspace. how to use it ? 1. set-up the imageStruct correctly (xsize, ysize, format/csize/type can be handled with the setCsizeByFormat() finction - you only need to now the format this way) 2. get a pointer to the image-data you want to convert. unsigned char* bgr_data=<pointer to my BGR-data) shouldn`t be image.data; then call image.fromBGR(bgr_data); et voila. however, great you have had a look on this. i will incorporate it when i am at home (in 0.5h) mfg.asd.r IOhannes

1 0

Re: [GEM-dev] Re: [PD] physical modeling objects for pd
by Cyrille Henry 02 Dec '03

02 Dec '03

B. Bogart wrote: > Yes please upload it, I would like to see. > ok, I'm comminting to the usual place (http://drpichon.free.fr/pure-data/physical-modeling/) a directory contening things needed for this patch. it's a NewWave clone. their is an abstraction wich create the matrices patch, and some already made matrices. tell me if there is someting you don't understand. this patch is old, and was not desing for distributing. so, you will have some work to make it work on your computer... you will nead somes abstractions. I use a lots of "lb" abstraction : it's just a loadbang. try the patch and tell me all the abstraction you are missing. the lia object was call liaKD when I made this patch, so I think you can make an abstraction "liaKD" contening only "lia"... so, if you don't get it to work, I'll do this job later... (documentation first!) > I've been interested in this stuff for years, but did not have the math to > deal with the details. > equation are very simple... > I'm interested in using a few input controls that drive a large number of > interelated (complex) other controls.. (ie physical environment) > yes, that's what's interesting with this objects : you can build high level comportement that would be hard to do with equations etc... you can do chaotic mouvement very easily. (02_2D should be chaotique : I'll pay a beer to the first one who can explain why it becames peridic ;-) > These seem pretty fast, fast for control data. Signal processing is not the same... > a 4x4 matrix should not have that much more > overhead? what do you mean by overhead? >A matrix over x-y with forces only applied to Z would be great to > see also. > > something on my long term to-do list is fluid dynamics... woww, that's complettly diferent cause each masse has to be conmected to all the other masses. > (maybe we'll have > to wait a bit for computers to get faster. ;) > depend of what kind of simulation you are using : I think using specific mouvements equations can be necessary. frinds show my very good things made using matlab! ;-) I nead to find where i put my fluid mecanique book... > Thanks for all the work! > thanks, cyrille > Ben > > ----- Original Message ----- > From: "Cyrille Henry" <cyrille.henry(a)la-kitchen.fr> > To: "B. Bogart" <ben(a)ekran.org> > Cc: <gem-dev(a)iem.kug.ac.at> > Sent: Tuesday, December 02, 2003 2:47 PM > Subject: [GEM-dev] Re: [PD] physical modeling objects for pd > > > >>hello, >> >>I've got a patch of 1D masses, in a 2D matrix. >>so only 1D force can be aply to them. >>it's very similar to newWave (but less efficient, cause of multi send / >>receive / route object I use.) >>this specific patch use primTri for rendering. >>using curve3d would alow to have less control points, but the physics of >>the mouvement will be less present.. >> >>to modify this patch for a 3D masses and lia is on my todo list... >> >>if your interested, I can upload it... >> >>Cyrille >> >>B. Bogart wrote: >> >> >>>Hey all, >>> >>>I'm sure your tired of hearing me say this but... >>> >>>I think Cyrille's physical modeling stuff would be most beutifully >>>(gem-wise) manifested in Chris's vertex_array stuff. I think the message >>>over-head of sending the data to pix_vertex for each vertex would > > probably > >>>be a performance anoyance, so how about something faster to allow these >>>physical models to be applied to geometry? (I suppose this is also > > closely > >>>related to the issue of high-performance GUI in pd for this stuff...) >>> >>>Well I'm just thinking out loud again... >>> >>>Oh and cyrille, do you have an example of a patch that uses lia and > > masse in > >>>a two dimentional matrix? (with the ability to put forces on it in 3 >>>dimensions). for example having each node in the matrix control a > > curve3d > >>>control point? :) >>> >>>Ben >>> >>>----- Original Message ----- >>>From: "Cyrille Henry" <cyrille.henry(a)la-kitchen.fr> >>>To: "Frank Barknecht" <fbar(a)footils.org> >>>Cc: "PDlist" <pd-list(a)iem.kug.ac.at> >>>Sent: Tuesday, December 02, 2003 9:52 AM >>>Subject: Re: [PD] physical modeling objects for pd >>> >>> >>> >>> >>>>Frank Barknecht wrote: >>>> >>>> >>>> >>>>>Hallo, >>>>>Cyrille Henry hat gesagt: // Cyrille Henry wrote: >>>>> >>>>> >>>>> >>>>> >>>>>>I have made a serie of objects for physical modeling with pd. >>>>>>The object send Data but NO AUDIO. >>>>>> >>>>>>the aim is to generate data for the control of an audio synthesis > > and/or > >>>>>>video synthesis... >>>>>>it's just like http://www.sodaplay.com/ for pd. >>>>> >>>>> >>>>>I always loaved soda. Now I have soemthing else to love: Yours are >>>>>wonderful objects, soo nice. >>>> >>>>thank's >>>> >>>> >>>> >>>>>>I upload sources + object for windows and linux + exemples here : >>>>>> >>>>>>http://drpichon.free.fr/pure-data/physical-modeling/ >>>>> >>>>> >>>>>Could you also provide a source archive? >>>> >>>>ok : >>>>I add a zip file in the src directory with the .C files. >>>> >>>>cyrille >>>> >>>> >>>> >>>>>ciao >>>> >>>> >>>> >>>>_______________________________________________ >>>>PD-list mailing list >>>>PD-list(a)iem.at >>>>http://iem.at/cgi-bin/mailman/listinfo/pd-list >>> >>> >> >> >>_______________________________________________ >>GEM-dev mailing list >>GEM-dev(a)iem.at >>http://iem.at/cgi-bin/mailman/listinfo/gem-dev > >

