] [Thread Prev
RE: Scripting/Command Chaining in the Max/MSP Kaiser Looper
> When you say parallel loop, do you mean a new buffer? When
> Mobius overdubs, isn't it just creating a new buffer along side the
> first one?
Sort of. A layer is basically a buffer, it is the management of
this collection of buffers to make it look to the UI as just one
loop with a scrollable history that is hard.
> I take it it is not replacing the original buffer and creating
> a new one with both old and new material, otherwise you would not
> be able to undo.
In Mobius, the previous layer is continually being copied into the
next layer, with feedback applied, and merged with live audio. So
yes, it is "creating a new one with both old and new material" but it
is also not replacing the previous buffer. This is what gives you a
Unlike phrase samplers that take the "autonomous parallel loops"
approach, there is no limit to the size of the layer history other
than memory. The other problem with parallel loops is that once you
have more than 20 or 30 of them the cost of merging them at runtime
starts to become significant and the CPU spikes.
Layer management in Mobius is actually far more complicated than this
but I won't bore everyone with the details (unless asked :-). It is
relatively simple of you're just doing endless overdubs with feedback,
but when you start throwing in inserts, unrounded multiplies, and
other structural modifications in the middle of the loop it gets, um,
hard to maintain a layer history and do all the flattening that needs
to be done in real time.
Another thing phrase samplers typically have a hard time with is
letting an overdub carry over seamlessly from layer to layer without
introducing a "fade bump" at the edges of each layer. If you
don't fade the edges of the layer, then say reduce the volume one of
layers, there will be a hard vertical wave form jump when the layers
are merged that will sound like clicks. But if you do fade the edges
you don't get a nice seamless overdub.