Do we really need more than 100 velocity layers?

[Rock Hardbuns]

But... is the question whether or not someone can spot the difference really interesting?

EDIT: A snippet from something I did recently. Drums are all single layer. Here

[BFDrummer]

But... is the question whether or not someone can spot the difference really interesting?

Well.... if you're not interested then no.

Some people like the sound of real kits because they sound different every time you hit them (even at the same velocity) and playing them at different velocities also changes the whole tonal characteristic of the drum sound etc not just its level.

BFD is tying to emulate this kind of thing - it's not trying to be 'better' than other drum machines, or kinds of drum sample libraries. IOW there is no inherent 'betterness' to multi layered drum sounds if you're not after the effect this gives you.

it is only trying to be better at recreating a natural acoustic drum sounds.

So if you're trying to make a realistic sounding virtual acoustic kit then yes it's better.

-What's that? .....OK .... yep, no probem...

Well I've just been told this subject is being closed now

[Rock Hardbuns]

Ok, all I meant to say is that wankage has a tendency to get in the way of making music. I did not mean to imply that being a wanker is moraly wrong.

[BFDrummer]

If you really think using BFD makes you a wanker and gets in the way of making music then I guess you're probably best off staying clear of it - in case you become one yourself!

- now your turn! etc etc

Oh KVR, you make us bad people!

Ok seriously Pukeweed, ahem .... cards on the table time .... yes I admit it. I am an acoustic drum wanker, some people are distiortion padal wankers, or guitar amp wankers or retro synth wankers or analogue circuitry wankers etc etc for me it's acoustic drums, always has been always will be...

But the thing is ...I like it! it's fun! it's healthy - and it *is* good for music -ding! dong! hello? -as long as you don't take it too seriously come on! lemme hear those layers say Yo! layers 1- 14 at the back - I can't hear youuuu!

[lion_cub]

If you really think using BFD makes you a wanker and gets in the way of making music then I guess you're probably best off staying clear of it - in case you become one yourself!

- now your turn! etc etc

Oh KVR, you make us bad people!

Ok seriously Pukeweed, ahem .... cards on the table time .... yes I admit it. I am an acoustic drum wanker, some people are distiortion padal wankers, or guitar amp wankers or retro synth wankers or analogue circuitry wankers etc etc for me it's acoustic drums, always has been always will be...

But the thing is ...I like it! it's fun! it's healthy - and it *is* good for music -ding! dong! hello? -as long as you don't take it too seriously come on! lemme hear those layers say Yo! layers 1- 14 at the back - I can't hear youuuu!

[hollowsun]

Ok.... I have held off this one for long enough.

Let's get practical for a momemt.

Most drums played in most musical applications are full on. There may be some dynamics but they are gonna be largely variations between soft, medum and loud.. certainly not 127 levels that most people can distinguish. Mostly (let's be honest) is gonna be the drummer hitting the sh*t out of the thing.

What is more important is not the level at which each drum is hit but where it is hit - a snare full central is very different to a snare towards the rim... but these are not 'velocity' based parameters... this is a 'zonal' thing!

Plus (as has been discussed) there's the fact that there are other physical and acoustic forces going on when a drum/cymbal/percussion instrument is hit that is way beyond just sampling a load of different velocities.

Go for 127 velocities by all means but in practice, with clever and intelligent programming - i.e. use of velocity to amplitude, filter cutoff, sample start, envelope attack, etc. - it is possible to create very realistic and usable results with just a handful of samples.

It's ironic that this thread should start just as another one started about sample library bloatware!

And I don't want to 'diss' the product (I'm sure it is truly excellent - especially if Steve Albini is involved) and more power to you but......

And I am sure that some 'real' drummers are gonna moisten their gussets over this one but let's get real guys - you're gonna load Gb of data and problably only actually play a few Mb of it as you hit full whack (as Ringo and John Bonham et al did) .

I am sorry if I sound a bit down on this one - I am sure it sounds superb but don't go rankling my tail proclaiming !that 127 velocity values is the only solution to realistic drum sampling coz it ain't

Steve

[james0tucson]

Ok.... I have held off this one for long enough.

Let's get practical for a momemt.

There are separate considerations here:

For a track that's being mixed into a song, there may be no problem at all with 3 velocity layers.

For a performer playing an instrument in real time, it will be much more important to have continuous feedback of dynamics.

I'm not saying there needs to be a separate sample for each of the 127 velocities, but more than two or three would be appropriate, if the real instrument has any dynamic range at all.

[Spyro]

45 seems to be a good compromise between quality and filesize

[herodotus]

Ok.... I have held off this one for long enough.

Let's get practical for a momemt.

Most drums played in most musical applications are full on. There may be some dynamics but they are gonna be largely variations between soft, medum and loud.. certainly not 127 levels that most people can distinguish. Mostly (let's be honest) is gonna be the drummer hitting the sh*t out of the thing.

