|
|
the george toledo sample thread I get a pretty consistent stream of people contacting me on vimeo, myspace, facebook, email, etc., for sample code. I have hesitated to post a lot of my stuff, not because of any unwillingness to "give back", but because I tend to get the composition where I want it to, and don't clean up rogue macros that aren't enabled, and I often times don't think ahead and make a folder for the composition, so my file paths are rarely relative. This makes it a little bit of a task to revise something for posting. My only caveats with anything I post here are... if you use it, give credit. That's kind of a scouts honor thing. The other is, if you use it and you can, give a donation to kineme, because this is using up their bandwidth space, and nothing is for free (even if they are nice enough to just put a "donation" option up instead of charging for all of their cool plugins). These are two kineme3D based examples, one with some GL Tools stuff, and one with some particle tools. The Collider example is straight out of Chris's with some tweaks for my personal tastes; a different particle renderer, a weird plate/bowl object I made, some movement, and a different orientation. The other is original, but at the same time, after seeing Chris's macro examples, it's not as if any of this is a gigantic stretch (for me at least). These also make use of toneburst's HLSL to GLSL conversions for shading. So, for the people that are curious about this stuff, and other qtz's, I hope you will ferret this out, because I may not remember to get back to all of you, as the list has become somewhat longish... ...And... the most important caveat... I have a decent background with computers, coding, etc., but I view myself mainly as an artist, so I truly don't mind making decisions that are "wrong" from a technical standpoint (though I typically am at least aware that they are "wrong"). :o)
|
This is a good paint example, with feedback... this is fun for using with the value historian... if you record what you do, you can "play back" the painting later....
This is a version with that uses it with value historian for replay... Try changing the blending on the sprite inside of the render in image for a little different look... I probably could have published that, oh well.
This is the audio reactive comp that I loaded on vimeo... if you want it to look like that screengrab, just turn the polygon mode onto textured for both sides, and disconnect the hsl color change that is happening inside of that patch, on the torus renderer. It's a decent example of how to use the read pixel patch to "read" some things and not others, depending on what "layer" you put it on... and also cool to me, because it shows how read pixels can grab the output of consumer patches, without the need for a render in image... so it makes a fairly natural coupling with particle patches (to me at least).
This uses the latest beta, so you will have problems if you aren't using that build.
No one asked for this one, but that is neither here nor there :o) ! But, if you have the latest Kineme3D beta, with parametric surfaces, it may be of interest...
... a "cartoon" style audio reactive "head".
What I kind of happened into as I was doing this was the realization that you can get a lot of "play" just from messing around with scaling on a simple shape, instead of actually doing it with the math expression. Off the top of my head, the GL Scale is something I plan on messing around with in this context as well...
I retained the polygons, removed the triangles and exploited the audio into HSL colour patches, added some patch time to each layer & incorporated an audio spectrum macro within a 3d transformation and - with all due credit for the original posting by gtoledo - I present a someways simplified and in other ways a more complex version of an audio reactive GL based composition. Oh and its put on some weight - a fat file at 49.63KB
Cool, haven't looked at it yet, you used the safe audio patch lol....ugh. Kudos though!
That is very visually appealing! You can just chop out the readpixels, since it isn't doing anything...
I have a feeling you could probably figure out something to get the same/similar effect without the iterators and get some higher fps... not 100% sure on that :o)
A simpler but definitely more active audio reactive composition employing polygon & splines.
Ah, this doesn't use kineme... lame of me! However, I DO think it is kind of cool for showing how zoom blur can sorta simulate the effect of peripheral vision/dof... I think the key is "timing".
I love it! The .fbx stays with it, I assume? errr .. Yep. I see the 3D Object Loader. Really nice, George. Thanks so much.
Oh yeah, I should note that one has to keep the qtz nestled inside of this folder, because of relative paths and such... unless you want to manually reload it or whatever.
You could also try this with other 3d files that are "multi-structured". It will "draw/grow" in the order of the structure, or reverse depending on the interpolate... or you could power the number of iterations with randoms, mouse, whatever. When I made the fbx, I kind of had to take that into account so that it would look like some kind of branching, and go from the base up or the tips down depending on whether the interpolate that is hooked up to the iterator is increasing or decreasing in value... it is still a little random at parts.
You can load the whole object and observe the total elements in the structure to decide exactly what "range" you want to "interpolate" over. You may want to not start at 1 iteration for example, you might want to start by loading a bit more...In this example, I also hooked the "final" value of the interpolate that's driving the iterator up with another interpolate so that the final value would shift at various points and not always be the same- that's not quite so direct a method.
If I was going to start to "set up" something like this I would probably observe the total number of elements in the structure, set the final value of an interpolate to that, the initial value to 0, and then linear interpolate over a longish time period, so that I could see what it looks like as it slowly unfolds with no timing push and pulls, or hijinx.... and then make more sweeping "artsy" decisions once I was sure what I was dealing with.
in a post from yesterday. awesome demo, george!
on that note, this does refer directly to the question i was asking in the other post-- how can the origin be set so the forms "grow" out of a stem-like structure? (i have not looked at the qc patch yet...)
many thanks!
Thanks. I made a quickie image filter with it. Works good. For my own use, I'll publish out some ports and try to incorporate as many of your suggestions as I know how to do. I can tell you this right now .. it looks just swell stacked as a unique layer in substitution for those bubble particle things I always use. It's particularly groovy in those Spooky-patch feedback stacks I was exploring*. Way cool, George! Thanks again! :-)
*( I should probably clarify that I don't have any 'in-use' quartz comps that aren't reconfigured as image filters. That's how I use them. My 3rd party virtual cam application republishes all the published ports and I control everything from within the app. Currently, there aren't any in-use stand alone compositions on this bus, and I wouldn't know what to do with them if there were. They would just sit there gathering dust. Hope you understand. heh )
The fact that "growth" looks like it is happening in some kind of logical manner is dependent on the actual organization of the structure of the 3D model. When I exported this file, I exported it in a way that ordered each "part" in the same order that I actually created the pieces in. When I made the object, I took into account that I would basically want to "see" the lower pieces load first, and then proceed "up" the object, so that it would look like it was growing.
