This earlier suggestion strikes me as predictable, but not really all that practical. Rather than having to specify an exact value for each hit between 1 and 127--a lot of extra trouble that does little to produce a more lifelike percussion track--I would urge you all to consider the virtues of doing more with less.
Have a look, for example, at the interesting French drumming program called Archibald, which substitutes PROBABILITY for MIDI's anally overdetermined numeric precision, and makes it easy to specify graphically a RANGE of possible velocities--also, a range of possible "locations" for each hit, and a range of likelihoods that the hit will actually occur!
That already sounds a heck of a lot more like what most live drummers I've known actually do, especially in repeating or looping patterns, rather than insisting "this particular hit must always be exactly at velocity 73!" Indeed, I'd recommend everyone download the free demo of this program and play with it a little, before we faithful DB users try to push poor Ben any further into Just-Another-MIDI- Sequencer Land.
-Sterling Beckwith
On 07 May 2006, at 02:36, Sterling Beckwith wrote:
This earlier suggestion strikes me as predictable, but not really all that practical. Rather than having to specify an exact value for each hit between 1 and 127--a lot of extra trouble that does little to produce a more lifelike percussion track--I would urge you all to consider the virtues of doing more with less. Have a look, for example, at the interesting French drumming program called Archibald, which substitutes PROBABILITY for MIDI's anally overdetermined numeric precision, and makes it easy to specify graphically a RANGE of possible velocities--also, a range of possible "locations" for each hit, and a range of likelihoods that the hit will actually occur! That already sounds a heck of a lot more like what most live drummers I've known actually do, especially in repeating or looping patterns, rather than insisting "this particular hit must always be exactly at velocity 73!" Indeed, I'd recommend everyone download the free demo of this program and play with it a little, before we faithful DB users try to push poor Ben any further into Just- Another-MIDI-Sequencer Land.
I would actually quite agree, despite having been the one who suggested the 127 point range. (Though I think that would be an attractive set-in-preferences option for folks already familiar with MIDI programming, and probably an easy preference to introduce).
I myself am _not_ familiar with MIDI programming :) and find a 1-127 point range of values for anything, let alone simply velocity, rather counter-intuitive :) I would be much happier using something that functioned along the lines you suggest.
For example, I might have 20 kick samples and 40 snare samples at different velocities (I picked those numbers arbitrarily). It would be nice to be able to define the dbkit such that it was aware that there were a bunch of sample files ranging from soft to hard, so that when one added a kick drum hit in the song editor and set a, say, 10- point velocity scale like we have to "8", on play back DB would be very likely to pick the 16th hardest kick sample, but might pick the 15th or 17th. (Though this should probably be an option, too, since loop-makers might *want* mathematical precision ;) Similarly, setting "8" for the snare might mostly likely get you the 32nd hardest snare hit, but might drift harder towards the 34th hardest snare hit or lower towards, say, the 30th hardest.
Something like hats could get more complicated, since the dbkit would have to be aware you might have multiple ranges of hat samples at both different velocities and different degrees of openness. ... but, hmm, I guess actually that about fits within the current dbkit paradigm, since you would just create openness variants of the drum and define ranges of samples at different velocity for each openness variant. I guess a drummer is much less likely to accidentally change how open the high hat is than to hit it slightly harder or softer :)
Cheers, Carl
-- Carl Edlund Anderson http://www.carlaz.com/