Hello,
I have a rather interesting 'feasibility question' regarding
integrating chromium (http://chromium.sourceforge.net ) and pd/gem.
I will be entering a new media graduate program, and the CS dept at
the school has mentioned it would like to work with the integrated
digital media program in building and creating tools.
We are interested in using cave systems/multipipe projection systems
for interactive art performances, but no tools exist 'out of box' to
create a cave system with 'n' projectors in 'y' configuration..
(afaik), that allow for realtime performance.
So I had the idea of potentially embedding the chromium engine in to
PD and Gem, which would allow for an interactive, dynamic, multi
screen environment that is now a platform for performance.
So, my question is, is it feasible to integrate chromium in to gem in
such away so that a gem patch could be run spanning multiple machines
and multiple displays? Im imagining a system where one could
generate PD patches for, and then plug it in to a multipipe system
running chromium-PD/Gem.
Thanks for any information either way,
vade///
halo
vade schrieb:
We are interested in using cave systems/multipipe projection systems
for interactive art performances, but no tools exist 'out of box' to
create a cave system with 'n' projectors in 'y' configuration..
(afaik), that allow for realtime performance.
if you'd not mind being bound to windows and to a licensing model that asks you to pay on commercial use i'd recommend you having a look at our http://vvvv.meso.net thing.
basically similar to pd/gem but already equipped 'out of box' with a multiscreen output system dubbed 'boygrouping'.
writing this i realize that we are really lacking documentation on this feature. but http://vvvv.meso.net/tiki-index.php?page=Boygrouping should give you an idea of how to operate a boygroup.
alles gute. joreg.
hi,
chromium is non-invasive. this means that you do not have to do anything to either Gem nor chromium to run Gem via a chromium-renderer. this means: it (should) work "out of the box"
i have tested chromium+Gem only on linux and it did work (but i admit that i didn't do a project with it: just rendering the Gem-window to 2 different screens)
the only issue was, that i had to explicitly preload the appfaker-lib; search the chromium-users and gem-dev archives for exact information (unfortunately i cannot do this personally, right now)
mfg.as.dr IOhannes
Iohannes,
Ah, I was under the impression, applicaitons needed to be recompiled
to work with chromium. Not sure where I became under the impression.
I will give it a shot.
Thanks much,
-Anton // vad3
On Jun 18, 2005, at 11:06 AM, Johannes M Zmoelnig wrote:
hi,
chromium is non-invasive. this means that you do not have to do anything to either Gem nor
chromium to run Gem via a chromium-renderer. this means: it (should) work "out of the box"i have tested chromium+Gem only on linux and it did work (but i
admit that i didn't do a project with it: just rendering the Gem-window to 2 different screens)the only issue was, that i had to explicitly preload the appfaker-lib; search the chromium-users and gem-dev archives for exact information (unfortunately i cannot do this personally, right now)
mfg.as.dr IOhannes
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/ listinfo/pd-list
Well, as long as you have a fixed configuration during runtime, then yes chromium is transparent to the application. But, for cave-like setups that have head tracking and use off-axis projection to adjust the viewing frustums, you will need to put some chromium calls in the application. For example, see: http://chromium.sourceforge.net/doc/nonplanar.html
BTW, what kind of system are you guys building?
It sounds quite similar to our work. We (at McGill / SAT) are building a similar interactive/distributed rendering environment using Pd. We started last year with the Pd/Gem combination, but as we started to get more elaborate with our rendering engine, we looked to chromium to help with distribution over the network. It was a viable alternative, but in the end we decided to use a combination of Pd & OpenSceneGraph, and write our own world-coherence engine using a Pd-message based system and [netsend]. It works quite nicely. Much more flexibility in terms of rendering quality and graphical extensions than Gem, and you can maintain higher level control with the slave renderers. ie, the renderers still have knowledge of the world, and not just streamed openGL commands. This is essential is you want to do extra processing down the line (eg. shadow detection/removal, multi-projector blending, etc).
Cheers, -Mike
Ah, I was under the impression, applicaitons needed to be recompiled
to work with chromium. Not sure where I became under the impression.
