Two problems help please


    Oct 18 2011 | 1:14 pm
    Hi,
    As the title implies I have two problems. Firstly my "round" object is not being recognised any more. I dont have any other object thats named the same and have even tried deleting the object from the externals folder and pasting a new one in.... I think i have to reset max(by deleting .plist and something else but i cant quite remember what you have to do....) any help would be greatly appreciated.
    Secondly I have been trying to create a number of sample accurate patches. The problem here is that when using a buffer~ to store patterns-as per the "mod squad" or "phasor step sequencer examples"-I get clicks and pops each time I change the pattern (I'm using a hanning window at the start and end of each division so its not that). The pops go away if I change the signal vector size or the I/O vector size after I've changed the pattern but that seems to just be resetting something because there is no particular setting of either that fixes the problem (the poppin only stops if one of them is changed after the pattern has been).
    Thanks in advance Simon

    • Oct 18 2011 | 1:26 pm
      If you're sure there aren't any badly-named abstractions in your search path, I dunno what's up with [round]...
      Re: Clicking
      Are you changing the pattern by changing the buffer~ that you reference with index~ or whatever? And what are you windowing exactly? It sounds unneccessary.
      If you're not doing so already, stick all your patterns in one long buffer, and jump between them by offsetting the 'read/write heads' of your sequencer.
      If you want to change patterns live, and keep it all nice and tight, you can do something like this:
      [number] 
      Any use?
    • Oct 19 2011 | 12:18 pm
      Yea i'm changing the buffer directly. Not sure I fully follow your explanation but i'll check it out. The only problem I see is that i'll always be limited to a certain number of patterns this way. Not the biggest problem in the world i know but I like the thought of being able too have as many as I like. As for the windowing I'm applying it to the beginning and end of each division of the sample, without it there are definite clicks that have nothing to do with the problem I mentioned-they're just a result of dividing the sample into 'x' number of pieces without worrying about zero crossings. I'll attach the patch that I was havin the problem with if that would make any difference. I ended up just implementing a similar(obviously not sample accurate) thing using Max objects but I'd like to solve the problem because I dont think it really has to do with the patch itself-as I said after i have changed the pattern and then changed one of the DSP settings the clicks stop so it seems to be something within Max... Anyway thanks alot for your help man
      PS If you do look at the patch all of the scripting stuff is pretty obsolete (I just wanted to see if it was easier but ended up going back to just sending things presentation 1/0 messages) so you dont need to worry about it.
    • Oct 19 2011 | 12:41 pm
      No Max at work, but I'll check it out this evening.
      I see what you mean about windowing now; I thought you had a general-purpose click sequencer, rather than a sample slicer-type deal.
      But you're going to have to switch to one giant buffer to switch patterns cleanly. The message you send to change buffers doesn't happen at audio rate, so the switch won't happen at exactly the moment when your window is at zero. Also, depending on how you use your pattern data further down the line it could cause you more problems.
      You may think you have "infinite patterns" but you'll run out of memory at some point anyway. I have a sequencer patch with 64 steps per pattern, 64 tracks, and 8 parameters per track.
      64 steps * 64 tracks * 8 params * 4 bytes = 128kB per pattern.
      I wanted 4096 patterns, but realised this would consume 512 megabytes, so I had to cut that requirement down!
      But if you just have one track and, say, 4 channels, 4096 patterns would only take 4 megs. Do you really need more than 4096 patterns, even for an entire album? Or are you making some sort of ADHD hyper-edited hand-sequenced glitchcore? ;)
    • Oct 19 2011 | 1:22 pm
      Na I was fully aware that an infinite number of patterns is impossible. I suppose my concern is more to do with the clicking itself rather than the number of patterns which are possible, also I want to be able to alter patterns on the fly so thats another reason why I'd like to sort the clicking out. I'll check out the suggestions you made though cos the reason I started making it in the first place was to try get away from doing roundabout solutions in Max rather than sample accurate MSP patches.
      Cheers
      PS dont mind glitchcore.... bit too lazy to make it though....
    • Oct 19 2011 | 8:25 pm
      I must confess, I haven't a bloody clue what's going on in your patch!
      It must make sense to you, but it seems like a very unintuitive sequencer to me. Perhaps someone else could shed some light on this one?
    • Oct 20 2011 | 1:39 pm
      Ha! Well i actually intended for there to be multiple instances of the same patch I loaded up along with a number of different sections. It is basically my attempt to create something similar to a kaoss pad... although i've never actually used one so its just from watching youtube videos n reading a couple of reviews(not the best way of going about things I know). Basically the pattern will always play left to right. The number in the top left corner is the number of divisions of the sample which you load and each toggle represents one of these divisions. If a toggle is checked it will be played, if not it wont. I havent given any other options of rearranging the sample yet so its kind of basic in that sense. I didnt get any further cos i pretty much instantly arrived at this problem Sorry should have made that a bit clearer.
