< Back to IRCAM Forum

Camu selection strategies

Chris Trapani asked this great question on the mubu group, which I’m transferring here:

This is a very broad and conceptual question, but even as an active participant in development I don’t have a clear answer, so I will pose it as an outsider:

I am trying to construct a patch which toggles between three types of selection in CaMu

1.) selection by setting 1-3 descriptors and sending values via line to construct a select message (1-3 values)
2.) imubu selection using the mouse, x/y axes as descriptors (2 values)
3.) mosaicing with a set of up to 9 descriptors (getting live values via pipo~) (9 values)

I would also like to have a multi slider of weights that is automatically reset when new “columns” are selected, and that can affect any of these selection types.

Given that we have multiple objects capable of target selection (camu.target / camu.select / mubu.knn), what would be the best strategy of routing to a single select object? I figured camu.select, but a note inside the object says that in the preferred patching, camu.imubu.control sends descriptor choices instead…

I am certain multiple configurations are possible but am interested in others’ thoughts on what the optimal chain would be. Thanks!

camu.select accepts all three on the left inlet: Selection (with any number of dimensions) can be controlled via the columns / select messages, or via weights / list of target values. You’d want to insert getweights after each change in weights to see what’s going on, like in tutorial 6 live control. Another aspect here is scaling: imubu and pipo~ will need scaling off, the line scaling minmax, presumably.

Don’t hesitate to start a tutorial stub with your use case for me to complete.
Best!

1 Like

Thanks, Diemo. Here I’ve attached an adapted tutorial patch I’m having trouble with and a short video to demonstrate an issue I’m having when I pilot selection using columns (sending “columns 1 6” to camu.select) here vs. when I send a weights message ("weights 0. 1. 0. 0. 0. 0. 1. 0. etc.). If I understand correctly the results should be the same, but they are not; the columns message seems to get the right selection, but not weights.

Not sure, furthermore, if I have the optimal setup re: camu.select vs. camu.mubu.control, so that selection is monitored in the imubu without affecting selection also.

Thanks!

endless-melody-v2.maxpat (913.6 KB)

Hi Chris, I think there’s a column mismatch/shift somewhere in the patch. Sometimes it uses columns 1 5 6, sometimes 0 4 5 (0 is duration). It helps to add getweights everywhere they are changed (and gettargetminmax, for that matter). I also replaced the ages old [p target] by camu.imubu.control and added the random selection visualization.
endless-melody-v2.maxpat (873.8 KB)