I will give it a shot.chromium is non-invasive. this means that you do not have to do anything to either Gem nor
chromium to run Gem via a chromium-renderer. this means: it (should) work "out of the box"
One of the research areas of the TOT project (the host of pixelTANGO development) is lightTWIST on the same old www.tot.sat.qc.ca. This will eventually work with pd/Gem and pixelTANGO as it is already openGL and controlled from PD.
Only problem being is that it is very HW Dependant, each projector has its own shuttle and render commands are sent over GB multicast. All TOT software projects are GPL and so this could be a great place to start. Also a group a McGill in Montreal is working exactly chromium integration in Gem to create an AV cube "save" check the gem-dev archives... I think its in there. I've cced Mike Wozniewski on this topic, I've not checked into their progress in some time.
Keep in touch about this project, sounds interesting.
<B>
vade wrote:
Hello,
I have a rather interesting 'feasibility question' regarding integrating chromium (http://chromium.sourceforge.net ) and pd/gem.
I will be entering a new media graduate program, and the CS dept at the school has mentioned it would like to work with the integrated digital media program in building and creating tools.
We are interested in using cave systems/multipipe projection systems for interactive art performances, but no tools exist 'out of box' to create a cave system with 'n' projectors in 'y' configuration.. (afaik), that allow for realtime performance.
So I had the idea of potentially embedding the chromium engine in to PD and Gem, which would allow for an interactive, dynamic, multi screen environment that is now a platform for performance.
So, my question is, is it feasible to integrate chromium in to gem in such away so that a gem patch could be run spanning multiple machines and multiple displays? Im imagining a system where one could generate PD patches for, and then plug it in to a multipipe system running chromium-PD/Gem.
Thanks for any information either way,
vade///
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
hi everyone i am taking a plunge here and attempting to learn pd , and i know that i will have a lot of questions ..some good and some that will be oblivious ....first question is , who can point me in the direction of some good working .pd files ? (other than what pd comes with)
if you look at pure-data.org, you should be able to find all the
tutorial packages, and help files for what ever libraries you might
be interested in. the gem docs are quite good, as are the help files
in the newer builds of the PD apps.
if you have specific things you are looking for/to do, feel free to
ask, but for any basic functionality (ie logic, how do I get audio i/
o, objects, etc), the basic help files should lay that all out.
Good luck and welcome to the community!.
On Jun 21, 2005, at 2:28 PM, dustin sedlacek wrote:
hi everyone i am taking a plunge here and attempting to learn pd ,
and i know that i will have a lot of questions ..some good and some
that will be oblivious ....first question is , who can point me in
the direction of some good working .pd files ? (other than what pd
comes with) _______________________________________________ PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/ listinfo/pd-list
Hi all,
I am in the throws of tetsing dozens and dozens of video codecs to find the one that is the most "right".
I want to decode 30 minutes of video (no sound) at 25fps (no dropped frames) at 640 x 280... from a 10gig scsi hard drive.... decoding must take up no more than around 30% CPU on a 2ghz processor (need the rest for audio)
codec must have no frame differenceing for fast access to sepcific frames... revers playback etc.
I'm using PDP .... which does yuv422 planar. ......
- mjpeg A (as encoded by transcode) chews 65% cpu
- mjpeg B (as encoded by Premiere) chews around 50% cpu
- component video (packed yuv422) chews around 30% cpu (acceptible) but
takes up too much hard drive space
- planar 4:2:0 is good ... its the closest ..... chews 25% cpu , takes
less hard drive than component but still too much hard drive.
theora is yuck, sorenson too much cpu, divx is small, fast and reasonable quality ....... but all these have frame differencing.
indeo 3 is poor quality but decode is very fast.... cant read indeo 4+ on linux (I think it is frame differencing anyway)
cinepak would do it, ..fast decode... but very poor quality.