Had I REALLY wanted to take alot of care... I could have made a stem... then made another single stem, then a branch off of that. Then go back to the base, make another stem... and if you export a 3d object that reflects the order that you created those pieces in, then as you "load" more of the structure by increasing the amount of iterations, it will kind of "repeat" the order that you modeled in.
I basically DID do that, but I didn't take an extreme amount of time to make sure that each piece was connected to the main "trunks", and "wrote" groups of branches at the same time, so there is a little more "something popping out of nowhere" than "slowly growing" at points... though it does get across the "growing/branching" feel.
Oh yeah, your comment did spur me onto posting this, but I didn't chime in yesterday, because it isn't 100% an "automated/procedural" sort of branching/l-system, just a way of achieving a similar visual result. At the same time, I had already been messing with this for a bit, as I am into all of that branching type of stuff that people do in processing.... flight404's stuff comes to mind.
makes sense that the 3D model has to be prepared properly in this case. thanks so much for the clear explanation!
This is one of those that I would almost hesitate to post because I think it's REALLY cool... yet, it DID come out of Lee's observation of a teapot feeding back in the safari renderer, which led to this line of thought for me...
This one IS pretty. Thanks.
Those darned Render In Images are tempting, aren't they? I use lots of them during my explorations but I'm sorta reluctant to promote them in here due to their unpopular status as resource hogs.
Opening the other referenced qtz in Safari was an accidental observation on my part, btw. I clicked 'open in new window' instead of 'download linked file'. heh
It was a funny incident, because I had noted that safari does feedback if you don't put in the clear, and I think that subject had come up a week or so before on something Chris posted and I had busted Chris's chops on it (like I have any business doing that, of all people)... There is something in the apple docs like "you should always start with a clear" blah blah blah. So then I subsequently post that teapot without a clear :o)
Anyways... :O) Once you pointed it out again, I had a bit of a flash on it. I've used this feedback for a number of things since, and I think it's a nice alternative to the accumulator for certain things.
I ran this one overnight (about 7 hours) one day, and it was fine. I run it and variations as a simple screensaver all the time. Didn't make anything get sluggish. I think spaces hesitated for a second after the 7hr run, but I didn't pin it to QC, since it didn't keep responding that way... I think it was my sleep mode settings.
I think that this qtz I am attaching would solve any kind of problems with the feedback eating up resources... this is a mod that flashes a clear every so often. You can control when by editing the period, which I made a published parameter.
Now, that's interesting and useful. Thanks.
semi-off topic but .. you've also tried your 'growing' as a source in some Render In Image feedback exercises, I'm guessing? Very attractive.
I was going to upload something, but after looking at the model size, I think I will spare kineme the bandwidth! Suffice it to say, I have... it can look cool. I've also used it with the KuwaharaNagao filter and I like that as an interesting twist.
Ah, I really liked that. It reminds me of all the cool little programs that drew pretty patterns like this back in the olden 8/16bit era. Of course this looks way cooler, but it's along the same lines :D
It inspired me to sit down and do a quick bit of tweaking with your comp to see what was possible. That turned into a complete rewrite with a new idea, and rainbow smoke was born :) This is a really slow moving comp, you need to watch for a while.
Yeah, it reminds me of that stuff too!
Man, I would love to see this... but it fails on my computer! It is failing at one of the render in images, according to debug. Not sure which yet. Everything looks somewhat reasonable.
Truth be told, I HAVE actually reigned this in to make more useful/utilitarian things... like clouds, smoke, etc. I am loathe to post that stuff, lest it end up in one of those bundles from that company that seems to go on Quartzcompositions and bundle/rip off everything they see into that lame program... forget the name of it offhand.
oops, noticed after i uploaded that there's a few deadly errors in it (I can tell from your reply that you're on an nvidia card ;)
Problem is, I go to fix stuff and decide to just make one more tweak.. and end up with a totally different composition an hour later!
So, fixed Rainbow Smoke, plus all-new Northern Lights!
What's this company packaging our stuff up btw? I'll have to check there's nothing of mine in there, and given them an immense bollocking if there is ;)
Oh, both of these are frame-rate dependent. They'll look dark if they run too slow, and very bright if they run too fast. Adjust the window size to control it a bit..
Well...rainbow smoke works, and but northern lights is giving an error, RGBAf extension not supported! Oh well. :o) The rainbow smoke is very cool, and not what I expected.
On the note of yours about editing till it's a new composition... always have to make sure to not save over too much and save various versions, because it's easy to over edit. Then, after that, sometimes it is cool to look at the variations, see what was good in each, and then make the "best" overall.(Just total theoretical/creative method blather here).
Also... it's an intel.
Tell you the funny thing about that program- only reason I noticed is because I have a habit of looking for QC related stuff, and occasionally even downloading random mac graphics apps and looking in the contents. Soooo.... I was poking through one, and was thinking that the stuff looked really familiar, then I noticed that it was a bunch of QC composition stuff. I will look through my hard drive for that sooner or later and get back to you on it. My HDs are like a no man's land of qtz crap... It's in a downloads archive, which are my least organized of all.
I don't get why rainbow smoke would work but not northern lights - they're both using 32bit render targets. 16bit causes issues on nvidia as QC uses 16h instead of 16f mode for some reason, and only ati supports that.
However - on intel cards, nothing other than plain 8 bit rendering is supported, so it must be falling back to 8 bit mode anyway. It'll "work" in 8 bit, but it'll look nothing like how it should. Infact having just tried it, it looks horrible :/ Try it on a reasonable ati or nvidia card if you can.
Did you just use "Reasonable" and "Ati" and "nvidia" in the same sentence? I think the only quirk-less renderer I've come across on OS X is the Software Renderer, and even that might be incorrect (I've not tried it exhaustively).
What's the difference between 16f and 16h? (16f = "16-bit floating", 16h = "16-bit half float", where the word "half" is used only because "normal" floats are 32-bit). In the GL headers, I only see
GL_RGBA16F_ARB, which is to indicate floating-point pixel format for texture upload, andGL_HALF_FLOAT_ARB, which defines a type/size for texture data (similar toGL_FLOATand friends). As far as I can tell, there is not other 16bit GL format except for 16-bit integer (either 5-6-5, 1-5-5-5, or 16-bit-per-component, depending on the setting).Well....(edit, I see Chris's post above me... this is for psonice...)