Yes and no. It depends on the style, and the song, and the drummers choice in sticks.

But I just have to take this opportunity (brought about by your very thoughtful post) to get pedantic about drumming. Making a lot of noise with a drum isn't about how hard you hit them, but how you hit them. There is (believe it or not) a technique to making drums sound good, and it has much less to do with proper tuning and such than it has to do with proper playing. This notion of drummers 'pounding away' on the drums is part of a rather common misconception of drummers being kind of stupid semi-musicians (which is true no more often with drummers that it is with any other species of musician).

What is more important is not the level at which each drum is hit but where it is hit - a snare full central is very different to a snare towards the rim... but these are not 'velocity' based parameters... this is a 'zonal' thing!

Yes it is that. And also has to do with the angle of the stroke, and the speed of the stroke, and whether the stick is heavy or light or has nylon or wooden tips, or one is playing with the butt ends. There are also different techniques for playing drums with different sizes, different types of heads, different head tensions, and so forth.

In conclusion, I just have to wonder why so many people are obsessed with doing the almost impossible (i.e imitating a trap kit played live), when currently available sequencers and samplers can do so very many, many, other fun and interesting and humanly impossible things.



But what do I know, I am just a drummer.

[james0tucson]

In conclusion, I just have to wonder why so many people are obsessed with doing the almost impossible (i.e imitating a trap kit played live)

Because you can get so close to a passable result! If I ever create anything worthwhile, believe me, I'll treat the synth drum track as a placeholder, and the whole thing will be re-done with a drummer, a bassist, etc.

Not an option here in my spare-room-office-studio, where there would not be room for a drumkit even if I knew a drummer.

[BFDrummer]

Ok.... I have held off this one for long enough.

Let's get practical for a momemt.

Ok buy BFD - £170 - whack it in the HD - an hour later - the most realistic sounding acoustic drums to date which can be programmed by midi..

So what's all this about 'most' drummers 'most of the time' play full on? and there 'may' be some dynamics but only really 3 kinds!

Blimey! Is there a war on? are we being asked by the government to ration dynamics in our drum recordings? This country needs you - remember wasting dynamics unnecesarily, costs lives! LET'S ALL WRITE UPPER CASE TOO AND SAVE LOWERCASE LETTERS THAT WE DON'T REALLY NEED TO USE.

The fact is enough (I said enough and not all) computers/HD's these days can handle 55GB libraries with 127 layers. That's why some company has gone and made one. If you want to fiddle around with your 'handful of samples' doing 'clever and intelligent programming - i.e. use of velocity to amplitude, filter cutoff, sample start, envelope attack, etc.' then that's fine , you'll save £170 quid and the cost of 55GB storage and probably a couple of GB of RAM as well. I really mean that sincerly.

But it won't sound as realistic as BFD can -BUT THAT IS FINE because you don't always need (OR WANT) total realism for many kinds of music. I use another drum library which only has 1 layer - for what I use it for it's perfect.

But like I said in my first post BFD's aims to recreate as far as possible the situation of a real drum kit in your studio all wonderfully miked up and being played by a real drummer. That is the starting point of the sound. Realism.

What is more important is not the level at which each drum is hit but where it is hit - a snare full central is very different to a snare towards the rim... but these are not 'velocity' based parameters... this is a 'zonal' thing!

Rubbish - how can that be MORE important than the differences between the tonal characteristics of a low velocity hit and a high velocity hit - given that ALL hits are going to be marginally in different zones anyway AND you get rimshot samples too. Make two drum sample instruments, one with zonal layers and no velocity layers and one with velocity layers and no zonal layers and see which works the best.

Plus (as has been discussed) there's the fact that there are other physical and acoustic forces going on when a drum/cymbal/percussion instrument is hit that is way beyond just sampling a load of different velocities.

Yes totally new subject - nothing to do with the title of the thread. Applies to all sample libraries equally, wouldn't you say?

And I don't want to 'diss' the product (I'm sure it is truly excellent - especially if Steve Albini is involved) and more power to you but......

And I am sure that some 'real' drummers are gonna moisten their gussets

I'm exited, yes, but not that exited

over this one but let's get real guys - you're gonna load Gb of data and problably only actually play a few Mb of it as you hit full whack (as Ringo and John Bonham et al did) .

Absolute rubbish again - you are not even thinking about that one, just remembering the obvious fact that JB could play loud when he wanted - just listen to any Led Zep album and you'll hear loads and loads of detailed, expressive - DYNAMIC - playing by John Bonham, all that shuffly stuff on the snare and toms. And all very audible in the mix too. (similar story for Ringo's drums) Yes, some LZ tracks are just total drum twatting but not that many at all - and loads of 12 bar blues with just drums and bass - occassional guitar and vocals so plenty of opportunity to pick up on all those repeated samples in a lesser library. I'd go so far as to say BFD could do a decent version of 'Moby Dick' - minus the gong, timps and hand playing! etc Thanks for giving me the perfect example to illustrate my point further!

