Is this not working as a matter of principle, or is there something that could be done in the Renderer? I'm using vade's TransparentBorderlessQCRenderer (by the way: thanks, vade!). In the composition it adds but then overlays to black when rendering to the screen. [...]]]>
Congrats! Having done several large projects, I can say this list is amazing. -- Jason Belec Sent from my iPhone On 2009-11-06, at 8:27 AM, Stephen Holmes <email@hidden> wrote: > quoted text [...]]]>
Home
Archives
FAQ
Terms/Conditions
Contact
RSS
Lists
About
Visit the Apple Store online or at retail locations.
1-800-MY-APPLE
Contact Apple | Terms of Use | Privacy Policy
Copyright © 2007 Apple Inc. All rights reserved. [...]]]>Well on Tuesday 1st at 10 am the Re-Rite:Be the Orchestra open to the public, this is the installation plays 30 camera feeds of the orchestra in sync across 3 floors of a deserted warehouse on the south bank of the river Thames. I have been working on this for 3 months and generally it has performed flawlessly over the last 3 days. The installation uses 9 Mac Pros running matrox triplehead2gos to drive 30 screens of 1024x768 content to a mixture of projectors and plasmas. This includes one Mac Pro driving 6 screens. There is touch screen control (running flash on PC using flosc to pass OSC to the mac) to allow the audience to control the volume of the separate sections of the orchestra, there are touch screens to allow the user to select different camera angles of the orchestra. There is even a room that uses a firewire camera and some clever masking to allow the general public to appear playing with the horn section. Anyway this was not a post to brag although I am very proud of it, but it is to say thank you to everyone on the forum for there help and patience (I know I have asked a lot of repeatedly stupid questions) and it certainly it is true to say that this installation would not have been possible without your help. I would also like to say a specific thank you to the kineme boys as this installation uses a number of their patches and it has been amazing the extra functionality and more importantly focus they have brought to Quartz Composer and I hope this continues even with the changes snow leopard has brought. The last person I would like to thank is vade for his customization of the vade v002 movie player that allows me to mix twelve audio tracks inside of a quicktime, it is very cool and works very well. (under 10.6 on a G5, still haven't got it working under 10.6 on my MacBook Pro. Anyway if anyone is in London before the 16th come down to the Bargehouse at Oxo Wharf and have a look at the installation. Thanks again everyone Steve Holmes]]>
Thanks for all the input. I've been busily digging into the many possible ways of handling things. And cursing a lot at 10.6 - not because of bugs for a change, but because every time I found a good way to do something it was labelled "Available in Mac OS X v10.6 and later." and I need 10.4 support ; [...]]]>
[...] I checked out the link Troy sent, using IOKit, and got a bit excited (because I also felt like periodic "don't sleep" messages are lame ;). After some testing, however, it looks like that only disables system sleep and display dimming, not disabling the screen saver. [...]]]>
I made my own FPS macro once to resolve the 'which is more accurate FPS meter question' and was edging towards FPS patch over viewer but not conclusively across all compositions ;)
2009/11/5 George Toledo <email@hidden>
Good call... On the iterated mesh example I posted fps is totally different (n/a in the Viewer, and around 5fps using the patch). I usually used to think the Viewer was more accurate, but in this case (iterated OpenCL mesh), I think the fps display reflects something closer to reality.
Going back through the other examples with the fps display patch, things are basically consistent whichever I use (the fps patch or the Viewer both reflect similar performance decreases).
-George Toledo On Wed, Nov 4, 2009 at 5:43 PM, vade <email@hidden> wrote:
Ive also noticed the FPS patch can offer very different results compared to what the FPS in the viewer reads. Not sure which to trust, but it seem like the viewer gives me more believable results to my eye.
On Nov 4, 2009, at 5:38 PM, George Toledo wrote: On Wed, Nov 4, 2009 at 5:15 PM, Christopher Wright <email@hidden> wrote:
Another test case that yields somewhat different results:
[...]
]]>I made my own FPS macro once to resolve the 'which is more accurate FPS meter question' and was edging towards FPS patch over viewer but not conclusively across all compositions ;)
2009/11/5 George Toledo <email@hidden>
Good call... On the iterated mesh example I posted fps is totally different (n/a in the Viewer, and around 5fps using the patch). I usually used to think the Viewer was more accurate, but in this case (iterated OpenCL mesh), I think the fps display reflects something closer to reality.