One thing I can't figure out... because it crops up in your comps occasionally- why not just leave it on default? I say that out of ignorance... will it not default to 32 on your system? (this brings to mind that a host info patch that output graphics card type might be desirable). Also, so you know... if you do that, it won't just "drop down" to whatever works on a system, it tend to not work at all. This is the same case with the SSAO stuff... I remember you checking off 32 bit- it simply doesn't render at all if you do that, and will actually crash in that case.
In this case, it's still fubar, and I can't figure out why I keep getting that RGBAf not supported message. I can't figure out what patch is causing that, which is semi weird...
NEVERMIND- I see the one render buried deep in- if I change that to default, I don't get an error/debug mode, but I'm not seeing anything (image wise) either.
I know it's a relative thing... but somehow I have a feeling that you checking off 8bit on your system probably looks way worse than my system on default (8bit, with whatever mitigating code is/isn't there).
Ironically, the nVidia cards do have problems that the intels don't and vice versa. That has kept me from jumping out and getting another to compliment until more dust settles with Snow Leopard.
Northern Lights works here, but Rainbow Smoke checkerboards....
Hahah.... the opposite of me...
Ok, psonice, now I am officially busting your chops- MAKE YOUR FRIGGIN QTZ'S WORK WITH DEFAULTS (hehehh, hope you know I say that in good humor... I see the point what you doing and that is a cool thought too... but mondo glitches for everyone is a super suck).
BTW, I'm sure many of you have read/heard of this, but for those that haven't, Intel and nVidia are having a legal squabble, that hopefully won't effect release of the next OS... Intel's argument is "curious".
http://www.maximumpc.com/article/news/shots_returned_nvidia_responds_int...
I only use 32bit mode where it's really needed - for these 2 comps, 8bit is nowhere near enough and you get a mess of random colours instead of nice smooth rainbows and clouds. 8bit is the default, so there's not much point in leaving it at that.
Really, for stuff like this it's a plain choice between a card that supports 32bit or not watching it at all :D
I can't remember the specifics, but if you look at apple's openGL capabilites table you'll see the 16f/h lines, with only ATI supporting h. I can't remember the difference, but it's probably 16bit/channel integer mode. I really wish apple had gone for 16f where it's well supported!
Better still, i'll make it so it works on anything with a video card that has features from this century at least ;)
Franz: that's really odd. What GPU do you have?
the legal junk is about SLI and chipsets, IIRC -- apple only uses intel chipsets anyway (or maybe the new MB/MBP line use nv? not sure). The argument is a hardware-level thing, while snow leopard is software -- thus it shouldn't affect the timeline of SL (though it might affect the feature set a bit... probably nothing too debilitating though)
Well, if that's your thought on it, that eases my mind a bit... nVidia has been so involved in all of the OpenCL/Chronos group stuff that I was wary about this situation.
Yeah yeah yeah.... I wish I could remember that glitch that everyone was complaining about on the dev list recently that works fine on my intel :o)
In all seriousness though... if I had any inkling of how into QC I was going to get when I bought this particular mac, I probably would have sprung for the top of the line, instead of the bottom of the barrel! I'm alarmed by all of the poor performance complaints of various things on the nVidia though...
Hmmm, that's interesting. You are making me curious, that's for sure. It really sucks that if you leave your settings on 32 bit, that it throws errors on non-32 bit systems, or will give no render/crash on occasion... or that you can't leave your comp on default and have it do 32 bit on your system... that seems like a bit of a QC bug.
i'm on a Ge8600 M GT 512Mb...
console says:
3/8/09 9:05:44 PM Quartz Composer[10064] ***: Render failed at time (some time)
3/8/09 9:06:06 PM Quartz Composer[10067] *** CGLCreateContext() called from "_CreateGLContext" returned error 10009 ("invalid share context")
Very odd! That should support 32bit mode, so I can't imagine what could cause it. The two are nearly identical really, so I don't get why that doesn't work.
Quick render, this is what it SHOULD look like (well, it should look less compressed and much more detailed + high res, but..)
I'm a bit behind the times I'm afraid: but do any nvidia cards support 16bit contexts? Have they ever? my mbp (pre-unibody-ification-edition)'s opengl driver info says it supports 64bit float color buffers (half float), but Render In Image doesn't work in 16bit mode (with franz's error above).
Additionally, Apple's software OpenGL renderer claims to supports 64bit color buffers (half float), but also errors with that error (tested in QuartzCrystal).
From my (lossily compressed) memory, only ATI has ever supported Render In Image in 16bit mode -- all other drivers refuse to share contexts (QC requires context sharing, but hard-codes some internal context (possibly the viewer?) to not-16-bit, and requires Render In Image to fit with that context as well...
An experiment might be to write a small app that loads up a composition with a 16bit render in image, and also manages its own opengl context, and sets that context to 16bit half-float, and see if the error persists. Perhaps it's a subtle qc quirk that is secretly a driver issue in disguise.
The plot sickens!
That looks nice... Is that a different tweak, the motion doesn't look the same? The blending does look better here.
This misses out on the cloud effect, because of the change in time from system to local.
Nope, because it uses system time, it will look different each time it runs.
The blending varies on a few things: 8/32bit (8 has blocks of pure colour all over the place, 32 has nice smooth gradients), how much it's moving (slower movement = more colour), and FPS (more fps = more particles rendered = more additive blend = more colour on screen).
It might be worth a go to record the system time with the VH to semi-reproduce for render...
You know what... I stay away from using that kind of setup exactly because of the reasons you do use it! But I like the concept... that's a really cool thing about this board- seeing the valid uses of things that may not have occurred to you (me, in this case). It is nice to remember to use that for when I want something to be unique each play.
This discussion inspired me to make a new idea based in feedback that I'm rendering a piece of right now!... and probably won't post the composition, lol. I'm savin' it for crapp2app (QuartzBuilder)!
