Reaktor VST output channels limit

Hi everyone!

I'm working a lot lately with multichannel systems, composing music for large arrays of speakers.
Unfortunately, i'm finding problems keeping reaktor as an efficient tool for multichannel audio, as it is limited to 16 channel output.

I don't understand the nature of this limitation. It doesn't look like it is a VST limitation, as other NI tools like kontakt can manage up to 64 channels.

Is there any reason for it?

 

no?

it's a pitty making an effort on learning an audio programing language and encounter such blockings. I would like to understand if there is any explainable reason for it, or if it is just an unattended feature.

It would be great to overcome this limits on future versions of the software.

Spatial sound design should be available for any serious audio development software. It is being developed more and more on the industry and it starts being applied widely on the entertainment industry. Ambisonics, Wavefield Synthesis, 4DSound... all of them rely on flexible environments where extensive multichannel capabilities are a must. I'm not asking for high level macros for any of this mentioned systems... just being able to output to more than 16 (ridiculous) channels!!!

Max made its smart move with the mc. feature, and i'm afraid NI is strategizing Reaktor just as a medium-level user tool for re-mixing synths, and not anymore as a cutting edge audio development kit... which makes me really sad, honestly...

 

Discussions here on the forum seem to have no influence whatsoever on NI in terms of features and designs direction. So maybe open a support ticket or something?

That aside, I would be willing to bet that the vast majority (like 99.8%) of professional and amateur users of Reaktor are making music and music tools that target end users using stereo equipment or up to 7.1 surround. Sure there will be a small minority who want to use more than 16 channels, but is it really more than a very small minority? I'd bet that at least 99% of users wouldn't have access to enough suitable speakers to test more than 16 channels.

I'm not saying that that should be a reason not to provide support for those people, but I guess it pushes it way down the priority list when it comes to requested features (it also makes it more difficult expensive and time consuming to beta test - how do you know that it works on a wide range of systems when so few users would be able to test it?).

I suppose that a compelling argument to support the feature would be that top pros want it, or use it with competing software and that adding it to Reaktor would be valuable from a marketing perspective... So which well known music industry gurus use more than 16 channels, and in what context?

 

I get your point and i totally agree that under a certain marketing positioning it is not attractive for NI to move Reaktor into a fully powered audio development suit.
It seemed to be that way back in the days of the release of Reaktor 5, when Core was added. It was attractive for an audio nerd to dig into Reaktor instead of other environments like Max, PD, Chuck or Supercollider...

I'm just sad that nowadays this strategy seems to be abandoned in favor of wider, more commercial orientation. I totally understand that 99% of the users will not be interested in extensive multichannel capabilities, and that they don't have access to big numbers of speakers (neither do I, as i work simulating in stereo and don't know what i will find until the gig setup), but i don't think there's a considerably bigger percentage of users who care about Chebyshev filters, or about many of the great and cutting edge developments being done in Core.

I can think of some situations, not that nerd or top-professional, where having more than 8 stereo pairs would be great for making music:

-Routing different synths (i'm thinking of blocks) to different physical audio outputs, for a higher integration between Reaktor and the eurorack ecosystem
-Using it as a multi-timbrical VST engine, in a similar way to that of Kontakt, where different instruments can be routed to different channels in your DAW for further individual processing

Sincerely, i don't think this is that "snob" or "pro oriented"... but maybe it is just my perception, and i'm in my little world asking for weird things

Anyhow, i'm just curious to understand if there is some kind of limitation on the VST protocol or on the Reaktor software architecture, or if it is just an arbitrary decision...

 

You're definitely wrong there!
I've seen a few threads on different forums in the last few years with questions like "why do Reaktor blocks sound so damn good". Go back to pre core times and the threads were more like "meh, all Reaktor stuff sounds the same", "that stale Reaktor sound".
So while folks might not know about the DSP details, their existence really has been noticed and is making a difference in the marketplace.

Yeah, totally. I'm not against the idea at all - I was just trying to be rational about it...
You could come up with all sorts of cool things to do with loads of outputs... Imagine wiring up a bunch of transducers and attaching them to different things In a room - furniture, the floor, window frames, a piano(a few in different places.... Then sending different audio streams to al of these and recording with a bunch of mics... you could do all sorts of stuff, doesn't have to be about spatial positioning.
And it really is better to have an environment with as few limits as possible. There are real practical limits, but 16 outputs seems like an arbitrary limit, so probably just a design compromise that looked good when that part of the applications was originally being developed.
Now with a bunch of other stuff built on top of it, it will probably be costly to update.

It's nothing about snobbery, more likely like you say, a limit of VSTi or similar that NI just decided to adopt as their own limitation, because it means very slightly less code to develop and maintain (otherwise you need to check if VST mode is active and check the number of channels, throw up warning dialogs etc.)

 

Are you sure about that? I don't have hardware where I can output 40 channels but it looks to me that theoretically Reaktor can do it.

 

that's just on the standalone version of Reaktor, which is limited to 40channels.
But in the VST version, you cannot use more than 16:


I'm interested in using more than 16 channels in the VST, just as i can do with Kontak for example.

 

Any statement from NI here?
Can we expect expanded multichannel support in future versions of Reaktor?
Or throw some light on why this limitation?

