Non-multitimbral version (EXS24 like) planned?

I'd really like to buy Kontakt, because of it's flexibility and the fantastic sounding HQI-algorithm. But the

multitimbral concept, copied from the ol times of hardware samplers, keeps me away.
I need Logic's EXS unbeaten one instance per sound concept and not be caring about instrument routings and all the

involved further complications, when it comes to freezing, ect. Therefore i'd like to ask, if there any plans from

NI, to release a non-multitimbral version of Kontakt?
I've read, that Kontakt eats so much memory that it forbids to open 10+ instances.

 

...if you'd use Ableton Live, maybe their new sampler would be something for you...

 

this has somewhat been discussed before, not the option of a monotimbral version, but the problem:

how much overhead does the multitimbral features of K2 introduce, when used monotimbrally? naturally it uses some, but is it negligible? NI hasnt answered anything official. personally i think its close to negligible and the benefits outweight whatever small price there is

i use K2 like EXS, one plugin instance for each sound or "channel" in a mix. i have never had any problems with too many instances, never met the "roof".

i just tested now, very informally: when i have 30 instances of K2 in Logic, each playing one sustaining note (the Modula Saw program in the Library, Classic Waveforms, ModulaM), Logic is using approx 8-10% of each CPU

if each instance plays a 4 note chord, CPU jumps to around 30% on both CPUs

if the instances are IDLE, but logic is playing - logic reports around 6-7% CPU.

so in a very unscientific report conclusion; 6-7 % cpu is the price for 30 instances of K2. theoretically i could run hundreds of instances, i dont bother trying. this on a dual g5 2ghz, logic pro 7.2

i dont know if there is a K2 demo? then you could try for yourselves. anyway there's always the 30 day return policy?

AAAAAAAANNYway. its a good suggestion. i wouldnt mind having it (monotimbral). but i dont think its a problem. so i guess it wont happen.

 

What about RAM?

 

Today i visited a studio, having K2. We checked with the Windows task manager the memory useage: each naked instance of K2 with minimal output configuration costs around 70MB of RAM!!! What a joke. If K2 wouldn't sound that good with HQI, i would call it bloatware.

Since days i'm searching for alternatives to EXS and it's hard to believe, that all companys only use the s..t multitimbral concept, although they should know in the meanwhile, that in software it doesn't cost more, to open more instances - in oposite to hardware samplers...

Thanks for the hint to the Ableton-sampler. Will it have a high quality Sinc-interpolation mode and be avail as VSTi, too?

 

I don't know the exact specs yet, but it looks very cool - given you're a Live user. It certainly won't be released as a VSTi, though, it'll only work in Live...

 

good point, i didnt think of it. i'll check tomorrow in studio

 

ok i did some RAM investigations while having my morning coffeee..

My empty Logic autoload uses 230 MB RAM.

With 1 empty K2 instance makes it 272.
With 2 empty K2 instances makes it 325.
With 3 empty K2 instances makes it 396.
With 4 - 455.
With 5 - 532.

So about the same as your experience on PC Schiller, on average 60 MB for each instance.

Adding EXS24 instances does not seem to add any significant amount of RAM to Logic (around 100 k pr instance it seems)

The Kontakt2.component is 30 MB in size, the same is the K2 standalone application. Why does it take so much RAM?

 

K2 is many many times more powerful than EXS, which is really a very simple sampler. EXS has a smaller footprint because it simply can't do a lot of what K2 does. It's really futile to compare the two.

I understand the request for a monotimbral version of K2, but it's unlikely that it'd be possible to create it with a smaller footprint, Therefore, if you want to use it monotimbrally, just load up one instrument per instance -- voila!

 

...I'm sorry, but I don't understand this discussion completely. I think, it's about visibilaty - so, if some people are bothered by the many outputs of Kontakt in their sequencer mixer, maybe a master output in Kontakt's internal mixer would be a solution for some (do a FR in this forum). Then you could apply one channel to each desired instrument within Kontakt to mix internally and you'd have only one output in your sequencer mixer. I don't know if such a solution would make sense, but neither does this request make sense to me ;-)... (but that's just personal).

But using one instance of Kontakt for each instrument makes freezing a lot easier - so why not freezing the tracks and temporarily switching off the respective instances of Kontakt in your instrument rack - I'm sure, that'll help saving RAM & CPU power...

 

This discussion is opening a can of worms, and from the perspective of NI (which I'm only guessing at -- this isn't said with any inside knowledge) they would probably not even want to open it.

So the request is: make a monotimbral version of K2 so that you can use one instrument per instance, for easier freezing, and without all the extras like multiple outputs etc.

But: it would have to have a lower CPU/RAM footprint. Because if it didn't, there would be no point -- you could simply use K2 as it is today, and use only one instrument per instance.