...and here it is...
Lovely stuff-- I get only black screen with rainbow smoke and northern lights. Painting in 3D works fine. Is it because of my MBP RadeonX1600 card? I don't understand...
Hey, I like that. It has a real 3d feel to it!
Hmm.. x1600 supports the 32bit render modes, so I don't know why it wouldn't work. One suggestion though: both are very slow-moving compositions, and sometimes the screen does stay black for a while (especially with a wide window - it can draw above or below the top of the window for a while) so maybe try leaving it running for a short while?
Just looked it up again, and according to the osx openGL info table:
http://developer.apple.com/graphicsimaging/opengl/capabilities/
the ATI cards support both 48-16 (presumably 16bit/channel integer) and 48-16h (half-float) modes. Nvidia only supports 48-16h.
I can't remember if how deeply i looked into this, but I have a feeling that I saw somewhere that QC was using this 48-16 mode, hence it only working on ATI (and you're right, it's never worked on nvidia).
I really shouldn't use 16bit mode at all, but it has less performance hit than 32bit, so it's great for all those quick effect tests if you have ati :)
Thanks! I'm happy with the way that it has a fairly "real" look to it (well, realistic in a fantasy sense... you know what I mean I think). It doesn't resemble too many qtz setups I've seen that's for sure! I can't bring myself to post it though... a little too cool to end up uncredited in some vj reel, or as a new iTunes screensaver or whatnot.
-heheh, I should edit the title.... "the psonice debug thread" :o)
http://www.quartzcompositions.com/phpBB2/upload/details.php?file=5879
This is an example by psonice that I think that anyone that is interested in QC should look at. I think it should stand in the QC "Hall of Fame".
I had to do another take...!
i appreciate the reply-- still no luck and it's been driving me a little crazy all day. i let it run for a long time, in different size viewer windows, in fullscreen... there's nothing in the qtz layout too too difficult for me to understand, eg, no complex javascript / CI filters, etc. sigh. i'll post back if anything reveals itself. thanks again.
and indeed, the 1988 qtz was a fantastic and generous demo for me!
Glad you guys found 1988 useful :)
I'll test this new stuff on an x1600 card later (we have an older imac in the office), but from now on please discuss anything to do with that here: http://kineme.net/Applications/Compositions/psonicescompositionsandstuff to avoid polluting george's thread any more :)
Tears of joy rolling from Satie's eye i'm sure. Feels like some frenchy-psycho-soap-drama! (that's good btw)
There is a bit of parody of art film in there. I mean.... what triggered me to do actually pair it with that music was that it made me laugh! It seemed a bit larger than life and over the top. The same with the second one with the audio collage, where there are the baby sounds, applause, etc... it struck me as nice and odd- and funny :o)
Okay I've watch the second now - they both made me laugh. When you parody art films in this way you're dangerously close to a recursive loop! (With accumulation no doubt). Love the Gelati Van audio distortion (and colours for that matter).
[Oh I think in US you call them Whippy Vans and they don't have gelati they have synthetic soft-serve but that's not what 'm saying.]
A recent post on OpenCV got me poking around again...
I was having too much fun with this not to plop it out there.
I am always a bit reticent to setup things like this with smoothing and number tolerance to setup enables, but it works ok.
The basic thing is... you move... render of GL materials happen. If you don't, after a bit, it peters out.
This makes use of Kris Harris's thing to ignore nan points, but I also went ahead and got rid of the kineme structure maker in that setup.
Totally abstract and random.
Another idea is to set something up like this to do a circle of gl points by feeding the x/y coordinates with lfo to do a simple circle (or oval, or square)... (which more often than not, circles around your head if you are working on a laptop)... and then stop. Sooo... you wind up with something that looks very much like head tracking/detection.
Depending on lighting, the thing that may be advisable to tweak would be the smooth patch that feeds the conditional running to the "enable" on the iterator... tolerance/decreasing duration most likely.
anyhoooo....
http://qctwit.blogspot.com/
I made a pretty straightforward quartz composer based visual, audio reactive application using Kineme's Quartz Builder that I'm putting up for download on my "box.net" account (to spare Kineme the bandwidth). We'll see how long "the box" holds up. I'm not sure how many downloads I get, but it seems like it has worked ok for Toneburst.
I don't know how active this blog will or won't be, but I decided to set it up.
It just flashes the window between a primary colour + white for me. Now and then a gaus gradient texture was visible filling the whole window.
! Ah that sucks.
If anyone else has tried this and gets this please let me know. I'm going to go ahead and post on the Quartz Builder debug thread.
Question that may help... do you get transparency at all? It may be that GL Tools isn't executing (speculation).
I definitely get transparency, but very little. Like 95% of the time it's just bright colour flashing to white, and very rarely I see a flash of the window behind it. Oh, sometimes I see a little bit of motion, but mostly it's just the gauss texture.
Hmm..what's it supposed to look like? My impression is that it's possibly working right, but it's like it's zoomed right in on the centre of it or something.
Edit : added a screenshot. It's like this, but flashy, with occasionally a glimpse of the background, or something that might be a few particles, but only for a frame or 2.
You are definitely just getting part of the composition loading. Ugh.
Well, it looks like it might look like that if it was zoomed out a whole lot :D
Are plugins bundled with the .app? That could be the cause if not, otherwise it could be that the app is trying to load newer versions of plugins I have installed and fails at a guess.
Alternatively, it could just be your settings not working quite the same on my PC. I've had that before, where things suddenly look totally different. Colour conversion is the worst for it (I did a demo ages back with really great looking noise filters, tried it on another mac and the noise wasn't even visible, same on a quartz crystal render!) but it can affect other things I think.
colour spaces - profiling - sRGB - pretty much a one size fits all colour profile for online presentations and computer screen delivered graphical media assets.
Won't cure all problems, should eliminate or make less likely some of the commoner colour conversion problems.