Going back through the other examples with the fps display patch, things are basically consistent whichever I use (the fps patch or the Viewer both reflect similar performance decreases).
-George Toledo On Wed, Nov 4, 2009 at 5:43 PM, vade <email@hidden> wrote:
Ive also noticed the FPS patch can offer very different results compared to what the FPS in the viewer reads. Not sure which to trust, but it seem like the viewer gives me more believable results to my eye.
On Nov 4, 2009, at 5:38 PM, George Toledo wrote: On Wed, Nov 4, 2009 at 5:15 PM, Christopher Wright <email@hidden> wrote:
Another test case that yields somewhat different results:
[...]
]]>Removing the 3rd party plugins was the solution! thanx! Some old patch was messing up things... [...]  Matthias Oostrik  www.magdatt.nl]]>
[...] Check out console logs, see if anything interesting gets printed there. Remove all 3rd party plugins (~/Library/Graphics/Quartz Composer Patches, ~/Library/Graphics/Quartz Composer Plugins, /Library/Graphics/ Quartz Composer Patches, /Library/Graphics/Quartz Composer Plugins). [...]]]>
I can't seem to get Quartz Composer 4 working on my MacMini (2ghz core 2 duo). When QC starts up it shuts down immediately. iTunes had the same problem, but that was solved by removing the "Quartz Composer Visualizer.bundle" from it's contents. Reinstalling Snow Leopard didn't help. [...]]]>
I've heard of issues with midi over network before, particularly when using ableton live. You might want to look at the vidvox guys free utility which does a similar job http://vvopensource.googlecode.com/files/MIDIviaOSC_0.1.0.zip to look into the thread where Midi over network is discussed (in relation to VDMX a Quartz Composer hosting app amongst other things) go here http://www.vidvox.net/phpBB2/viewtopic.php?t=3781&highlight=midi+network On 5 Nov 2009, at 13:32, Matthias Oostrik wrote: I did some more tests with the midi notes receiver. The problem seems to be in the combination midi over network and the QC notes receiver. When Ableton Live and QC are on the same computer there is no problem. But in different setups all using Midi over network (from audio midi setup), the QC note receiver only receives one note. This is both the case with osX.5 and osX.6, using different midi sources. Where should i file a bug report? Thanx Matthias On Tue, Nov 3, 2009 at 12:41 AM, Troy Koelling <email@hidden> wrote: The Midi Notes patch should listen to note-on messages and note-off messages and only flip bits when those messages come. That means that it should work like you expect, and I haven't seen any other complaints. It could be a special case with your setup? I see we handle "all-note-off" messages so perhaps your hardware is sending extra messages that we are handling as such? Please file a bug if you can. Thanks for the heads up. Troy On Oct 30, 2009, at 9:28 AM, Matthias Oostrik wrote: [...]]]>
I did some more tests with the midi notes receiver. The problem seems to be in the combination midi over network and the QC notes receiver. When Ableton Live and QC are on the same computer there is no problem. But in different setups all using Midi over network (from audio midi setup), the QC note receiver only receives  one note. This is both the case with osX.5 and osX.6, using different midi sources. Where should i file a bug report? Thanx Matthias On Tue, Nov 3, 2009 at 12:41 AM, Troy Koelling <email@hidden> wrote: The Midi Notes patch should listen to note-on messages and note-off messages and only flip bits when those messages come. That means that it should work like you expect, and I haven't seen any other complaints. It could be a special case with your setup? I see we handle "all-note-off" messages so perhaps your hardware is sending extra messages that we are handling as such? Please file a bug if you can. Thanks for the heads up. Troy On Oct 30, 2009, at 9:28 AM, Matthias Oostrik wrote: [...]]]>
Good call... On the iterated mesh example I posted fps is totally different (n/a in the Viewer, and around 5fps using the patch). I usually used to think the Viewer was more accurate, but in this case (iterated OpenCL mesh), I think the fps display reflects something closer to reality. [...]]]>
Ive also noticed the FPS patch can offer very different results compared to what the FPS in the viewer reads. Not sure which to trust, but it seem like the viewer gives me more believable results to my eye. On Nov 4, 2009, at 5:38 PM, George Toledo wrote: On Wed, Nov 4, 2009 at 5:15 PM, Christopher Wright <email@hidden> wrote: Another test case that yields somewhat different results: [...]]]>
On Wed, Nov 4, 2009 at 5:15 PM, Christopher Wright <email@hidden> wrote:
Another test case that yields somewhat different results:
[...]