.... any suggestions? ........... anyone found a mjpeg compressor that deocdes really fast?
does anyone know of a codec that is PLANAR .... but compresses each Y U and V plane? ..... something like that would be real fast for pdp.
etienne
Try Photo-JPEG if available. Also, those CPU numbers look really high to me. Maybe something is not configured optimally?
cgc
On Jun 21, 2005, at 9:44 PM, etienne deleflie wrote:
Hi all,
I am in the throws of tetsing dozens and dozens of video codecs to find the one that is the most "right".
I want to decode 30 minutes of video (no sound) at 25fps (no dropped frames) at 640 x 280... from a 10gig scsi hard drive.... decoding must take up no more than around 30% CPU on a 2ghz processor (need the rest for audio)
codec must have no frame differenceing for fast access to sepcific frames... revers playback etc.
I'm using PDP .... which does yuv422 planar. ......
- mjpeg A (as encoded by transcode) chews 65% cpu
- mjpeg B (as encoded by Premiere) chews around 50% cpu
- component video (packed yuv422) chews around 30% cpu (acceptible)
but takes up too much hard drive space
- planar 4:2:0 is good ... its the closest ..... chews 25% cpu , takes
less hard drive than component but still too much hard drive.
theora is yuck, sorenson too much cpu, divx is small, fast and reasonable quality ....... but all these have frame differencing.
indeo 3 is poor quality but decode is very fast.... cant read indeo 4+ on linux (I think it is frame differencing anyway)
cinepak would do it, ..fast decode... but very poor quality.
.... any suggestions? ........... anyone found a mjpeg compressor that deocdes really fast?
does anyone know of a codec that is PLANAR .... but compresses each Y U and V plane? ..... something like that would be real fast for pdp.
etienne
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
chris clepper wrote:
Try Photo-JPEG if available. Also, those CPU numbers look really high to me. Maybe something is not configured optimally?
it feels that way ... but the question is what ... are there performance parameters to pdp ?
I have re-compiled, from scratch, the latest versions of libquicktime, pdp and pidip .... (as well as ffmpeg)
... and I have also tried photo-JPEG which yields similar results to motion jpeg A ...
Etienne
cgc
On Jun 21, 2005, at 9:44 PM, etienne deleflie wrote:
Hi all,
I am in the throws of tetsing dozens and dozens of video codecs to find the one that is the most "right".
I want to decode 30 minutes of video (no sound) at 25fps (no dropped frames) at 640 x 280... from a 10gig scsi hard drive.... decoding must take up no more than around 30% CPU on a 2ghz processor (need the rest for audio)
codec must have no frame differenceing for fast access to sepcific frames... revers playback etc.
I'm using PDP .... which does yuv422 planar. ......
- mjpeg A (as encoded by transcode) chews 65% cpu
- mjpeg B (as encoded by Premiere) chews around 50% cpu
- component video (packed yuv422) chews around 30% cpu (acceptible)
but takes up too much hard drive space
- planar 4:2:0 is good ... its the closest ..... chews 25% cpu ,
takes less hard drive than component but still too much hard drive.
theora is yuck, sorenson too much cpu, divx is small, fast and reasonable quality ....... but all these have frame differencing.
indeo 3 is poor quality but decode is very fast.... cant read indeo 4+ on linux (I think it is frame differencing anyway)
cinepak would do it, ..fast decode... but very poor quality.
.... any suggestions? ........... anyone found a mjpeg compressor that deocdes really fast?
does anyone know of a codec that is PLANAR .... but compresses each Y U and V plane? ..... something like that would be real fast for pdp.
etienne
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Hey Etienne,
Just to chime in, I don't know the solution to your problem, but I've not heard of people using PDP at such a high-resolution.
Anyhow what comes out of your comments here is the fact that there is a size/decoding speed trade-off. I think your seeing the best options and I'd suggest going with the bigger files... If the HD is not big enough... :( I just got my new AMD linux machine and got a RAID0 array of dual SATA drives.. I'm hoping this will work well for what your considering. I'll be doing some steps once it is up and running, but the raid setup was a little more complex than I was hoping/expecting!
I have not tried it but you can control the compression quality of a photo-jpeg video, to match the size, and then try and use a well optimized jpeg decompression library. photo-jpeg videos are big though... :(
Tried the same tests with Gem? Any difference?