Thanks, good tip. That was actually a really extreme case (going from VERY grainy black + white to smooth greys). I fixed it by putting the whole comp in a render-to-image, and disabling colour correction. Not ideal, but the deadline had already passed by then :)
Oh it definitely wouldn't be a zoom thing... I can recognize from seeing it that only one part is working... a sprite with gaussian, but at that, it still doesn't look like the transparency is running correctly, and one of the layers just isn't executing (which would be something in a render in image piped to a sprite).
Yeah, the plugins are in the app and should be loading from the app not /Graphics, so theoretically that shouldn't be a problem.
I appreciate you checking it, and giving thought on it. I don't think that's any of the issues though. I'm venturing a guess that it's vade's optical flow plugin failing, but it's hard to say since it doesn't fail on my system in this context. It also looks like GL Tools isn't working correctly either.
I've built something very similar that's worked on a few different build macs, but that didn't incorporate the Vade optical flow... hmmm.
I would appreciate if anyone could let me know if this does run fine on their system.
Ok, it's looking like a GPU issue. Not 100% sure of course.
It is working on my X3100, and on a friends' mac mini- GMA950. So, I'm half relieved.
Have you used Vade's optical flow at all on your system psonice? I'm trying to track down possible culprits, and that is the only plugin/non standard item in the chain that could break to make you not see the "fluid". The other plugins are incidental and wouldn't "break the chain" so to speak.
doesn't work here on a MBP Ge8600 2.8 / 4gb. No fluids...
However, vade's plugs. do ALL work on my system.
Nope, not got that plugin at all.
You've already stated you have them all, so it is a moot point to specify the optical flow...
Hmm. 50%. Not too good. Intel yay, nVidia, nay. Weird.
I made another build of this... not sure if it will make a difference.
I got rid of the v002 blur which wasn't being used, and was an oversight on my part. Made sure to explicitly load a gl tools patch inside of the composition in case it wasn't loading the gl tools plugin correctly in your environment (though... it doesn't seem like it should make a difference). I also changed the window level to normal, instead of floating.
None of this should really make a difference though, so it's puzzling.
So... I'm including a slightly different version, hopefully that runs on the nVidia.
I'm also including a related thing that totally cuts out the v002 altogether, and only makes use of GL Tools, plugin wise. This has run fine on mac mini's and also on Lee's gpu(ati I think?), transparency and all.
Petri works beautifully, like it did before, please note this is on an iMac with NVIDIA GeForce 9400 GPU | Intel Core 2 Duo CPU.
Liquid - well it has not changed much for me, but as it is, I can clearly see that mouse focus to bottom left keeps the mouse dependent iterations going, but move elsewhere and the fun kind of stops, just a middle of the screen blur.
I'd be thinking about having a second billboard with a similar iteration running that would continue and compliment the mouse dependent top layer, though possibly keeping the blur , zoom thing out of that top billboard and play that in the lower layer's billboard
Petri seems to work fine (I get some nice blobs on the desktop following the mouse, which invert colour/transparency. There's no motion as such, the blobs appear and spread a little, but it looks right).
Liquid2: similar to before, but I get a little more transparency now. Still no 'liquid' effect as such.
I'm on an ati 2600 btw.
Oh, one thing that's a bit 'wrong' with petri: it only affects my main screen. I know the issues with dual screen, and performance, and all the rest, but it still looks a bit 'wrong' :)
i see the fluids for about two seconds then the window starts flashing colors then moves towards white over the course of ten secs or so. this is true for both versions, although the colors last longer before completely turning white in the second version. transparency works fine. i have all the plugins, i think.
This actually doesn't matter --
QuartzBuilder bundles in the plugins it detects (it can miss GLTools if you're just using the Alpha blend mode thing, but otherwise it finds everything), and also disables loading system-installed plugins. This happens so that if the user has conflicting plugins (an older version, or a dangerous plugin, or whatever), they never get loaded to cause problems. The only plugs that load in a QuartzBuilder app are the ones bundled inside [and a couple system ones, like the ExtraPatches.plugin].
FWIW, there is no zoom blur on the top. I'm not sure what you are getting because nothing is mouse dependent. It sounds like you aren't getting the particle system either, but you aren't quite clear.
Thanks for looking though.
I don't think the visual suggestion makes sense given what is actually going on the qtz... but it's fair enough.
Let me know if it runs on yours...
As is either QBuilder or v002 is, as they say, "nsfw". But I think it's the combo.
I've also recently had a v002 zoom blur not attach correctly to a port on state restoration on someone else's machine (in qtz form, not quartz builder).
Thanks for looking Jersmi... appreciated. Too bad it isn't working correct.
_double post...
I'm going to go ahead and use this to upload the actual qtz file. If anyone whose system this "doesn't" work on in Quartz Builder app form, and has optical flow could try this to see if it works in qtz form, it would help isolate the problem more.
I would like for v002/QBuilder to work well together without having to worry too much about it!
The one thing I do note after looking at this issue more is that the v002 optical flow is GPU based. I'm not sure if that could effect the way the the app works... but the fact that it has run fine on two Intel builds, and not on the nVidia and ATI is making it seemingly fall along GPU based lines. It's too small a sample to make any conclusions yet.
Attached is : liquid.qtz, which is basically the same as this, with some minor changes.
You will see a dark background on this one, since I was using alpha to setup transparency in the app. The clip should illustrate something close to what the particle system should look like with the optical flow.
No zoom blur then - my presumption borne of the following :-
I actually was referring to the centre of the presentation, after it has run for a while, it does seem to rest, temporarily ,as an ellipse, lighter than surrounding area, before recommencing its busy activity. Sometimes seems to pulse
Tried to get a screen grab - but just too difficult to catch when I wanted to.
I had assumed that mouse input was included - so I guess I was just fooling myself. :-)
See your point about the visualization, I was just thinking of how to avoid too much whiteness / brightness .
Not seen the .qtz, [I don't think], but it actually looks pretty brilliant as a standalone application.
Perhaps too much brightness, or luminosity being employed, [get lots of white / light colouring at certain points.
Mainly get lots of wonderful random and fluid motion graphics generated.
A really interesting piece.
Oh and , maybe its a subjective response, but the transparency seems to be really - well transparent - beautiful.
