(the product formerly known as) Epoch

Official support for: loomer.co.uk
Locked New Topic
RELATED
PRODUCTS

Post

I think I would prefer using the two different colors and when selecting a chord you make it discernible from the rest. Easy to see where it starts and ends. Why not Bigger/other color/flashing(joking of course).

Post

I did experiment with rounder modules, but in the end favoured the rectangular style. The modules are aligned on a grid, making it easy to sit them together tightly, and the closely packed shapes seem both aesthetically pleasing, and a group way of grouping closely related modules together. Contrary to the few contrived teasers I posted, most patches aren't that messy: it's easy to make them look tidy, and with the ability to macro a collection of modules together, things tend to stay pretty regimented.

On saying that, feedback on the routing UI will be taken on board once during beta: I'm absolutely not locked into a single style (but I'm also hoping most people will, after playing with it for a good while, agree that the rectangles work!)

And as a little progress update, today I'm evaluating a few new random generation functions. I found some literature on generating a variety of coloured noises from a single exponent parameter. It generates nice landscapes a la Minecraft, and I'm hoping it'll make good MIDI data too.
Architect, the modular MIDI toolkit, beta now available for macOS, Windows, and Linux.

Post

colin@loomer wrote:the ability to macro a collection of modules together
Hey, you just answered the question I was going to ask. Macro capabilities! Great, just what I was hoping for! But if this plugin will answer my questions before I ask them, then I think you have a surefire winner on your hands :tu:
Thanks for the updates btw, they're a really good insight into what sort of beast this is going to be...
Textur for ACE
After Hours for Lounge Lizard EP-4
Prism Sticks for Chromaphone 2

Post

colin@loomer wrote:
And as a little progress update, today I'm evaluating a few new random generation functions. I found some literature on generating a variety of coloured noises from a single exponent parameter. It generates nice landscapes a la Minecraft, and I'm hoping it'll make good MIDI data too.
I agree with ThomasHelzle re the cables so it is great to have the choice. Round v rectangular seems to me primarily aesthetic with a slight bias favouring the rectangular (which I favour) for the reasons Colin mentions.

@Colin - I would appreciate a link to that literature. As I have mentioned before I am a fan of the beta distribution but would like to see some info on other options

Post

I personally love the pseudo-fractal noises used in 3D applications which are mostly based on Perlin noise. They create very natural structures that make sense for midi as well (to my ears that is). Easily being able to define the complexity by increasing or decreasing the octaves and combining them with other stuff from the 3D-imagery world like Bias and Gain, mirroring etc. makes them quite versatile.
Also interesting: since they are based on fixed tables, they can be made repeatable, so a generative piece can sound the same the next time it's played on demand. They also have the capability to be loopable.
One of my favorite topics actually ;-)

Cheers,

Tom
"Out beyond the ideas of wrongdoing and rightdoing, there is a field. I’ll meet you there." · Rumi
UrbanFlow.art · Instagram · YouTube

Post

Tom wrote:
a generative piece can sound the same the next time it's played

Sorry, but this makes no sense for me, this sounds for me like play a record again and again.
In my eyes a generative piece should not sound the same as before, more like a on going composition were the basic elements are to recognize and there are a endless numbers of variations to explore.
Personally like generative pieces were i could controll the distances of this variations from the original in realtime with the possibility to come back to the starting point in time.

Post

ThomasHelzle wrote:I personally love the pseudo-fractal noises used in 3D applications which are mostly based on Perlin noise. They create very natural structures that make sense for midi as well (to my ears that is). Easily being able to define the complexity by increasing or decreasing the octaves and combining them with other stuff from the 3D-imagery world like Bias and Gain, mirroring etc. makes them quite versatile.
Also interesting: since they are based on fixed tables, they can be made repeatable, so a generative piece can sound the same the next time it's played on demand. They also have the capability to be loopable.
One of my favorite topics actually ;-)

Cheers,

Tom
thanks Tom - I have not explored Perlin noise but will now - looks like there is plenty of code available for Matlab, which is where I do my midi programming.

Post

jowen wrote:Tom wrote:
a generative piece can sound the same the next time it's played

Sorry, but this makes no sense for me, this sounds for me like play a record again and again.
In my eyes a generative piece should not sound the same as before, more like a on going composition were the basic elements are to recognize and there are a endless numbers of variations to explore.
Personally like generative pieces were i could controll the distances of this variations from the original in realtime with the possibility to come back to the starting point in time.
It is totally up to you how you use it. That is why I wrote "can sound the same".

Perlin noise is (can be) pseudo-fractal and pseudo-random. In 3D graphics, it usually is based on a relatively small fixed table of random integers. The human eye isn't able to recognize repeating patterns when the period is long enough and the patterns aren't too strong, so for 3D shaders (where the only point is, that one can't see repetition) it's more important that it's fast than if it's really endlessly non-repeating.
The ear is more sensitive there, but if the patterns aren't very recognizable or when several such noises at different speeds are overlapping their influence, you get something that is as close to "random" as the brain can discern.

And there are many different ways of working in this realm. For instance: if you use such a noise to create notes, you can get the same notes each time you play the piece. You then can pick out a part that sounds good to your ear and work with that. Or if the implementation allows, change the random seed until you hear something you like and then work with or from that.

Perlin noise can be made to have any period you like or need, so you can have defined loop lengths and then animate an additional axis of it to create as much variation as you need but come back to where you started anytime you wish to.

