Darwin Grosse

Editor's note: This was the piece that our friend Darwin Grosse was working on at the point of his untimely passing. It's about as perfect an example of Darwin at work as we can imagine, geeking out with a friend about something wonderful done in Max. Our friend Garth Paine wrote to say that he sees this series as a tribute to Darwin and his warm and generous mentorship of the community at large.

Darwin: All right. Let’s take a look at the Cycling Three project, which is the only one that you've done that opens up in presentation mode. And I think if you do, in fact, go look at patching mode, it's because there's a lot of complexity in there. But just even in presentation mode, it's clear there's something going on here that we just simply aren't used to seeing in your everyday Max patch. This pattern mechanism is almost a mini language of some sort. What does that mean and what object is using this stuff?

Garth: The mc.assign object is using the pattern messages. As I’ve done before, I kind of think about this in the modular synth-type environment, where you might have a mixer that's actually panning across multiple inputs and so you could use a control voltage to randomize the feeds that are coming from different modules. You could do that rhythmically. You could do that linearly. You could do that randomly. You could change your time. It seems to be that's how I understand the MC assign object — it's taking a whole bunch of messages there, but it treats them within three separate modes. 

You've got just your normal mode, and then the step mode where it'll step through the instructions that have been sent, or the list iteration mode, which gives you, again, a different kind of output. So you can really switch up the way it's responding. And then, all of these kind of cryptic messages are telling it which inputs to bring up in this case to turn it up. And so, I'm sending it a seven-channel audio file. Well, there's two assign objects running here. One with percussion on it and one with kind of melody stuff that I made in Ableton. And so, it's turning up and down the volume on those seven channels, so we hear different combinations of them.

Oh, okay. Got it. So both of those audio files are seven-channel audio files?

No, the melody one is seven channels. The other one is just - 

Oh, just two channels. Yeah. Okay. I see it. Now, what's interesting is that you didn't really have to create snapshots as you’ve been doing in this series because all of the action in this case just comes from selecting different pattern messages. I think that you've done kind of a great service here just by showing us kind of a breakdown of what the language is with the asterisks, the plus, the up carat, and the dollar sign. Each has their own purpose within the pattern set. I was surprised at the variations that came from this. It was really quite remarkable… when you first look at the patch, it kind of belies the fact that some real complexity can come out of it.

I completely agree with you. And actually, one of the things that I love — even though I've written a few pattern messages that I think are interesting — one of the things that I found even more interesting is the variations on randomness that you can send. The plus pattern alone is going to kind of go through without repeating any, or the asterisks pattern, and so on and so forth. So I actually find myself kind of playing the patch, going between the patterns that I've written and then sending random instructions so that the patch mixes it up itself. I've also linked some of the melodic patterns to the percussion patterns so that when you send some of the melodic patterns, it automatically switches the percussion patterns automatically switch.

And you could have any number of assigned objects here with different content, of course. You could use this not only as a way of creating a performance patch, but you could use it as a way of structuring a whole performance — like a whole gig, a whole set. You'd have a pattern which would be your first section, another pattern for your next one, and so on. You could actually have it as a kind of really complex routing system for a whole set rather than just for a single-performance environment.

Interesting. I've seen a lot of people who do live coding performances, and quite a lot of their work has this kind of nature to it. You have this compact language that represents pattern management and you can kind of type those in and group them and fling them out in real-time.

Yeah, I guess it is a bit like that. Here I chose to use. Going back to the MIDI stuff, I played around a little bit with that and then I decided that I actually wanted to make some sample material as a source that I was manipulating that also all locks temporarily together. I decided that was a more interesting way of going about it in this patch. And I really love the way... I never know what those straight up and down lines are called, but I love the fact that you can combine channels with those-

Oh yeah, the little bar things you mean? I don't know what they're called either.

I'm sure they have a proper name. But I like the fact that you could then just group any set of channels actually. Whatever's between those all get played at the same time.

When you're in step mode, can those be treated as one step? So you could have different groups of combinations all in a row, then?