Hahah, yeah I've been there (re:thinking something is doing something and it's not). Last night at 4am I was tweaking something, and then realized I was tweaking inside of a disabled macro :o/ At that point I was thinking, "ok, need to go to sleep". Did I? No....
Well, it's heartening that it's apparently running on your system. That still leaves it a 50/50 thing basically... which isn't great.
The qtz won't look like quite like the app, because of the aforementioned alpha=black instead of transparent issue.
Yeah, the white/bright flashes are something I considered taming, but I liked the intensity of it... it makes the background gradient blend into the particle system as well. With just the gradient along (like psonice, franz, and jersmi are getting)... not too exciting.
Do you think that the original plugin locations could be having any effect on this problem? I have some in system/library and others in library... still others, in my actual user folder (as you know). It definitely has all the necessary plugs in the app.
Being that it won't malfunction on the laptop I'm using, or the mac mini I have access to, I can't get any useful console info.
I don't think the original locations matter -- though I've not tested. As far as I recall, it loads the composition using QC, then asks all the patches which plugin they belong to (they're loaded when the composition is loaded, by QC), then copies the plugins that aren't built-in ones from wherever they are. If they're all getting into the app, then it's finding them just fine.
hopefully someone that has it malfunction can post a console dump or something? (if anything even shows up)
ahh, the joys of app development :)
george, I think I owe you an apology, wasn't seeing the wood for the trees.
when I start the liquid.qtz, I get the following response in QC.
I have got the following plugins installed
I have not included the Image Units or QC Patches.
So I guess you were spot on earlier when you suggested that I , and others, might not be loading the full version of the composition.
Hardware / Plugin issues / differences :-(
Oh well - bleeding edge software :-) Love it really
Yeppers. Wunderbar....lol. Ugh.
FWIW, anyone can go in the internals of the app, and pull out the v002 optical flow plugin and install in plugins, if they want to check out the qtz. (which may or may not work at that point... most curious to find out).
Oh yeah, while I'm thinking about it, since you are doing QBuilder stuff too... Image Units, like NI, and the FreeFrame plugin are two things that it's not going to handle unless the person has the stuff installed (Quartz Builder will not import/run that stuff.... the image units, or the frf folders for that are needed for the free frame plugin to work).
So, it is interesting, because it sounds like GL Tools is at least initializing for you. It sounds like both GL Tools and the Optical Flow are failing all around.
I get those as well when loading the composition in QC -- not sure which plugin I need to grab to see if it is missing it somehow (rather busy at the moment :/)
They are in the internals of the app, but I will load it here to make it easy for anyone that should want to check out the qtz... to see if it loads in regular QC land, and not Quartz Builder.
It should go in the plugins folder.
I will feel BEST about it if it doesn't work in regular quartz composer for those that it's failing in quartz builder for... but I'm not certain that will prove to be the case :)
Incidentally, in trying to put together what is happening, I've had another scenario where a composition breaks noodle connection using v002 plugins, EVEN with all of the same plugins installed. So, I'm postulating that there is something with the v002 plugins that can occasionally break connection when moving to another machine. If so, it seems that this follows pattern. The v002 optical flow connection to the sprite may be breaking on restoration when you move it to another machine. Why (?), is the head scratcher.
The one machine besides mine that it DOES work on apparently, doesn't have any QC installed at all.
Fellow Forum Readers and Contributors, Please allow me to enlighten you on the basis of my experience to date with this Liquid thread :-)
According to received wisdom one should place the vade plugins into /Library/Graphics/Quartz Composer Plug-ins/ or ~/Library/Graphics/Quartz Composer Plug-ins/ - the latter being the place I had installed the plugins.
Some of them worked from there - they really did.
The one's Liquid.qtz needs, didn't work, at least, not until I moved them up to the /Library/Graphics/Quartz Composer Plug-ins/.
That explains why the .qtz was not working , but the Liquid app was [I think ....]
What a relief.
Does raise a question for all QC Plug-Ins creators though - most plugins work from either folder. Why don't these ?
I also found that Glitch wouldn't work if it was in the ~/Library/Graphics/Quartz Composer Plug-Ins/ folder.
!!!
Can anyone confirm this? if so, it's a complete disaster that QuartzBuilder can't deal with (plugins expecting to be in a specific location).
To answer your question as to why they won't work, it's possibly trying to load resources from a hard-coded path (Really Bad Idea), or doing some other hard-coded path expectation tricks. Not sure, to be honest.
Cybero, I'm way too tired to think right, but you are my angel right now.
In my improper state restoration scenario (NOT with this app/qtz, but a different qtz)... I can see that in that scenario one of the v002 plugins is in system/library, the other is in library.
In this scenario, my optical flow plugin location is currently exactly where you describe having to move it to. You ideally, should be able to place it in either place and have it work on your end (not even addressing Quartz Builder here, just the raw qtz).
This makes me really curious to move all of my plugins to my user folder, reopen the qtz, and see if any connections break, repair, and rebuild/compile. It shouldn't be a problem, but I have feeling that may actually work. It also reliably works on a system withOUT QC installed, so that further makes me thing that something funky is going on with state restoration/plugin location/v002.
It's also worth noting that there may be an issue when more than one patch of the same plugin from v002 is used on a qtz, and the second loaded patch from that plugin failing. However, your observations seem to add a new layer to this.
(for record, to anyone looking at this: If you are seeing mostly a particle system with a little gooshiness, and an orb that blobs up from the center and is bright every so often, then it's working as an app. If you are running it as a qtz, it should be similar, but with a dark background.)
Chris, do you think that given this... that moving all plugins to one level folder (like my user folder or something), reopening the qtz, fixing connections if any break(not sure if that will happen or not) and recompiling in Quartz Builder may actually cause the app to work when it compiles?
I am going to definitely try this, but I've been at it for a long time here, and need to take a break I think!
liquid.app seems to be working for me regardless as to whether I place v002 Optical Flow.plugin in the User location or in the System location. Same is true for liquid.qtz. At least I assume they are both working .. both liquids are very flashy, eh? Sometimes I suspect that George drinks even more coffee than me, but that's unlikely. heh. Anyway, I don't see any observable difference in either plugin placement circumstance on my particular Mac.