 

We did limit it for a reason, which is performance: if you plug in a 40 channel plug, into a 40 plugslot on host side, the host delivers 40 buffers a x samples to the plugin (x = number of samples per buffer), which the plugin HAS to fill, regardless if it is really using the channel or not. So for the "dead" channels, the plugin has to zero those samples. It is a simple task of memset 0 from programming perspective, no problem. As a matter of fact, this operation still causes some cpu load, and it accumulates if you have to do it for more channels.

AFAIR while Reaktor 6 development we were able to reduce that overload in such that at least we were able to handle the case when our 16 channel plugin is loaded into a host's 2 channel pluginslot such we do not generate unnecessary cpu load on our code base side, which was badly handled in Reaktor 5 times.

Nevertheless, we decided at that time that 16 is enough, also to lower the risks of possible unknown problems, as we never did release something with more channels before.

Perhaps not 100% satisfying, perhaps it would be possible to increase the channel number to 40 without causing problems. But it was said in some other posts here: this is not on our top priority list.
Back some months ago when I was replying to some flame war postings after the 6.3 release, I was also stating that I can not tell or make promises for WHAT on our priority list IS and how likely it is for this or that to be implemented, and that has not changed till now, sorry.
If we would be a little 10 person company with only Reaktor as a product this could be different, but as you can see on our web site, this is not the case. No smily here as I am also not really satisfied with this answer.

 

Thanks for your answer Georg!

Got it... I see it has some implications in terms of performance.... And I can understand your position and development priorities. I understand what NI has become. I understand salaries need to be paid and investors need to be satisfied. No blame to you or the development team. You've been doing a great job evolving the Reaktor platform since the old Generator days, and for me, even this whole blocks and racks look like an amazing universe and an interesting line of development!

I remember that reaktor 5 had several vstis and vstfx with different numbers of channel inputs and outputs.
I guess that was an easy solution to keep performance low for users that didn't want a high number of IOs.
Wouldn't something like that be possible for Reaktor 6? a 64ch IO version for the vst?

I'll keep the hope that this specifical claim is attended in the near future, although I understand 98% of the users don't give a **** about multichannel... by now...
I'll try to find a way to share parameter data through OSC between 2 VST instances of the same Reaktor ensemble. Maybe that way i can have my 32 channel output being controlled from a single GUI... but it's a crappy solution and i'm sad this has to be this way... (If you are curious, this is the kind of projects i use all this for: )

Anyway, thanks Georg!
at least there is some explanation now!
Really appreciate you participation!

 

Sure, we just were actually quite happy to ger rid of most of those different vsts with our mentioned code changes and were not wanting to take unnecessary risks by increase the max channel number of the remaining ones. Again, with the huge amount of todos we have, chances are low that we will resolve this.

 

Very interesting looking lightshow Santiago.. didn't know you were into Max as well. I think I read a brief quip from David B. the developer behind ToyBox that he was very close to a nearly audio-rate multiplexing solution that could transfer synchronized streams between multiple instances of Reaktor.. so maybe there will be some relief for you there.. I often find with most things that 1/8 of audio-rate is suitable for typical modulation targets.

 

Do you see his vid from the concert 'Lux Mundi'?

 

I think I did some months ago, but I will go rewatch.. reminds me I need to spend more time on Vimeo.. less on YT

 

Wow! that sounds interesting! do you know if there is any published ensemble that features that multiplexed signals? sounds a bit sci-fi and outside my capabilities, but would love to give it a try!

 

yeah, we used to use MAX/MSP as the main control signals generator... with this control data we fed DMX light fixtures and Reaktor engines, making sure we have a system controlled by the same automation envelopes. But as the complexity of the setups we're managing has been growing (larger arrays of lights and speakers) we've been finding the performance and features limits of MAX to generate control data, thus forced to move to C++.
And looks like now we also found the limits of Reaktor in terms of large arrays of speakers and i assume we'll be forced to switch to MAX (or Supercollider) for the audio engines... which makes me really sad... fingers crossed to try finding a reliable solution and elongate the use of Reaktor for some more years....

 

I was kind of having the same idea, you know to do a layered "hypersynth" with coordinated reaktor instances.. since we can't have multi-core like VCV due to architectural reasons. He mentioned his solution in passing but I think he's focused on updates to his module collections at the moment.. with any luck he'll chime in here!

Yeah, big fan of Max myself because it enables me to do sequencers pretty easy but I've still not really gotten my head around Gen yet and tend to use Reaktor or other VST's for synthesis still, although I'm getting into MC now. Need to find the time to really read some literature on it and learn the Gen paradigm and how best to leverage it in conjunction with Jitter.

That performance you did looked really nice.. truly beautiful atmosphere there.. Not sure if you've looked at Touch Designer at all but it's pretty straightforward from a Max/Reaktor background and I think it might be better for DMX if that's what you're aiming for.. a Max patch in a live context would make me a bit nervous.. just because the program has rapidly grown by leaps and bounds-it's quite a sprawling beast now and not entirely stable as a result imo. At least Reaktor almost never crashes..

This is why I mentioned recently in that other thread if nothing else please just remove unnecessary and arbitrary limitations from Reaktor.