Exactly. So you see there, I've got one, two, and I've got five, six… I've got a single channel seven. But the ones with the asterisks… If you're in step mode, they'll step through one and then two and then three based on the delay time that you set.

Ah. I Got it. Interesting. This is really quite fascinating. One question I have, though, is this: The mechanism that you're using to adjust the volume — it looks like you're using an mc.adsr~. Is that right? Do these patterns send out a pulse or something? Why are they using ADSRs to control the volume?

The mc.assign object is basically triggering an ADSR and they're both multi-channels, so the ADSR is per channel. And then, the mc.* is basically multiplying that ADSR on the incoming seven-channel file. So the reason that I've got a pattern volume meter there is so that you can actually see the effect of that ADSR bringing up the volume and down again. The timings that are set in the ADSR are the envelope and is what’s generated when that pattern is being triggered.

So none of the audio goes through a sign. A sign is just basically organizing using ADSR.

Right. It basically tells the ADSR which channels to receive messages.

Right. And how fast to turn them up, and how long to leave them up, and so on.

That's much, much clearer. That's awesome. 

You know, I still don't think ADSR makes sense in the way it's set up….

Again, the kind of the development of it kind of goes back to the idea of the original ADSR. It was meant to be sort of like a keyboard-played device. You have an "on" and an "off," and the sustained thing has no duration assigned to it. It's directly related to the key being pressed.

Yeah, so I guess you could look at it another way and say that it's interesting because you can trigger the attack and decay part of the envelope, and then separately trigger a sustain and release. When you leave sustain at zero, it just runs through - 

No, I don't think it does. I think it goes AD, and then your sustained level is zero, so it'll sit at zero then.

Okay. I think it might do the release. I have a sense it just jumps sustained, but as soon as you put a number in sustain, you have to trigger the sustain-release separately.

Right. Yeah. All right. Well, I'll be digging into that a lot more. Thanks so much for providing us with these really cool patches. It was really, really neat to dig into it, and really fun to hear inside the way your brain works.

Learn More: See all the articles in this series  

image1.jpg

Garth Paine and Darwin Grosse sit down to work through a patch that combines the mc.assign object and some clever use of pattern messages. They produce a patch that plays and pan melody and rhythms in a complex routing system for a whole set rather than just for a single-performance environment.

garth-paine-patches-for-max-8-3-part-3

Garth Paine Patches for Max 8.3, Part 3

For the second article in our series on Max 8.3 patching, Garth Paine has created a patch that can claim an ancestry from the astounding player piano work of Conlon Nancarrow and gives a nod to its Black MIDI cousins: Cycling Two Black MIDI train (if you're new to the genre, this article will help you get up to speed)  It's quite remarkable because it brings Max users something they may have not even thought they needed — an audio rate midi player, which is really cool. Buckle up!

Note: this patch also requires the installation of the JC_rev~ object from the PeRColate package, which you can download using Max's Package Manager.


Darwin: Can you explain a little bit what captured your attention about MIDI at audio rates and why you decided to build a performance system out of that tool?

Garth: I really love taking — even in the modular environment — instruments that are really designed to do one thing and then driving them much faster, and seeing how they self-modulate or what other timbres they produce. And the whole no-input mixing desk practice of resonating the actual infrastructure of the device itself is super interesting to me. I see this kind of black MIDI approach as, in the past, being a little clunky, but this player, and I guess with the computational pair that we have now, suddenly opens up this amazing ability to trigger it with the phaser at audio rate. And again comes back to this idea of just getting these incredibly rich timbres.

I started playing with that and I was really interested. You'll see that I'm feeding like 10 notes as a kind of basic structure, or 12 I think, in there. And I was getting these really interesting timbres out of there and started changing those a little bit. And you can see the velocity is driven also by a cycle object, so we're getting a little bit of randomization or variation in the tamp, so it's not too static.

