IK Multimedia releases MODO DRUM

[lasteno]

Yes, we need a Demo for the HiHats¡

[twitewhite]

"combination of modal synthesis and advanced sampling"

Hmm. ..not really interested anymore.

Only sampled portions are cymbals. All drums in MODO DRUM are physically modeled.

That's a disappointment, as generally everything but the cymbals are good enough sampled. It's the cymbals/hi-hats and specifically how they interact with previous hits that usually gives away the fact that it's a sample, not the real drum sound.

BFD3 has modeled (no one has defined this and we've beat it around a bit here) cymbal swell for some years now, and Superior Drums has something as well. Your point is pretty well mooted by this at least in the BFD3 usage I'm familiar with for a few years now.

Modeling cymbals themselves with no sampling would be quite suspect to me: I implied earlier that it's a combo of modeling and samples that is required.
I'm kind of interested now.

I'm not sure if you specifically looked in my history to see that I use and am a big fan of BFD, and in particular the way its cymbals work. Yet, I haven't seen any indication that MODO Drum's cymbals are anything but sampling. I know it says "combination of modeling and sampling" but then other posts in this thread seem to imply the modeling is the drums and the sampling is the cymbals. Correct me if I'm wrong, IK Multimedia.

If the cymbals are samples with some modeling applied, that would make me more interested. But really I was hoping for everything modeled (guess that's impossible to wish for in 2019, maybe in the near future). But still, unlike with basses, in which there was no real competition (Trillian was years old and nothing else was close), for drums there's BFD, Superior, and quite a few others that MODO has to be better than. For me personally, it has to beat BFD.

I won't be preordering, but I will likely try it out once the demo is ready.

[jancivil]

I wouldn't be moved by 'modeled cymbals' in and of itself. samples with some modeling, why not. But I am seeing 'drums modeled, cymbals are sampled'.
I'm not hurting for drums, but I find the modeled shells of interest.

[plunk&boomaudio]

I am very intrigued by this! Any idea of what the CPU hit would be?

[drvc]

Very interested in buying this - but would need to know more about the GROOVES section first. What tools will there be to find the right groove/MIDI pattern for my projects?

- A "tap to find" feature like in Superior Drummer?
- A "grid search" like in Addictive Drums?
- Something else?

At the moment, the screenshots on the MODO DRUM page only show a simple filter based on genre, leading hand, time signature, but nothing more precise...

[Peter - IK Multimedia]

I am very intrigued by this! Any idea of what the CPU hit would be?

We currently have Specs posted but I do not have specific real-world CPU data as we have not released the final production version.

[Peter - IK Multimedia]

Very interested in buying this - but would need to know more about the GROOVES section first. What tools will there be to find the right groove/MIDI pattern for my projects?

- A "tap to find" feature like in Superior Drummer?
- A "grid search" like in Addictive Drums?
- Something else?

At the moment, the screenshots on the MODO DRUM page only show a simple filter based on genre, leading hand, time signature, but nothing more precise...

I've relayed this to the product manager and developers as a good idea for showing more detail in a video to better construe the Grooves section's functionality.

[xoxos]

circularmodes1.png

i was just going back over circular modes when i saw that you're doing this. watched a video, recorded the audio and had a look at your snare.. wouldn't have been able to tell it wasn't a sample from looking at th wav and spectrum, i've been looking at snare wavs analytically for a fair time.

been watching some videos on drummers setting up their snare, how the choice of snare wire affects the sustain, different effects you can get eg. by using a wide wire and tightening laterally to the wire. worth doing for any computer musician who likes to work emulatively, understanding the variables. from some of the videos i watched, even some drummers aren't clued into how some parts can be tuned. we expect perfection from a key.

[lasteno]

so .. ModoDrum has only Tombs and Snares Modeling ? no Hi Hats and Cymbals Modeling ? half of the idea Done Then ?

[benjamind]

Everything except the cymbals is entirely modeled. I believe the cymbals part uses a hybrid sampling plus modeling similar to Samplemodeling.