Well, if you want to really get miffed, check this out (OpenCL Mesh iteration). I don't even get an fps reading on it, though changing iteration amounts obviously changes the apparent visual speed in the Viewer. Really odd. I can crank this thing up so that things even get a bit non responsive (I don't want to push into a freeze), and I still won't get an fps reading.
-George Toledo
Attachment:
Color Ball.qtz
Description: Binary data
]]>Another test case that yields somewhat different results: [...] FWIW, I see analogous performance differences (Snow Leopard fps = 0.5x * Leopard fps) for both Sprite and Billboard. Steve -- Steve Mokris [ email@hidden ] http://kineme.net/]]>
Another test case that yields somewhat different results: [...] Ahh, this is interesting - it's line based (not sprite based), so perhaps it's a function of which consumers/renderers are inside? curiouser and curiouser... :) -- [ christopher wright ] email@hidden http://kineme. [...]]]>
Similar results, different hardware. MacPro1,1 (2 x Dual-Core 2.66GHz Xeon, GeForce 7300GT), with disableVBLSyncing = YES. [...] 10.5.8 (9l30) = about 77fps 10.6.0 (10a432) = about 36fps 10.6.1 (10b503) = about 38fps 10.6.2 (10c535) = about 35fps [...] 10.5.8 (9l30) = about 6fps 10.6. [...]]]>
I'm creating the base mesh outside the Iterator, then passing it in, and adding per-vertex colour with the same OpenCL kernel that distorts the mesh, so I guess it's a different setup. a|x [...]]]>
Not to sidetrack, but for clarity on that issue, placing the mesh render patch in an iterator breaks the color values created by the mesh creator outside of the iterator. To be clear, if the mesh creator isn't actually setup to create color value - no structure input on the "Colors" input - everything works "ok", and one can do all the typical iterator color stuff by using the color control on the actual mesh render. I'm venturing a guess that you aren't creating your color values with _javascript_ input into the "Colors" input of the mesh creator and then iterating the mesh render (when I "unhook" that, I can at least control color on the mesh render inside of the iterator). Let me know if you are (offlist or not), because it may lead to a valid workaround for a bug that I need to file. -George Toledo On Wed, Nov 4, 2009 at 4:02 PM, Alex Drinkwater <email@hidden> wrote: [...]]]>
Funnily enough, I have managed to get generative mesh stuff working with an Iterator and OpenCL, with different coloured mesh per-Mesh Renderer instance. It's not as fast as I would like, but since it uses OpenCL, there's not way to compare with QC 3. a|x [...] [...]]]>
Pretty slow on my machine, too (flips between 19>21fps and 30fps). MacBook Pro 3.06GHz GeForce 9400M MacOS X 10.6.1 I was starting to think that Iterator performance was slow, too, actually, from other experiments, though I haven't done any formal tests. [...]]]>
I'm getting the same basic fps results using the same hardware to test both qtz's - performance is pretty much halved in SL (as in, half as fast). I could post more examples, but you've covered it.
I haven't found any scenarios where the new iterator is quicker, and the iterator also always seems to be a problem patch in scenarios where other things break, though it may just be coincidence.Â
Off the top of my head, iterator+shadows are a no-go, iterator + mesh creator results in the color values from the mesh creator getting broken (still need to file bug officially), and a particular queue+iterator thing that worked for me in 10.5 no longer works. This could all be side issues to this performance problem, but when taken together, it makes me dubious about the new iterator, in general.