Good luck.
B>
PS: anyone put a wiki on pure-data.info about codecs?
etienne deleflie wrote:
Hi all,
I am in the throws of tetsing dozens and dozens of video codecs to find the one that is the most "right".
I want to decode 30 minutes of video (no sound) at 25fps (no dropped frames) at 640 x 280... from a 10gig scsi hard drive.... decoding must take up no more than around 30% CPU on a 2ghz processor (need the rest for audio)
codec must have no frame differenceing for fast access to sepcific frames... revers playback etc.
I'm using PDP .... which does yuv422 planar. ......
- mjpeg A (as encoded by transcode) chews 65% cpu
- mjpeg B (as encoded by Premiere) chews around 50% cpu
- component video (packed yuv422) chews around 30% cpu (acceptible) but
takes up too much hard drive space
- planar 4:2:0 is good ... its the closest ..... chews 25% cpu , takes
less hard drive than component but still too much hard drive.
theora is yuck, sorenson too much cpu, divx is small, fast and reasonable quality ....... but all these have frame differencing.
indeo 3 is poor quality but decode is very fast.... cant read indeo 4+ on linux (I think it is frame differencing anyway)
cinepak would do it, ..fast decode... but very poor quality.
.... any suggestions? ........... anyone found a mjpeg compressor that deocdes really fast?
does anyone know of a codec that is PLANAR .... but compresses each Y U and V plane? ..... something like that would be real fast for pdp.
etienne
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
B. Bogart wrote:
Hey Etienne,
Just to chime in, I don't know the solution to your problem, but I've not heard of people using PDP at such a high-resolution.
.... other than the video decoding, it seems to perform very well.
Anyhow what comes out of your comments here is the fact that there is a size/decoding speed trade-off. I think your seeing the best options and I'd suggest going with the bigger files...
ya, ya, .... I've got two 5 gig scsi disks (RAID 0) running off a PCMCIA scsi card off a laptop ... so disk access speed is fine.
In the past (in jmax) I had a similar setup, reading raw component video (same frame size/rate).... with about 5% cpu. Big files, low compression, fast.
Tried the same tests with Gem? Any difference?
same kind of results for Gem.
although I encoded one video using mencoder .... forced ffmpeg mjpg as the codec. Mplayer plays it back with really low cpu (4% or something). Mplayer reports that it is using ffmpeg as the decoder.
when I try to open that same file in pix_film ... I get a message "ffmpeg: unsurported codec" .......wot? so pix_film is using a different version of ffmpeg? so I downloaded the latest ffmpeg and compiled it. No difference, pix_film still cant read it.
I got Gem from the demudi apt tree ..... does Gem come with its own version of ffmpeg statically linked inside it or something?
when people say that mjpeg is fast are they talking about avi ones or mov ones? ..... because xmovie decodes the libquicktime encoded mjpa files at the same speed as pd_qt
..............its all so god damn confusing.
Etienne
Good luck.
B>
PS: anyone put a wiki on pure-data.info about codecs?
etienne deleflie wrote:
Hi all,
I am in the throws of tetsing dozens and dozens of video codecs to find the one that is the most "right".
I want to decode 30 minutes of video (no sound) at 25fps (no dropped frames) at 640 x 280... from a 10gig scsi hard drive.... decoding must take up no more than around 30% CPU on a 2ghz processor (need the rest for audio)
codec must have no frame differenceing for fast access to sepcific frames... revers playback etc.
I'm using PDP .... which does yuv422 planar. ......
- mjpeg A (as encoded by transcode) chews 65% cpu
- mjpeg B (as encoded by Premiere) chews around 50% cpu
- component video (packed yuv422) chews around 30% cpu (acceptible) but
takes up too much hard drive space
- planar 4:2:0 is good ... its the closest ..... chews 25% cpu , takes
less hard drive than component but still too much hard drive.
theora is yuck, sorenson too much cpu, divx is small, fast and reasonable quality ....... but all these have frame differencing.
indeo 3 is poor quality but decode is very fast.... cant read indeo 4+ on linux (I think it is frame differencing anyway)
cinepak would do it, ..fast decode... but very poor quality.
