A new competitor to Uhe Diva has arived!!!

[Urs]

Hmmm, the purpose - if any - of this thread strikes me as odd, but if I see anything blatantly disrespectful against NI, I don't have the time to argue or edit. I just delete.

[Krakatau]

fair...

[TomTwohy]

I've been reading this pile of opinions, arguments and overwrought teenage angst... and for my money, Monark kinda makes sense.

Diva is probably never going to be released in the Reaktor format... so NI figured they would make their own version JUST for Reaktor and pick up the few users U-he didn't feel like pursuing (for whatever reason). They might have reverse engineered Diva the day it came out and have been trying to figure out how they can use the results without U-he suing the PANTS off 'em.

It seems (by reading the brochure) to use the same predictive elements that Diva uses (though a much less robust feature set) and in essence, for Reaktor users, allows the same Moogy experience.

For my money, Diva seems to be nicer and somewhat more useful... more features/control, nicer looks, broader scope of potential users.

Can't we synthesists all just... get along together?

[digitalboytn]

Yesterday I was travelling from Dali to Kunming in SW China and I saw a HUGE billboard...

A baby boy was taking a long loopy piss into a shiny new toilet bowl..

There was only one word for the advertising on the billboard..

"MONARCH"

I did have a bit of a chuckle and I wished that my camera was ready to roll...

It was a classic

[chilly7]

Yesterday I was travelling from Dali to Kunming in SW China and I saw a HUGE billboard...

A baby boy was taking a long loopy piss into a shiny new toilet bowl..

There was only one word for the advertising on the billboard..

"MONARCH"

I did have a bit of a chuckle and I wished that my camera was ready to roll...

It was a classic

Lol, i think NI will be really pissed of if u would post it.

Do it!

[chilly7]

I've been reading this pile of opinions, arguments and overwrought teenage angst... and for my money, Monark kinda makes sense.

Diva is probably never going to be released in the Reaktor format... so NI figured they would make their own version JUST for Reaktor and pick up the few users U-he didn't feel like pursuing (for whatever reason). They might have reverse engineered Diva the day it came out and have been trying to figure out how they can use the results without U-he suing the PANTS off 'em.

It seems (by reading the brochure) to use the same predictive elements that Diva uses (though a much less robust feature set) and in essence, for Reaktor users, allows the same Moogy experience.

For my money, Diva seems to be nicer and somewhat more useful... more features/control, nicer looks, broader scope of potential users.

Can't we synthesists all just... get along together?

So NI hired Diva engenirs to creat something similar but for Reaktor to see what happend?
Interesting

P.s. We need all Uhe synth in Reaktor
at list i need them all there.

[Meffy]

Hmmm, the purpose - if any - of this thread strikes me as odd, but if I see anything blatantly disrespectful against NI, I don't have the time to argue or edit. I just delete.

Thanks for the deletion. Hakey was right, the poster whose post(s) you deleted was a returning troll, banned at least three times so far.

If you want to move the thread out to Instruments, where it wouldn't be so weirdly out of place, I wouldn't object. Or not, as you prefer; it makes no difference to me.

[Urs]

It seems (by reading the brochure) to use the same predictive elements that Diva uses

I highly doubt that. Their approach to zero delay feedback filters is based on Vadim's work, which IIRC is an analytical approach. If I'm not completely mistaken, his approach can make an accurate estimate of the output for any input and state, under the condition that there's a limit of non-linear elements - not sure if that's still the case, I think he also proposed Newton-Raphson refinements, which hints at something similar to our method.

#---

Because this still comes up every now and then and because I believe that the majority of "Diva competitors" use CPU-friendly analytical approaches to model zero delay feedback filters:

No-one can accurately predict the filter output without delay if there is more than one non-linear element involved using analytical methods. It's either a wild guess in our books or just not distorted right. There's a great approach discussed here on KVR which mixes zero delay feedback for the linear filter but adds a unit delay to the difference of the distorted part of the signal. But that's not strictly zero delay feedback.

Diva otoh uses a numerical method, i.e. she uses a "good guess" and then iterates through the filter, improving the guess time by time, until the result is good enough. The advantage is, with the VCA model, the external input amp and the filter we have up to 7 non-linear elements and up to 8 filter poles per model. I doubt that anyone can do this in is an analytical way. I also doubt that anyone can do this sigificantly faster than we did. Our algorithm is really good and really fast. If someone is significantly faster he therefore must either use a less complex model, or a less accurate integration method (or less oversampling/more aliasing, naturally).

I usually suspect "less complex model" when people proclaim Diva-quality at much lower CPU impact. We have yet to hear a less complex model though that comes sufficiently close to our goals and ears.

I have a hunch that developers who start on a numerical approach quickly give up and go analytical. Because initially the numerical approach might not render in realtime. To those I say, it took us a few months to understand the behaviour of the analytical approach well enough to make it as fast as it is. Time well spent.

- Urs

[penguinfromdeep]

[medienhexer]

I have a hunch that developers who start on a numerical approach quickly give up and go analytical. Because initially the numerical approach might not render in realtime. To those I say, it took us a few months to understand the behaviour of the analytical approach well enough to make it as fast as it is. Time well spent.

- Urs

I don't understand all of this but my powers of deduction tell me you meant to say it took you months to understand the behavior of the numerical approach well enough?

[Urs]

I have a hunch that developers who start on a numerical approach quickly give up and go analytical. Because initially the numerical approach might not render in realtime. To those I say, it took us a few months to understand the behaviour of the analytical approach well enough to make it as fast as it is. Time well spent.

- Urs

I don't understand all of this but my powers of deduction tell me you meant to say it took you months to understand the behavior of the numerical approach well enough?

Yes. The approach itself is fairly simple. Making it fast isn't.

[EvilDragon]

I think that medienhexer wanted to point out you said "understand the behaviour of the analytical approach", where you talked about "numerical approach" for the whole time.

[hakey]

I also doubt that anyone can do this sigificantly faster than we did. Our algorithm is really good and really fast.

Sounds like an example of computational irreducibility?

[aMUSEd]

Anyway I have it - K9 package just arrived!!

[bmanic]

@Urs: Because it's a reaktor ensemble you can just open it up and take a peak at how it's done. Perhaps there is something to learn? Or not. Never know.

Cheers!
bManic