-George Toledo On Wed, Nov 4, 2009 at 2:48 PM, Christopher Wright <email@hidden> wrote: > quoted text
]]><puts on asbestos suit ;)> I'm finding a lot of cases where iterator performance in QC4 is dramatically worse than in QC3 (typically about half as fast). Attachment: Checkers.qtz Description: application/quartzcomposer (this does a 50x50 grid, with half the sprites disabled, so it's [...]]]>
[...] it inherits all of OpenCV's bugs, plus whatever we've introduced. Given the less than exciting results I've had from randomly spot- checking CV's source code, I'm going to err on the side of caution with regards to stability :) Speed-wise, using CI to track a single point (using all GPU-resident filtering) should absolutely run circles around CVTools (which has to do a costly texture download, then filtering via CPU only). both methods should be satisfactorily fast (real-time), but that texture download really is killer (and PerformanceInspector will attest to that). -- [ christopher wright ] email@hidden http://kineme.net/ Attachment: smime.p7s Description: S/MIME cryptographic signature]]>
[...] After a bit of searching, it appears to be from nvidia (but using CoreImage/OS X, etc) -- this isn't where I originally found it, but it's using the same example image images and illustrates the same idea. Anyway: http://http.developer.nvidia.com/GPUGems3/gpugems3_ch26.html -- [ [...]]]>
[...] The Kineme CVTools 0.2 release works with Snow Leopard --- if some component isn't working on SL, it's a bug (please let us know specifically what the problem is; email email@hidden offlist). (Also I contend that it's not quite as slow or buggy as cwright claims above.. :^] ) [...] [...]]]>
Hey, [...] I saw that a couple of weeks ago and didn't worked because i'm on Snow Leopard or something, so I couldn't even try... [...] This sounds more like I'm looking for! :-) [...] Yes of course! :-) many thanks! WE are trying to make an installation [...]]]>
Is there a URL for the Apple Core Image example? On Nov 3, 2009, at 1:05 PM, Christopher Wright wrote: [...]]]>
[...] There are 2 general ways to do this that I've seen: use CVTools (freely available at kineme.net), and hope some kind soul helps fill in the blanks, or use Apple's CoreImage method. CVTools will provide you with accurate point tracking, and provide numeric outputs in a structure. [...]]]>
No it's fine. I've just seen motion trackers that analize 2 frames and use this difference for something, I haven't found quit yet a 'single point' tracker. I'm usind OSC Reciever and sender to communicate already... but I'm missing this detail... [...]]]>
[...] What you probably want to do is use the motion tracking stuff you've seen, and tie its output to the OSC patch -- that can provide output from QC via OSC. Look for "OSC Sender" to find the OSC output patch. (apologies if I completely misunderstood your requirements) -- [ [...]]]>
Dear QC devs and users, This is my first post/message, so please forgive my ignorance. I'm looking for a Plug-In or anything alike to track just a single point (area or pixel) and send these coordinates via OSC. I've seen motion trackes that work wonderful within QC (to control filters and [...]]]>
On Mon, Nov 2, 2009 at 6:43 PM, Troy Koelling <email@hidden> wrote: > quoted text
Thanks for the response, I really appreciate it. Will do! Best regards, George Toledo  > quoted text
]]>OK, I'm with you now. Another way might be to just draw the images in 3d as I suggested, turn off depth testing (so the sprites draw over the background), and scale the size by the inverse of Z to prevent them shrinking. That would put more of the maths into QC rather than [...]]]>
Hiya Chris Allright, I dug into the maths and seem to have solved it. I'll share it here, as I feel it is useful. QC file is attached this email. To project a 3d point onto space, use this formula and z-axis scale factor: Â Â Â var scaleFactor = 0.58; [...]]]>
[...] I can get it to happen with "Show both editor and viewer" in place of "show viewer only" -- also, if I launch QC (no windows become visible), open a composition via finder (windows then become visible), and then trigger the patch creator via cmd-enter, it's empty. [...]]]>
I've hit this one too. Can't remember what I did to get this to happen, unfortunately... I don't get it when following your steps though. I always have the QC prefs set to 'Do nothing', and 'Show both editor and viewer', so conditions wouldn't have been exactly as you've described. [...]]]>
set QC settings to "Do Nothing" and "Show only viewer". Close all open compositions, and the patch creator. then quit QC. PNG image Open a composition by double-clicking it in Finder (on the desktop is handy) -- only a visualizer should open (if the patch creator opens, [...]]]>