I am sorry if I sound a bit down on this one - I am sure it sounds superb but don't go rankling my tail proclaiming !that 127 velocity values is the only solution to realistic drum sampling coz it ain't

Steve

I don't think anyone ever said that. If I said anything it was that 127 is more realistic than 60 or 20 or 3 (just as a real drum kit is more realistic than BFD) and that is what BFD is about - realism.

I think the question which titles this thread is a dumb one anyway.... 'Do we really need more than 100 velocity layers?'

Who is 'we'? We're all making different kinds of music. There is no 'we'.

'really need' - what does that mean? need for what? what style of music?

No one is forcing these sample on anyone - you even get the choice of 3 different instalations on your HD. lite/medium/the full set of samples. And BFD allows you to limit the max no. of layers inside the software.

Like I said in my first post 'Horses for Courses'.

or 'Different Drums for Different Ones"

[Max el Belga]

I don´t need multi layered drums: I only need good samples.

[SKoT_FX]

A little input from FX in response to several points raised in this thread:

* The "Bloat Factor" is user-controllable - BFD Deluxe's install can be tailored to 32 ("BFD" detail), 64 ("XFL" detail) and 127 (Deluxe) velocity layers as a maximum to suit your resources. Until we get 64-bit BFD finished, I personally will probably only use 64 "at home".

* Cross fading between velocity layers introduces phasing / flanging effects, and isn't really viable. Even if you phase-align the fundamental frequency of the two drum samples, there are non-harmonic modes of vibration that are unlikely to match up between the two samples.

* Velocity to Amplitude scaling, Velocity to Damping, Velocity to Pitch are all in BFD v1.5, which is required to run BFD Deluxe. There are many other "shaping" aspects of BFD's sample playback algorithms hidden from the user which introduce more subtle variations on the samples - for example, the new LP-enveloping damping algoithms in v1.5 are deployed in a number of mutual choking situations, and repeated cymbal hits during a swell have their attack env time lengthened slightly to take the "initial impact" of a non-vibrating cymbal away slightly, giving rise to a more natural swell sound.

* Some drummers and composers like having a full range of dynamics to be able to vary between subtle pitter patters upto outright smacks. As BFD determines its audio content at runtime, you can remove the lower velocity layers from your filesystem if you only need loud to very loud. However, we observe every real drummer that visits our stand at trade shows invariably tries pitter pattering to test the more subtle layers, then moves up thru to the whacks.

* Playing complex patterns in BFD can actually be lighter on the your resources than slower ones, as there are more opportunities for mutual choking before disk streaming has to kick in to finish the long tail of the sample.

* Steve Albini was responsible for drum tuning, mic selction aqnd placement, and overall engineering, the actual production: slicing, dicing, velocity ordering and BFD-er-izing was done by our long time audio partner Steve Duda (whose ears may be relatively "fresher" for sorting sounds ).

* Remember BFD will give you a range of hit positions as well as velocity depth. One future update we are sitting on (because it requires updating the BFD engine first) is an additional snare hit type in the "halfway" position between center (BFD: "hit") and the rim. We hope to make this available in the future.

* If you think 127 is excessive, for the Bison snare in BFD Deluxe, Steve Duda managed to extract and sort 360 useful hits from the Electrical Audio sessions, and delivered them to FX (BFD's engine will handle any number of hits - we deal with velocity as a "real" number between 0 and 1, and the various humanization and anti-machine algorithms provide scope for hitting any number of subtle degrees betwen 0 and 1). We anticipated the reaction on KVR about bloat, and had concerns about computer resources and even just shipping the data, so we set the ceiling at 127.

* Deluxe cymbals typically have 40 or so layers. The much longer tails that cymbals require place practical limits on storage resources and recording time.

* BFD allows you to tailor the number of velocity layers it loads. Deluxe gives you the sound of a bunch more kits recorded in a smaller tighter room by an engineer with a long history and reputation for great up-front drum sounds. To what degree you want to work with the data's subtlety is up to you.

[Angus_FX]

In the way of truly lifelike drums , ten years from now it's all gonna be physicall modelled and we'll be laughing at the internet archives

Doubtful. Physical modelling hasn't really come on as well as people hoped (look at the number of people still clamouring for a VL-1 VSTi.. that's a ten-year-old implementation based on 25-year-old algorithms). My guess would be we'll see men return to the Moon -- possibly even men on Mars -- before we hear a decent pure-modelling cymbal with convincing OH and ambient channels.

That said, the approach of using a combination of samples to provide most of the character of the sound, and "clever" performance modelling or a thin layer of (re)synthesis to fill in the gaps, is certainly one that works well. As Skot explained above, BFD is starting to do that with the damping, cymbal choking, hi hat control, mic positioning etc.

[waiting man]

Most drums played in most musical applications are full on

turn that into something between "most musical applications are ignorant shite" & "the world is how you make it".