And then, I started thinking, "Okay, if we're playing those, how do we change the distribution of pitch so that we can really bring the timbre alive so it becomes kind of fluid?" It becomes kind of compositional tool. That's why I use the ease object. Basically, the ease object is allowing me to do non-linear transformations of the pitch sets. If you change the ease function, you get completely different sound even with the same pitch set that's coming out because you're redistributing that pitch set.

Right. So, what the ease function kind of does is you have a note that says, "This is the distribution curve for the pitches," but it's kind of acting like a mixer, right? A formula-based mixer?

I guess you could see it that way when the distribution of those pitches is really low, for example. You're right, but it's going back into the note numbers, so you are actually changing the frequencies themselves.

Right. You've kind of got this really nonlinear ability to change the whole timbral makeup of the sound, even with the same note input at audio rate. And then you can move that midpoint, as well. So, in this patch, I started finding that if I just started moving that midpoint, even though it doesn't really show for quite a while in the graph itself, you hear this amazing timbral variation. And I was thinking, "Oh my God, I could write a whole film score just with this patch."

Because it's changing the MIDI, it's actually changing the harmonic content of what's being generated. That's wild.

It's not even shifting at a full MIDI pitch and we're still getting these amazing variations. I found that, again, once we get that main sound content, and then we kind of manipulate that so we can more fluid, it's just the opportunity to vary things that's extraordinary. And then, you could put any VST object you like in there really. At the moment, we are just running a basic, general midi object there and that does a great job.

I loaded up the Arturia DX7 emulator and stuff, and it's different, but it's an incredibly weird sound. What's funny is when I saw what you did, I sort of expected it would sound like running a fork across a rake or something. And sometimes it does. But sometimes it doesn't. Sometimes it has this like real structural chordal sound, and sometimes it has a very ethereal, kind of foggy sound.

Yeah.

Going through the snapshots for this patch was a real treat. Is there something in the snapshots that was particular to how you liked engaging with this program?

In this patch, the snapshots are even more varied than the other patches because of the amount of variation that's possible here. And like you say, in the early snapshots, it's almost Stravinskyesque in the kind of Rite of Spring, like big chord or kind of rhythmic things, right?

But I actually like some of the ones that are more minimalist, and I like the kind of Different Trains reference, which comes back to the famous Steve Reich piece. Because it sounds almost ensemble-like, there's such a richness to the sound that it really sounds like you've got a whole orchestra there or a whole ensemble of instruments there. Iin the latter one, slowed the frequency of triggering down and you start getting this whole kind of unfolding of the sound. So some of the cross rhythms that come out with that are super interesting.

In the "cycling 01" patch you created that we looked at last time, you used generate to sort of create a function that worked across channels. In this case, basically you used the ease function spread across an MC line. Is that what I'm seeing?

That's really interesting. What drew you into this sort of mathematical function use?

That's a very difficult question. I guess I wanted really very fine control. The approach of sweeping the midpoint for instance is set with a step rate of... I'd have to check it, but it's like .00001, right? So it's almost microscopic. It's like watching grass grow, but you can really hear it. I think it's just that level of control that we can get by doing that. I was kind of interested in sweeping as I listen to it... I guess it's informed by really listening to it. As I was sweeping through, I felt like I wanted to go back and have these ranges sweeping through these timbre spaces... almost like ocean waves.

It's so interesting. And it's neat, because you think of these functions as having a high level of precision, but it had a very organic sound to it — especially as I played around with some of the different curve functions and with that varying center point. It was really cool.

Right. Yeah. The ease function itself — because the resolution is really just the 12 pictures that have been put in there — is pretty coarse. But then you have this ability to really tune that in terms of the higher and lower range, so you can bring the whole function down in range, and then sweep that midpoint, it starts getting super interesting.