1 0

Re: [PD] physical modeling objects for pd
by B. Bogart 02 Dec '03

02 Dec '03

Hey all, I'm sure your tired of hearing me say this but... I think Cyrille's physical modeling stuff would be most beutifully (gem-wise) manifested in Chris's vertex_array stuff. I think the message over-head of sending the data to pix_vertex for each vertex would probably be a performance anoyance, so how about something faster to allow these physical models to be applied to geometry? (I suppose this is also closely related to the issue of high-performance GUI in pd for this stuff...) Well I'm just thinking out loud again... Oh and cyrille, do you have an example of a patch that uses lia and masse in a two dimentional matrix? (with the ability to put forces on it in 3 dimensions). for example having each node in the matrix control a curve3d control point? :) Ben ----- Original Message ----- From: "Cyrille Henry" <cyrille.henry(a)la-kitchen.fr> To: "Frank Barknecht" <fbar(a)footils.org> Cc: "PDlist" <pd-list(a)iem.kug.ac.at> Sent: Tuesday, December 02, 2003 9:52 AM Subject: Re: [PD] physical modeling objects for pd > > > Frank Barknecht wrote: > > > Hallo, > > Cyrille Henry hat gesagt: // Cyrille Henry wrote: > > > > > >>I have made a serie of objects for physical modeling with pd. > >>The object send Data but NO AUDIO. > >> > >>the aim is to generate data for the control of an audio synthesis and/or > >>video synthesis... > >>it's just like http://www.sodaplay.com/ for pd. > > > > > > I always loaved soda. Now I have soemthing else to love: Yours are > > wonderful objects, soo nice. > thank's > > > > > > >>I upload sources + object for windows and linux + exemples here : > >> > >>http://drpichon.free.fr/pure-data/physical-modeling/ > > > > > > Could you also provide a source archive? > ok : > I add a zip file in the src directory with the .C files. > > cyrille > > > > > ciao > > > > _______________________________________________ > PD-list mailing list > PD-list(a)iem.at > http://iem.at/cgi-bin/mailman/listinfo/pd-list

2 1

pix_film and pix_video
by Mirko Petrovich 02 Dec '03

02 Dec '03

Hi all, in latest Gem CVS, [pix_film] is not working, I get : pix_film:: quicktime support pix_film:: libmpeg3 support Direct Rendering enabled! opening ../data/anim-1.mov with format 1908 quicktime_check_sig: No such file or directory quicktime4linux: unsupported! mpeg3io_open_file: No such file or directory MPEG3: unsupported! If I compile with: configure --disable-new , [pix_film] works but then [pix_video] doesn't work. In the other side, [pix_video] is working great (thanks IOhannes !!) but I can't do alpha blending. What can I do to have both objects working at the same time ? Thanks for your time Saludos Mirko Petrovich Osorno, Chile

1 1

FTGL and freetype2
by ben＠ekran.org 02 Dec '03

02 Dec '03

Hey all, I'm *still* trying to compile gem CVS. I guess it's more interesting that I get a different problem eveytime I update! Now its FTGL and freetype2 I have freetype2 v 2.0.8-30 installed but FTGL stops on configure at: checking for freetype-config... /usr/bin/freetype-config checking for FreeType - version >= 9.0.3... configure: error: FreeType2 is required to compile this library Now the latest version of freetype on sourceforge is 1.3.1, the latest of freetype2 is 2.1.5, but I'm getting nothing anywhere near 9.0.3??? Is this a typo in FTGL configure? What version of freetype2 do I need? Oh and I'm using FTGL 2.07 So whats up?? Thanks Ben

3 2

gem cvs alpha-depth bug?
by ben＠ekran.org 01 Dec '03

01 Dec '03

hey all, I just updated my gem cvs again today. I've noticed an odd bug with alpha (again). this time the alpha object seems to turn on depth buffering when its in the chain! I'm using lighting (because lighting tends to get messed up when using "depth" to show the effect. without a depth object in a chain, the cube looks identical to its depth counterpart, as long as the alpha object is in the chain. If I turn off the alpha object it works fine. (looks as it does without using depth. Here is an example patch. Ben

4 10

Fw: [sc-users] OpenGL Sneak Peak
by B. Bogart 01 Dec '03

01 Dec '03

Hmmm! Chris, are you a prophet? B. ----- Original Message ----- From: "Jeremy Rotsztain" <jeremy(a)mantissa.ca> To: "Matthew Nish-Lapidus" <matt(a)ekran.org> Cc: "B. Bogart" <ben(a)ekran.org> Sent: Monday, December 01, 2003 9:17 AM Subject: FW: [sc-users] OpenGL Sneak Peak > > It looks like OpenGL is being introduced into SuperCollider! > > Jeremy > > ------ Forwarded Message > From: ccos <ccos(a)alphalink.com.au> > Reply-To: SuperCollider users mailing list <sc-users(a)create.ucsb.edu> > Date: Thu, 27 Nov 2003 22:05:02 +1100 > To: SuperCollider users mailing list <sc-users(a)create.ucsb.edu> > Subject: [sc-users] OpenGL Sneak Peak > > http://swiki.hfbk-hamburg.de:8888/MusicTechnology/569 > > i'll be back in a week ( or two ) > > _c > > _______________________________________________ > sc-users mailing list > sc-users(a)create.ucsb.edu > http://www.create.ucsb.edu/mailman/listinfo/sc-users > > >

1 0

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

GEM-dev December 2003