      Ive attached another version with a few more comments-its basically just the "mod squad" example file though with a couple of extras so as you arent always stuck with 16 divisions and 16 steps. Again though I'm not convinced it has anything to do with the patch itself-I've checked quite a few different things and it seems to be something with MSP, otherwise I dont understand why changing the either of the vector sizes would fix the problem but then it would return once the pattern was changed again.
    • Oct 20 2011 | 2:59 pm
      I have a hard time understanding it, rhythymically. Every time you switch a toggle on/off, the number of "steps" changes... seems odd to me.
      There's an awful lot of complexity in that patch. You obviously know your way around Max, but it looks awfully inefficient. Can you describe in words what you want your sequencer to do? I'll try and knock up something simpler.
    • Oct 20 2011 | 4:35 pm
      Hey, here's a patch that uses mxj buf.Op and polybuffer~ to create an arbitrary number of empty buffers which you can write into and then switch between.
      This works with your patch by just opening it alongside because the buf.Op object references your #0-steps buffer.
      You should be able see the waveform~ in your presentation mode changing when you change the second umenu.
    • Oct 20 2011 | 5:55 pm
      Here's a patch a bit like "mod squad". It uses a slightly more elaborate step counter, which seems like overkill at first, but it's a flexible system to build upon.
      For example, by [bitand~]ing the per-step clicks with the gate output, you can extract click triggers for each active step. You can also use [
      Max Patcher
      In Max, select New From Clipboard.
      N.B. I only have the Max 6 beta on my work machine, and no speakers, so who knows if Max 5 will open this, or if it even works at all! I'll be home to make sure it actually works in a half hour!!!
    • Oct 20 2011 | 6:36 pm
      D'oh!
      I understand the purpose of your patch now! It's pretty cool; especially for remangling things into strange new time signatures. A whole bank of these would make for fun polyrhythmic remixing :)
      I reckon it ought to be do-able with far fewer objects - scripting the creation/deletion of those toggles is definitely overkill; you could use a [matrixctrl] and just change the number of columns, for starters. I'll try and re-build it without clicks (probably quicker for me than reverse-engineering your patch). Might not get round to it tonight tho.
    • Oct 21 2011 | 1:52 pm
      @davidsmith Cheers man i'll have a look. I generally try and stay away from externals though... I've had a bit of trouble with them crashing on me... also just havin a quick look at it now it seems that it will solve the problem of getting clicks etc while changing the pattern but not sure if it will solve the issue of the clicks that i'm getting while the pattern is actually playing. As I said changing either of the vector sizes(or even just reselecting the current setting) eliminates the clicks and the only return if i change the pattern again.
      @christripledot Yea I got rid of the scripting in the second version i uploaded. If you look at that one i'm just sending presentation messages to each of the toggles. As i said I have actually implemented the same thing using metro's etc. I created a similar patch to the "mod squad" example using multisliders recently and basically did exactly what you were sayin about just changing the length depending on the number of steps. I know doin that would save havin to send all the presentation messages to the toggles but because they're only on/off I just figured toggles would make a neater user interface. As for the actual section which makes the pattern if you can come up with a more concise way of doing it that'd be great. Bar combining some of the maths objects into "expr" objects I dunno how you'd be able to do it while still retaining the ability to change the length of the pattern so easily.
      As i said though-considering the fact that the clicks disappear once i alter either of the vector sizes (actually just creating/deleting any MSP object does the same thing) I dont see how it could be something to do with my patch. I'm more than happy to be proved wrong but it just doesnt seem to make any sense if its not an issue i'm having with MSP itself....
      PS thanks for takin the time to try help me with this guys!
    • Oct 21 2011 | 3:07 pm
      Hey - no problem re. the time... sorry I can't help with the clicks. I couldn't detect any clicks when I opened your patch... other than audio clicks due to the waveform being sliced.
      Not sure what you meant exactly by avoiding externals, but both the mxj buf.Op and polybuffer~ are a part of the standard cycling74 distribution. Although polybuffer~ seems to be a new addition.
      cheers!
      d.
    • Oct 25 2011 | 12:42 pm
      Yea thats what i thought man, dunno what the deal with the clicks is so.... might have to try reinstall everythin n see if that solves it. Otherwise i'm just gonna have to stick with the Max scheduler...
      As for the externals I was referring to polybuffer... I definitely dont have it so I figured that it was an external. I am still on Max 5.1.0 though so maybe its been added since then.