Ah, sweet, I was hoping you would check it out Lee! Hmm, well at least that confirms that it can work on other systems besides mine and the mac mini.
I just got done updating OS to 10.5.7... I had been putting it off. I also went ahead and took every one of my plugins/patches, and migrated them all to the user location ... save for my image units and frf files.
Interestingly, when I open QC now, patch creator doesn't pop up by default, and I swear that it used to! It's no big deal though, it does pop up. It could be the reboot, the update, or the fact the plugins aren't loading across three different folders, but things seem a bit smoother actually.
So, plugin placement isn't proving to a problem for me at this point, and things (qtz) are restoring OK even with different plugin locations (which should be expected). The apps also still play ok, so they definitely can't be linking to my old plugin locations (since they are all different now... only kept one or two plugins in my user folder in the past, and now they are pretty much all residing there).
In the past, I have a memory of some kind of issue between v002 plugins in the library vs. system/library, that goes along with Cybero's observations (I wish I would have stopped myself right then and written down/documented). However, moving everything to the User library has not proven to be a problem.
I am considering making another test build since my update, and plugin relocations, but it doesn't seem like it should make a difference for those whose system it doesn't work on. I am most curious if someone reports that the app doesn't work, but the qtz does.
The plot thickens now that it does work ok on your system Lee! At least it makes it a bit more fun :)
FYI, I probably do drink more coffee than you! I don't even want to publicly disclose the amount of coffee I ingest, lest everyone become upset that my ingestion level has raised global coffee prices.
trash that .plist :-)
Well, what's funny is this... it's not the same problem that fella was describing on the list (where patch creator wasn't working at all).
What is now happening, is actually what I think should happen :) When I would open up QC in the past, patch list would automatically open, and in a couple Spaces scenarios, depending on what Space I opened up a qtz in, it would not be in the correct Space (eg., the same Space as the editor)! Now, it waits until I select it, and pops up in the correct place. So I definitely don't want to trash the plist! It is slick now.
edit.... OK... I went ahead and did all updates and reopened, confirming that nothing in the qtz broke after my changing my own plugin paths. I was almost hoping that something would!
This time I pulled in plugins via Quartz Builder all from the User location, instead of across multiple locations. To be clear, I'm dubious that this would make it work for anyone that it's not currently working for.
I also took away the circle gradient. So if it doesn't load, you would maybe get a translucent colored background, shifting level of opaque every so often.... more obvious if nothing is happening to someone who hasn't seen it.
Yeah about duplicate plugins in different addresses, we ough to just charge them rent :-)
I had to weed out a whole load of stuff when Migration Assistant did its magic last Friday.
By the way anyone noticed the fixed date issue, or is in fact just me ? Friday 15th May sound at all familiar ?
In this respect I am bound to ask if this needs to be addressed.
See attached screen shot for example - taken today at a little after midnight.
Well, hmmm ... with Liquid 1_2.app. I see a more subdued variation of liquid.app. lol. No observable difference when the v002 Optical Flow plugin location is swapped out. Particles are vague-ish but definitely discernible after it runs for several seconds. Is that correct? It loaded, right? My default plugin location is in System, btw.
I have also noticed the date issue. I am turning this into a Quartz Builder debug thread! We should probably post note about that in one of the QB threads.
I have that problem too (though, I guess that I don't particularly care that much about it... I would rather have all plugins load, and even stuff like a way to handle some published inputs way before the date thing) - though you are right, it's not really supposed to be that way.
Yeah, and it should load plugins from the app, so it sounds like it's right.
It could look more subdued! I took out the background, so there is no color shift happening there any more (or less, rather). Since the app is translucent, different desktops can interplay with the perceived colors a bit.
It sounds right. It is very clear, and after an initial flush of color, it will look light hue for a bit, and the colors of the particles start getting more intense at certain points.
If you swapped around the plugin, it reconfirms that it's not linking to the location (seemingly).
It's seemingly GPU dependent, and based on the optical flow working or not in the context of Quartz Builder, on a given system... but if the app doesn't work, and the plugin does on someone's system, that points to the Quartz Builder I guess.
Clear, eh? Nope. It's not clear except for some infrequent moments, and then only partially.
Frankly, this is rather difficult video imagery to be trying to assess verbally, no? I sent a screen grab to Vimeo (for your viewing only. I don't want it posting itself to facebook and god only knows where else) but they say it's a half hour before they'll convert it. This type of circumstance is why I was pushing A/V conferencing over in the welcome thread, eh? I can stream it instantly and anywhere ... but folks aren't there. C'est dommage. ;-)
Edit: it's up: http://vimeo.com/4739770
That REALLY helps. It's not working for you either. But I guess my color change on the clear is really snazzy.
Interestingly, I seem to see some particles in your clip way into it. But barely. There is no transparency (GL Tools).
Lol... I am SOOOOO over this at this point almost :) I feel like I want to load up Windows XP.
Well, there is something very awry, and I feel limited in determining exactly "what". I'm curious if the qtz runs and "looks" correct in any way for anyone that the app isn't running correct for (which is everyone minus me, and a mini with no QC install).
Your note an A/V conferencing is taken as well, I guess I'm still a luddite when it comes to certain things. I have to enter the computer age, hahah...
I'm putting screengrab of how the app works on my machine for posterity.
It does say on the optical flow that it is GPU based... so I'm wondering if perhaps the "throw" of the controls is just wildly different depending on the machine. This might explain it... changing the amount of CI dissolve, or decreasing optical flow sensitivity might make it look the same on a given machine, if that was the case.
As is, I think that the chain that comes from the renderer, to the optical flow stuff, and then the sprite, is just breaking somewhere. From what people are posting, that looks most likely... but it miiiiight be an effect amount issue varying by GPU. I wish I had a couple different builds to try this stuff on to reach these conclusions quicker. Guess that's going to have to happen eventually...
The qtz would look like this if it is working, but anywhere there is clear, there would be alpha black.
OK!
I'm posting these... which hopefully will work on other setups. We will see apparently.
This is built as a desktop level, and should pop up behind your icons, and ignore mouse events. Moving desktop wallpaper basically.