Learn More: See all the articles in this series 

Garth_EMPACcropped.jpeg

Garth Paine has created a patch that can claim an ancestry from the astounding player piano work of Conlon Nancarrow and give a nod to its Black MIDI cousins. It's quite remarkable because it brings Max users something they may have not even thought they needed — an audio rate midi player.

garth-paine-patches-for-max-8-3-part-2

Garth Paine Patches for Max 8.3, Part 2

To celebrate the release of Max 8.3, we contacted artist/technologist/educator Garth Paine and asked him to review the new features, find some objects that intrigued him, then create example patches that used those objects in an interesting way. In the first example patch, Garth put the new mc.generate~ to the test, then taking advantage of mc.teeth~ for some effects, giving us a curious variety of sounds from a single patch.

We spoke with Garth about this patch, and here’s what he had to say:

Darwin: Hi Garth - thanks for working on this project! Let's start by looking at the "Paine_Generate" patch. It has a sequencer and it's got a bunch of other functional modules, and it can range from a very tight sequenced sound to a very spacey sound. What is the Max tooling that you used in order to make this happen?

Garth: The focus here was really on the mc.generate~ object. One of the things that I hadn't previously gotten my head around was the way that multichannel objects really kind of act as a poly~ object on steroids. So, in this case, it's a way of wrapping up multiple instances of a particular generator.

I really enjoyed the way that the mc.generate~ then would feed any set of generated numbers - in this case frequencies - into mc.cycle~. But I also wanted to have some control over that; that's why I introduced the sequencer into that patch - so that I could set pitch ranges and add a kind of melodic structure to what was going on. That structure really opens up a second option, which is to use the mc.phasor~ to drive the triggering of the mc.generate~  object and to be able to set up deviations around both the triggering and the upper range of the frequency content.

That system generates an amazing range of sounds; depending on the rate of triggering and the deviations that are set up, you really get some extraordinary timbres coming out of there. And then, I put the multichannel mc.teeth~ in there, because I felt like it would be good to take the raw material and just expand it a tiny bit more. mc.teeth~ really allows us to get into these highly modulated comb filtering environments and to create completely crazy feedback.

Right. One of the things that I liked about this patch is the new approach: people typically create cloudy instances of the mc.cycle~ using the deviate messages. But this use of mc.generate~ actually allowed you use a number of different modulation devices, including the sequencer and including a sort of LFO. It feels like controlled-yet-chaotic behavior.

Right, absolutely. I think that's really one of the points of mc.generate~ is that you can have a lot more control, and as you point out, you can have control on multiple levels. So what I tried doing was setting up a kind of backbone with the pitch set that's in the sequencer. And then you can really modulate on top of that, but you've kind of got something to build from.

And the mc objects all the way through that patch just make it so easy to wrap all those things up and send all those channels through to mix down at the end.

Now, this patch contains a number of snapshots. Is there one of those snapshots that represents, for you, the canonical sound of the patch? And what is it about that preset that makes it so interesting to you?

I think, for me, my favorite one is the "Creep Chime," as I called it. It's really slow, but you can really hear the evolving timbres that are in there. I think that's my favorite. And once you've hit that snapshot, you can select different effects presets on the mc.teeth~, which will really change the sound a lot.

If I could select another one, it would be the last one: "Bounce Space." They're both slower, but you really get this sense of the timbre filling into the space.

Actually, I like all the others, too. They kind of demonstrate the control and the rhythmic structure that we get from the sequencer.

This is especially true with the presets named "Nervous" - they really do live up to their names! It's interesting that for doing the effects, you chose to use mc.teeth~. This is not something I've played around with a lot. Is comb filter manipulation a typical tool in your creations?

It's something that I really love, because of the kind of phase correlation that you can get across the comb filter. So like any other tool, I guess, I use it sparingly and when it seems that it's appropriate, but what I think it does here is it just kind of adds a big dose of “fertilizer” to the (sonic) garden!

Suddenly the whole timbral space just kind of blooms into a much bigger space. I think comb filtering is exactly perfect for this kind of thing, because we have so much functionality available. The feedforward and the feedback controls within that comb filter can completely transform sound that's coming out of it.

Garth_Paine_patch.png

Garth Paine dives into some of the new functions of Max 8.3 and provides some playable patches that show off his unique voicing. Part one of three.

Search Results