The original form has only hard jumps for the number of octaves, but in my own shaders and tools I implemented them in a way that they can be smoothly raised, so you can for instance start with a very low-variation form, then increase the complexity and end with lowering it again...

I personally think that the most interesting area lies in the realm between complete randomness and complete order.
Total randomness is meaningless and boring to me.
Total order is meaningless and boring to me.
But take a forest. Neither random nor ordered but so beautiful. :-)

Bazon Brock once put it in brilliant words, it went something like that: "Data" as ones and zeroes has no meaning until you put it in formation. Then it becomes Information."

The fascinating thing for me when working with generative things is, that something emerges I may not have planned or foreseen and I knead it and form it so that it has meaning to me, tickles my ears in the right way or is otherwise interesting.

But hey, that's just me ;-)

BTW. I wrote a modulation source for Usine quite a while ago that uses some of these principles and is a lot of fun to work with: http://www.design4audio.de/home/aon-random/

Cheers,

Tom
"Out beyond the ideas of wrongdoing and rightdoing, there is a field. I’ll meet you there." · Rumi
UrbanFlow.art · Instagram · YouTube

Post

Tom wrote:
The fascinating thing for me when working with generative things is, that something emerges I may not have planned or foreseen and I knead it and form it so that it has meaning to me, tickles my ears in the right way or is otherwise interesting.

This is how it should be used!

So my experimentations always have a recorder running in background to keep the magic notes or events when they happen. Afterwards the work could go further from there.
And you have to be patient, often it starts very confused but leads into something very interesting some time later.
Did you see the Numerology projects from jue at the Forum there?
He shared some so called Discrete Sequencer Projects were the note repetition has very large cycles.

Post

Ah Woggle, I think it was you who pointed me towards the Beta distribution in the first instance; thanks: it is now my distribution of choice in most musical applications. As requested, I included a nice, quick, Beta Distribution module.

The current random source modules include:

- Uniform random (Mersenne Twister 19337) [0 .. 1)
- Normal / Guassian distribution
- Closed uniform integer random [n .. m]
- Beta Distribution
- Bernoulli Distribution

For those who aren't up on the terminology of probability, these are all different ways of selecting from a range of random numbers. Sometimes, you'll want each number to be as likely as each other: this is called a uniform distribution. Think of rolling a perfect dice, or flipping an ideal coin. More often, particularly when generating music, you'll want certain numbers to appear more often than others - in the case of generating pitches, we probably want the root note to appear more time than the sixth. This is called a non-uniform distribution.

Bernoulli Distribution is probably the easiest to understand. It has a single parameter, P, and generates either False of True values. When P = 0.5, False and True appear in equal rations. When P = 0.75, we are 3 times more likely to generates True than False. At P = 0.1, False is 9 times as likely to appear.

How could this be applied musically? Let us pick a value from the Bernoulli Distribution, one value for every 1/8 note, by connecting it to a Sample and Hold module clocked to a 1/8 beat source. This value can then be used as a trigger for a Pack Note module. So, at every 1/8th note, we have a certain chance to generate a True value, which can then be used to trigger a MIDI note. You may then connect a CV Sequencer's output to the P input of the Bernoulli Distribution module, allowing a different chance per step. Maybe we always want a note on step 1, so the first step of the CV Sequencer will be 1. The second note, perhaps we only want a note here every 1 in 4 times, so we set it to 0.25. And so on. (Continuing this, you'd most likely then convert all this into a macro, with a single input for the Pack Note pitch, and a single MIDI output. You can then duplicate several instances, each producing a different MIDI note as governed by the macro input, and each with it's own CV pattern. Connect all the outputs to a drum machine, sit back, and enjoy.)

@Woggle RE noise function literature, my starting point was the chapter on Brownian Fractals in Game Programming Gems 2, and then plenty of internet searching. When I have some time to spare, I'd like to collate this all into an article and some corresponding source code.
Architect, the modular MIDI toolkit, beta now available for macOS, Windows, and Linux.

Post

2 weeks until release! How exciting that early 2015 is coming to an end! :wink: :wink:

Post

You're not far off: I'm fast running out of bugs to fix.
Architect, the modular MIDI toolkit, beta now available for macOS, Windows, and Linux.

Post

:tu:
"Out beyond the ideas of wrongdoing and rightdoing, there is a field. I’ll meet you there." · Rumi
UrbanFlow.art · Instagram · YouTube

Post

I'm in favor of confetti mode. With only two signal types I find it more helpful than color-coding cables by type (could do as Max/MSP has and put zebra stripes on one of them - I would suggest the MIDI cables, as they're more discrete and likely to be fewer in number as well). I've used 3D software as well as other node-based UIs and what works in one context doesn't always translate well to another. Eventually is closer to a modular synth than a 3D renderer, and if you look at modular synths (or soft synths with cables such as Ace or Bazille) you'll see confetti used there too.

The confetti takes a little getting used to but I think it's mostly a case of unfamiliarity - in the example of the MIDI delay feedback above it took me a moment to understand where the feedback was targeted, and it's much clearer with confetti cables. I suspect, based on my experience sequencing with Max, that the MIDI data type is most likely going to appear primarily at the inputs and outputs, while the bulk of the processing in the middle will be done by CV. In most cases I expect it will make sense to unpack the MIDI immediately and not repack until just before it goes out, so aside from a single pass-thru cable there's not likely to be much mixing of signal types.

At any rate, making it a user option keeps everyone happy. :clap:

Post

I do prefer slightly rounded corners on nodes, so that the boundaries from one node to another are clear. Perhaps that might be a user preference as well.

Locked

Return to “Loomer”