This other thing is a mod version of this:
Both of these will probably flash a 512x384 for a sec, and then "do their thing", popping to full size.
Let me know if any more major issues occur (like not running at all/no images). I really appreciate the help that everyone has given in vetting some of this stuff in the Quartz Builder environment. Hopefully it will be useful for us all in the long run!
These both require Kineme3D.
Aha! Yeah, as you saw .. I don't have any of that transparency.
Thanks for taking note of the argument in support of A/V conferencing. It's like this: [ text query ] "Dude. Go grab my app from this here link and show it to me with a live screen grab on your rig". [ text response ] "On it". [ video response 90 seconds later ] "take a look". [ two minutes later ---> end conference ]. That's just too useful as a tool to go to waste, imho.
Both are good here. Lego is quite impressive on the entire desktop! Woweee!
I'm going to bed. :-)
hairball works brilliantly :-)
lego just hangs :-(
screen size perhaps?
would originating platform really make a difference I wonder seeing as how width and height are set effectively to 'any' in Quartz Builder?
Hmmm. Does it hang when running the same time as the other, or independently?
Some of the 3D apps seem to take how ever long it would normally take to load the model (or a little less than usual actually)... but then have fired off ok. I think I didn't put any default size on the Lego Desktop one, just fullscreen. Hmmmmmmmmmm m m m.
It's the date, It's the date 30/3/09 - or perhaps that isn't all that significant, but lego is not 15/05/09 - it's 30/03/09
re your query about loading I had tried initially loading them together, without success with lego, then quit hairball, quit lego and tried running lego alone, without success.
Just trying the application again - it's taking time to load, so will update this post if it happens before I have to dash to catch a train.
It's the date, It's the date 30/3/09 - or perhaps that isn't all that significant, but lego is not 15/05/09 - it's 30/03/09
re your query about loading I had tried initially loading them together, without success with lego, then quit hairball, quit lego and tried running lego alone, without success.
Just trying the application again - it's taking time to load, so will update this post if it happens before I have to dash to catch a train.
Both of these work fine for me, at decent speed on a 1920x1200 desktop.
One major bit of criticism though: this page is getting WAY too long :D
That it is! It gets hairy to navigate any thread where responses happen above the others... gotta put this one to sleep :o)
I kind of like that, but it makes time organization weird, since you can't look at everything in a linear time line.
Last post from me on this score, having just got back home from work.
I had tried to run lego without the actual plugin required installed.
However, when I downloaded the Texture patch plugin, installed it to my usual home folder, lego bounced up and ran AOK. I am somewhat relieved , yet puzzled.
As we know, that action is not meant to be necessary, at all :-)
Oh and both of the applications can run concurrently without noticeable snarl ups.
It also runs AOK on my PPC G5.
Geez, really? Crap man. Nope, shouldn't be needed.
It is not a good day.
Hi George! To me work both well...
The hair ball is stocked at the begining with a grey window, after 5 sec runs good.
see u!
I've got a QC Farm of about 22 laptops (varying MacBooks and MB Pro's, old style and unibody) that I'm getting ready to test this on. I'll see what I find .
Odd, I've never heard of this happening, nary once. I dont think I am doing anything non standard in my non skank SDK coding that would cause plugins to not load from one location or another. Very very odd.
It's hard for me to say if it's the optical flow, the Quartz Builder, or the combo (in the context of the app).
Cybero's problem was interesting, because it was problematic with the actual qtz (not in app context). If you dl that, you will see that it's based in your optical flow feedback, with extreme minor tweaks on that effect, with a render in image feeding that chain, and then to a sprite or billboard (can't remember what I used offhand).
It sounds like Cybero has both the qtz and the app working, but had a plugin placement issue (with the qtz, not the app).
Others seem to not have the app working, but none have tested to see if the qtz works alone (that I am aware of). So, I don't really have any report yet that pegs it down. If the app doesn't work, and the qtz standalone does, some kind of constructive conclusion might be able to be drawn. At present time, it's very ambiguous. I will be very curious if anyone tries out the qtz and it works, but the app doesn't.
On a separate situation, I was having qtz problem with the difference between the old fbo glitch and the new having being keyed differently, leading connections to break. At first I thought it might be a general problem with the v002, but I see these are two unrelated incidents. I thought perhaps connections for the optical flow were breaking on different machines (which still seems like "maybe" is the case).
Since the source that feeds the optical flow goes through your v002 optical flow chain and then to a sprite... it's the only plugin in the chain of the qtz/app, I am thinking that it's breaking.
Now... I note that the Optical Flow is GPU based... could this have anything to do with the problem? Would settings be really different from GPU to GPU? It doesn't seem like it would impact too much, but something is going awry.
If someone was using a different GPU, it wouldn't break connections, correct? That seems really crazy, and I'm almost sorry to ask that (it seems like a dumb question), but I'm miffed.
I really appreciate all the feedback from everyone a great deal. I would like to be able to know what is ok and what isn't when it comes to the Quartz Builder app. It's hard, because the app works great on my machine.
Addendum.
If I open up the qtz, as is on my machine, after I've stored it... a million and one times, it's perfect.
I downloaded the qtz that is up here, that I uploaded (and looks fine, and I've closed, opened, etc, all fine... and I'm getting just what you guys describe!
In web shorthand, wtflol?!
Ah, love it.
Oh, awesome, I kinda see what the problem is.
Yay. I will be able to update this and get it working most likely, the liquid qtz that is...
I was going to put this into the repository, but I couldn't figure out what section to put it in; none of the groups seemed to fit. "Vid Post" means "post production" to me, not a "post of a quartz composer video", correct?
I didn't want to turn into into a Graphic Animation protocol thing for the sake of shoehorning it, and it's pretty tenuous to posit this as a tutorial. I don't feel comfortable putting it up in a QC Award setting either, since it's a kind of nascent idea.
This is a Quartz Composer piece that's aim was to make use of no plugins, but also to explore different ways of manipulating the particle system using Replicate In Space together with Replicate In Time.
Simple, clear and beautiful. Thank you.
sweet, works directly in the browser too.