*** EDIT: Post #7, below has the solution attached ***

hi all,

I'm trying to make an abstraction that makes improvements on [grid], namely the ability to record and playback motion.

Perhaps there is a standard way of doing this in programming science. But I've got round it by using arrays that record vales and an array that records [timer] values, for playback.

Would somebody, (for the sake of the pd community) go through the patch, it is neatly written, and try and stump out why, playback doesn't seem to be accurate. it gets worse the faster a motion you record into it. Leaving me to think that maybe it has something to do with the limitation of the control rate, and the frequency of information that is trying to be recorded.

How can I get around this. Let's talk about it... It would certainly be a handy feature to have!

patch attached.

http://www.pdpatchrepo.info/hurleur/grid+.pd

Cool idea! Here's what I found:

I noticed that it played back too fast, and the problem seemed to be with [alternate]. I replaced it with [t b b], and that did the trick.

The playback stopped prematurely because of the way you were determining how it should end by counting the number of zeros in the timer array. I see where you're going with that, since the end of the arrays should be filled with zeroes. The problem, though, is that it does actually record a few zeros. I guess [grid] can be that fast. So it causes the playback to end prematurely. I fixed this by just getting the list length of the timer list and using a [sel] to stop playback once the counter reaches that amount.

Hopefully the patch makes it clearer.

http://www.pdpatchrepo.info/hurleur/grid+.mods.pd

Nice one. Reminded me of this : http://puredata.hurleur.com/sujet-566-value-recorder

here's my 2 cents. time works well, could be a bit more precise, perhaps.

http://www.pdpatchrepo.info/hurleur/grid+auto.sunjification.pd

Good call on using [qlist], sunji. Makes the whole patch simpler.

nice
will you make an abstraction out of it or do i have to do it?

Great input people. Thank you!

[qlist] seems most CPU efficient and is so easy to implement into other areas of a patch, so it has to win here for automation.

Sunji, the double timer works great. But I can't work out why it's working better than just using one, can you explain?

Here is the patch, in abstraction form, code-neatened up and only starts recording on first movement (so no un-useful pauses at the beginning).

You can use multiple instances, but saving/loading will only work for an instance with the same initial argument.

(note to mod, should this thread be moved to abstractions?)

http://www.pdpatchrepo.info/hurleur/grid++.pd

arif, regarding your [timer] question: I've always done it that way, but now that I think about it, i developed that method long before I had a better handle on [trigger]. I'm pretty sure 1 gets you by just fine.

hey what does [onebang] do? what library is it in?

onebang is from cyclone. it will only let through one bang in the left per bang of the right inlet, and so on.

I used [coll] to record my grid data.

http://www.pdpatchrepo.info/hurleur/grid.ricky.zip