[lasteno]

Everything except the cymbals is entirely modeled. I believe the cymbals part uses a hybrid sampling plus modeling similar to Samplemodeling.

Sad news...... a Drumkit it is not only snare, kick, and tombs.

[Scotty]

Everything except the cymbals is entirely modeled. I believe the cymbals part uses a hybrid sampling plus modeling similar to Samplemodeling.

Sad news...... a Drumkit it is not only snare, kick, and tombs.

I am assuming because no one has done satisfying physically modeled cymbals (non electronic type) that either it isn't possible to do to a realistic level or the cpu hit would be too high for this generation of computers.

[xoxos]

/\ that there

this post is to lend some historical background to this field. yes, basically, modeling a drumkit to "expectations" is tremendously computationally intensive. i have no clue what methods of rendering this company are using and i don't want to co-opt the subject, but many consumer expectations don't seem to have sight of what is now being offered -

http://www.falstad.com/circosc/
this interactive java applet gives instantaneous modalities for a 2d ideal circular membrane. the audio component is bullshit. but to generate modal magnitude and phase from a given striking position requires (and i believe this to be the most efficient method)

(FFT operations + bessel calculation + multidimensional integration) * number of M modes

say a 16 * 16 array of m * n modes is 256 partials, which might only extend to 10kHz. you gotta render all those partials. with individuated decays and modulation depending on the complexity of the model.

i don't think there's a single vst (maybe in some dsp packages like the stk?) that does a circular osc simulation (apart from my own 'radian' which uses an approximation of amplitudes and "can be improved" as an instrument and model).

for a snare, you could say render partials up to 4 or 5k then use more efficient generalised algorithms for the noisier bands past (maybe, i mean to try this sometime). or render spectrally in groups, i dunno. anyway. i don't know what methods they are using (or why they need so much space) but this is the best drum model i've heard by far, ever, being an participating enthusiast in teh field. yeah i can see it would be cool to include a full retinue of woodblocks et c. but imo this product is offering something the market has waited for since the market began. maybe they can do some updates but it's just like wow, someone finally pulled their finger out.

according to bilbao (his examples sound convincing to me) cymbals, idealised as curved shells, "move one group of modes down and another up" by curvature. people have mentioned desiring interactivity.. and my chief observation is that cymbals rock around.. (well mine did) okay, take a look at the applet or the piccie up above, same thing... modes, described (iirc dissecting then radiatively) as m and n, so you have m * n modes... and you need a lot of them to accurately describe a spectrum...

.....you'll notice the blocks in those diagrams sort of run in pairs. the first row (or column in the other) are radial modes, circles around the center... then each pair beyond this describe an angular mode, which travels around the center..... these modes are described as a pair of weights in this cartesian map (instead of polar map) so that between the two we can describe one partial of arbitrary phase. so 2x as many partials if we want something we can modulate (at least, one way to do it, fortunately, quadrature hapens). ..that's a lot of partials to render.

if you slam a cymbal, it rocks around on its posty thing and that modulates sound radiation _for each mode_. brute force, for say 1000 partials for a convincing cymbal, well, i have a very old computer and don't use graphics accel and stuff so i don't know but it won't run for me.

then you have 3d cylindrical modes for the airspace. (i think this can safely be rendered once the size is set to a finite partial count but it really depends how discriminating the listener is, what a task, people have been listening to rock kits for omg years) rim and second head. and the snare. its a lot of dsp to run. i mean, i'm bullshit and i'd try something super cheesy like filtering XOR pulses for more complex spectral regions.

sure, my pov is irrelavent, consumer expectations are at the fore, and its pretty cool to see this first entrant as a set that can do resonances. i'd say, there's probably many interesting ways to augment emulative quality of cymbal modeling methods, but yes the market won't see them because they're not going to sound like they want and that's not astute to do commercially.

[jancivil]

Yeah, I think the people deciding the appeal would be cymbal modeling don't realize at all it doesn't really happen, for reasons.

[xoxos]

wtf tried to edit