If you really want to do it manually, you'll need to learn some maths. You need to calculate the distance of a point on a sphere from the camera (the camera position in QC is (0,0,-1) if I remember right) to determine size and perspective transform. [...]]]>
Sorry if I was unclear: How to translate an (x, y, z) point into (x, y) screen coordinates -- For example, you have a rotating globe and want to know the screen coordinates of a certain city on the globe. Best regards Sune Watts --- On Tue, Nov 3, 2009 at 11:02 AM, Chris Wood <email@hidden> wrote: [...]]]>
QC supports 3D space already. There's 2 ways to show a 2D image: Billboard (this is best for overlays and the like, and is 2D based) and Sprite (this is similar to billboard, but exists in full 3D space so you can move it on Z, and rotate on all 3 axis. Most likely, sprite will cover your needs. [...]]]>
Hello all; I need to pin a 2D billboard onto a (moving) point in 3D space. I haven't been able to find any info on the z-axis scaling QC employs. Has anybody got a handle on mathematically converting a point in 3D space to 2D screen space? All the best -SW [...]]]>
[...]]]>
I wasn't sure myself, but it looks like this method might help. Note: I haven't tried it so YMMV http://www.cimgf.com/2009/10/14/the-journey-to-disabling-sleep-with-iokit/ On Nov 2, 2009, at 2:23 AM, Chris Wood wrote: [...]]]>
I'm afraid the 3D image type, like the z-axis on the Mouse Scroll patch is there "just in case" and we don't have support for it in QC. If you can think of an awesome use case for it, please do write an enhancement request. Thanks! Troy On Nov 2, 2009, at 1:44 AM, Alex Drinkwater wrote: [...] [...]]]>
That's very strange and I'm sorry that bug hit you. Could you please file a bug and private message me the number? Thank you, Troy On Oct 31, 2009, at 1:01 PM, George Toledo wrote: [...]]]>
The Midi Notes patch should listen to note-on messages and note-off messages and only flip bits when those messages come. That means that it should work like you expect, and I haven't seen any other complaints. It could be a special case with your setup? [...]]]>
Nice one. The Mac Research tutorial looks pretty good. On 2 Nov 2009, at 10:48, Memo Akten wrote: Hey, there is a good tutorial at http://www.macresearch.org/opencl It does cover a lot of the 'host programming' side too though, which is the stuff you need to write in order to use OpenCl - which you won't have to do when using OpenCL in QC. The apple dev docs have a pretty good intro too http://developer.apple.com/Mac/library/documentation/Performance/Conceptual/OpenCL_MacProgGuide/Introduction/Introduction.html#//apple_ref/doc/uid/TP40008312-CH1-SW1 again it covers most of the 'host programming' side which you can ignore if you want to use it in QC. For the actual OpenCL language itself (Which is what you'll use in the QC OpenCL patch), the official OpenCL Specification doc is the best: http://www.khronos.org/registry/cl/ particularly the chapter about the OpenCL Programming language On 2 Nov 2009, at 00:33, Josh Sophrin wrote: Holding it down in NYC and I have to say, mega super awesome visual company! Your web signature rocks! The dude with a mustache looks like cheech.... Where is a good place to get a good description of opencl? I've been working primarly with modul8 and I'm having to head back into composer for more detailed engineering experience. Kind regards, Josh Sophrin Sent from my iPhone On Nov 1, 2009, at 6:57 PM, Memo Akten <email@hidden> wrote: Hi all, I just thought I'd chip in quickly regarding this. I've found OpenCL to be quite unstable overall, even outside of QC - but only if you do things wrong. E.g. sending thousands of commands to a queue without flushing can cause your system to become like this http://www.flickr.com/photos/memotv/sets/72157622563642705/ (see the original sizes to see the problem) - only solution after that is to force shutdown. Obviously in this case the blame is with the programmer, but I would expect the results to be not as drastic, I"m guessing future versions of 10.6 will address these kind of issues. In QC too i've found it very easy to lock up the entire machine and cause kernel panics, but only if editing a kernel while the composition is running. Once a kernel is running properly with all inputs connected etc., I've found it to be quite stable (both in QC and outside of QC). Cheers, Memo. The Mega Super Awesome Visuals Company memo.tv 29 Ironworks. Dace Rd. London E3 2NX, UK mob : +44 (0) 7958 783 832 tel : +44 (0) 20 8123 9986 fax : +44 (0) 20 8986 5496 work: www.msavisuals.com play: www.memo.tv On 19 Oct 2009, at 20:31, Alessandro Sabatelli wrote: Are you having stability issues once deployed or only during production? .xX On Oct 19, 2009, at 7:40 AM, Troy Koelling wrote: > quoted text [...]]]>