But: the minute NI releases this, some users who utilize surround sampling will lament that it only works for the stereo folks because multiple outputs are taken away. And multiple outputs are just one feature that might be taken away in a slimmed-down one-instrument version of K2. The more things NI takes away in an effort to make it more lean, the more users will complain that the features taken away are the features they absolutely need.

But: you have to assume that it's possible to program K2 to use a lower CPU/RAM footprint -- and this is not a fair assumption. I know NI is always working to optimize their apps, and they work to improve things over time, but I give them a fair shake that as of today, K2 is as optimized as they can do it **for today**.

But: it would carry the fundamental limitation that you can only use instruments, not multis. Some entire libraries rely on multis.

But: most major hosts allow for ways to independently control K2 multitimbrally -- today -- most people who use K2 as a plug-in probably already use it multitimbrally.

But: there are already provisions for outputting separate instruments to separate outputs.

In the end, my opinion is it's not worth it at all. It causes too much confusion and opens a can of worms. It's based upon a questionable premise (that it could be done leaner). And anyone who wants to use one instrument per K2 instance can do this TODAY -- with no tweaking.

My opinions only.

I do understand the simpler interface of EXS and the simpler mindset of one-instrument-one-instance. I just don't see a feasible way to mold K2 into that mold and have it be a success.

 

Good points. But if Emagic could do it, why shouldn't NI do it, ten times less effectively? No one says, a slim K should use only 250KB.
We are talking about factor 100-300!
And btw, the whole library concept with sets, is marketing driven only. Have you ever seen, how studios have their drum-libraries organized? Not per set, but per instrument.

I admit, that with my unfounded knowledge i simply expected NI being able to make a slim monotimbral sampler like Emagic did. That is obviously pure speculation.

In the meanwhile i found a sampler, with a monotimbral option: DirectWave. It also has Sinc-interpolation, but the monotimbral instance of the demo here uses the same amount of RAM, as the multitimbral version: 30MB.
Too bad, because this thing can read the EXS-format.

 

The main problem will be sharing sample between the instance of K2.

I dont know if you can do that being an AU ?

The EXS can do that because it is hardcoded in Logic.

Instead NI should look on how they can improve K2 :
- Work with Apple. (multi with 16x3 midi, CC 127 problem, crashes ....) watching the core audio dev list, I have never seen a post of NI
- Be able to freeze the 16 x 3 instruments in one go
- Change the DFD and use MAC OS X VM handling, like does VSL VI and the EXS
- Allow to key switch using program changes
- Make a script that does Legato and machine gun killing in the same script
- Allow to convert CC 127 to CC 11
- Allow automation of each parameter in track automation

......


Best
Cyril

 

Hi all,

the large RAM footprint of each Kontakt instance is also due to your DFD settings. set these to 0 and you should use about 45 MB. that is the minimum. period.

we already discussed the possibility of a smaller engine during the last months. it's not possible. you will only gain overhead if you use several instances of Kontakt (see below).

stay with EXS or reap the power of Kontakt ;-)


best,
Markus

 

Danke Ihnen Markus, für die ehrliche Antwort.
Verdammt schwere Entscheidung!

 

How do you convert .nki and .nks to EXS ? LOL

Kontakt is starting to be good product but it can be improved.

There is definitively a problem when you play 2x60 simultaneous notes.
K2 is having hanging notes with only 50 % CPU usage (in K2/engine)

One information for Logic Users :

You have to be very careful of the value of the "I/O buffer size" in Pref/audio/driver bigger it is more voices you will get.

With Logic 7.2.2 with "I/O buffer size" set to 128 you get the Error sync between audio and Midi messages (see picture attached) playing only 2 x 36 simultaneous notes. (using QLSO Platinum Strings 70 PIECE STR programs)

 

Does the Kontakt Player 2 solve any of these problems, i.e. does it use less ram per instance, etc?

 

No improvement on CPU and Disk I/O load, sorry !

Speaking with the "Apple Logic" guru at an expo in Paris, Emagic has bought license of the "DFD" équivalent and has include it into Logic
This is why the EXS is SO FAST

Best

Cyril

 

I could not disagree more
For EXS and VSL you need several instances/tracks of exs to get several articulations - in Kontakt you can host several/all articulations
in one instance.
There is not difficulty in routing - just but all the MIDI regions on the kontakt track - they just need to talk on different MIDI channels ....
Or route the MIDI tracks to the Kontakt instrument track ...

I think exs design is overcome - because it is lame to load 12 exs for the solo violin and compose distributed over several tracks (of course one can manage that in the environment). This way one really needs 128 instrument tracks and more ....

best

 

Have you heard of EXS "keyswitched instruments" ?
Use those you will load many articulation in one EXS

Best

Cyril