At 3:07 PM -0700 11/20/03, the wade wrote:
OK, I know that pd/pidip is cool for linux 2-D stuff. What about winXP? I was trying to map video onto cubes in gem, but I could not avoid the frame refresh popping.
You mean tearing - like no vbl sync? VBL sync is a feature missing from the Win32 and Linux versions of GEM, but present (and the default setting) in the OSX version (it's literally one line of code in OSX). I think it could be added to both platforms, although I don't personally know how to do it.
No, I mean the audio stream being interupted for a screen refresh. I have a line object move from 0 to the last frame of the movie, using pix_movie and if I set make [gemwin 30] for example, there is a 30 Hz clicking in the audio stream.
a 720x480 avi is no texture I think, even with my 512MB ATI 9800 and AMD 2600 w/ 1GB ram.
You have a 512MB 9800!?! I didn't even know those existed - I'm envious. Anyway, that setup should do at least one stream of 720x480 no problem. Have you tried setting the 'frame' message to the refresh rate of your montior and then using a counter object to drive pix_film playback (in other words don't use 'auto').
Where is the syntax and list of messages for the pix_film object? Will this help?
Whats the tool for windows? (as my linux laptop is pokey with its usb-audio card that doesnt really work well... >100ms latency)
Well, consider replacing that laptop with a Mac laptop if you are desirous of doing real-time video. All of the current models will do at least one DV size stream even the iBooks. The Powerbooks will do two or more depending on the hard disk used. Both Jamie and I develop GEM on Tibooks, so the code is well tested on them.
cgc
If theres no x86 64bit machine out by the time the g5's are in laptop form, then I will get a g5 and put linux on it. I just remember coding being a pain on Mac's because of the protection of all their proprietary hardware. Haterid of the black box approach I guess. Viva open source!
Thanks for the help! -thewade
Ya, what I said ^ | dito
At 8:21 AM -0700 11/21/03, thewade wrote:
No, I mean the audio stream being interupted for a screen refresh. I have a line object move from 0 to the last frame of the movie, using pix_movie and if I set make [gemwin 30] for example, there is a 30 Hz clicking in the audio stream.
I see. That's no damn good, and unfortunately there's nothing that can be done about it currently. Pd blocks audio output when the GEM render chain runs. Threading might help this out if both processes were light enough that the CPU could complete them both in the designated time period.
Where is the syntax and list of messages for the pix_film object? Will this help?
The help file for pix_film needs updating, but my tutorial covers the counter playback method. You can find it here: http://taproot.dyndns.org/~cgc/tutorial_gem/index.html
If theres no x86 64bit machine out by the time the g5's are in laptop form, then I will get a g5 and put linux on it. I just remember coding being a pain on Mac's because of the protection of all their proprietary hardware. Haterid of the black box approach I guess. Viva open source!
The only 'black box' issues I've run into on OSX are related to lack of documentation of APIs. I haven't seen any direct evidence that this is meant for 'protection of all their proprietary hardware' (x86 hardware is all proprietary anyway - you or I can't change Asus' motherboard design or affect Nvidia's GPUs in any real way - plus they own the rights to all of there wares which is the meaning of proprietary). There are benefits with having control over both hardware and software as well, and this is no more evident than the brilliant CHUD performance tools. Shark/Shikari has been responsible for pointing out nearly every one of GEM's performance problems, and in many cases we never would have found these issues without these tools. Intel has VTune for x86, but I don't see an open source set of tools that come close to doing what CHUD can do on OSX and PPC.
Thanks for the help! -thewade
Sure, any time.
cgc
Ya, what I said ^ | dito
PD-list mailing list PD-list@iem.at http://iem.at/cgi-bin/mailman/listinfo/pd-list