.... any suggestions? ........... anyone found a mjpeg compressor that deocdes really fast?
does anyone know of a codec that is PLANAR .... but compresses each Y U and V plane? ..... something like that would be real fast for pdp.
etienne
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
etienne deleflie wrote:
Tried the same tests with Gem? Any difference?
same kind of results for Gem.
although I encoded one video using mencoder .... forced ffmpeg mjpg as the codec. Mplayer plays it back with really low cpu (4% or something). Mplayer reports that it is using ffmpeg as the decoder.
when I try to open that same file in pix_film ... I get a message "ffmpeg: unsurported codec" .......wot? so pix_film is using a different version of ffmpeg? so I downloaded the latest ffmpeg and compiled it. No difference, pix_film still cant read it.
I got Gem from the demudi apt tree ..... does Gem come with its own version of ffmpeg statically linked inside it or something?
on debian, ffmpeg (libavcodec-dev,...) is only available as static libraries. Gem will link against these.
however, i think that mplayer links against its own version of ffmpeg (and having a look at the marillat's package description for mplayer-586 i also find that it depends on libavcodeccvs; probably it is worth to try to compile against this)
mfg.asd.r IOhannes
IOhannes m zmoelnig wrote:
etienne deleflie wrote:
Tried the same tests with Gem? Any difference?
same kind of results for Gem.
although I encoded one video using mencoder .... forced ffmpeg mjpg as the codec. Mplayer plays it back with really low cpu (4% or something). Mplayer reports that it is using ffmpeg as the decoder.
when I try to open that same file in pix_film ... I get a message "ffmpeg: unsurported codec" .......wot? so pix_film is using a different version of ffmpeg? so I downloaded the latest ffmpeg and compiled it. No difference, pix_film still cant read it.
I got Gem from the demudi apt tree ..... does Gem come with its own version of ffmpeg statically linked inside it or something?
on debian, ffmpeg (libavcodec-dev,...) is only available as static libraries. Gem will link against these.
however, i think that mplayer links against its own version of ffmpeg (and having a look at the marillat's package description for mplayer-586 i also find that it depends on libavcodeccvs; probably it is worth to try to compile against this)
now that's really getting confusing!! ..........I thought libavcodec was part of ffmpeg ...... (at least it is in the ffmpeg latest cvs downloads.... and as doc'ed on the ffmpeg site).
would someone have a small MJPEG video I could download to see how much cpu my PD/GEM/PDP set up takes to decode it? ... then at least I would know if my problem is in the encoding or the decoding stages.
etienne
mfg.asd.r IOhannes
vade wrote:
if you look at pure-data.org, you should be able to find all the
good one!
pure-data.org has been detached from the pd-community for quite some time. actually the domain has been grabbed by evil people in early 2004. you can find more information onthe pd-list archives.
i've thought everybody knows this by now. anyhow what you are referring to is located at:
alternative names are http://puredata.org and http://pure-data.info
mfg.a.sdr IOhannes
PS: if you feel in the mood, you are welcome to buy pure-data.org back to the community.
Doh!
Sorry for the mis-information. muscle memory and all that. Anyway the
rest of my post stands ;)
On Jun 22, 2005, at 3:59 AM, IOhannes m zmoelnig wrote:
vade wrote:
if you look at pure-data.org, you should be able to find all the
good one!
pure-data.org has been detached from the pd-community for quite
some time. actually the domain has been grabbed by evil people in early 2004. you can find more information onthe pd-list archives.i've thought everybody knows this by now. anyhow what you are referring to is located at:
alternative names are http://puredata.org and http://pure-data.info
mfg.a.sdr IOhannes
PS: if you feel in the mood, you are welcome to buy pure-data.org
back to the community.
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/ listinfo/pd-list
-
B. Bogart -
chris clepper -
dustin sedlacek -
etienne deleflie -
IOhannes m zmoelnig -
Johannes M Zmoelnig -
joreg -
Josh Steiner -
Mike Wozniewski -
vade