Diversion 1.4 released

[Sampleconstruct]

I didn't feel the need to state the obvious. If they were keys or leads that would likely extend into the higher octaves, I'd completely agree. But, with pads which are usually played within a few octaves, it is doable. I mean, what is the alternative -- having full blown unison with maxed out voices which would be hard to fit into a busy track? In this case, I'd say the benefits outweigh the costs. About waveforms not blending/looping smoothly, when was the last time you checked out diversion? This seems to be a non issue, I'd never had problems with this -- it blends perfectly for the most part. Stream longer segments into the oscillators if need be.

I use Diversion on a daily basis, so I am fully aware of what it can do and what not, "seems to be a non-issue" sounds pretty vague, but just go ahead, if this works for you -> fine, imo people need to know about the facts, it's not all about opinions.

I gave a suggestion on how to save cpu, a process that works and is therefore a fact. You saying that higher octaves would alias so therefore that option is not valid, is your opinion bro

I'm not your bro, the fact I was referring to is the fact that Diversion does not offer any kind of crossfade looping or even alternate looping. Over and out.

[Touch The Universe]

I was refering to your original comment,"if Diversion were a sample player with multiple key zones, fine, but it isn't, so your suggestion doesn't really work." Why so hostile? bro is a casual friendly referral and obviously not intended as some deep symbolic gesture. It was simply an olive branch to relieve tensions and leave some room for the argument. why you'd need to feel compelled to retort it, is beyond me

carry on gentleman

[Sampleconstruct]

I was refering to your original comment,"if Diversion were a sample player with multiple key zones, fine, but it isn't, so your suggestion doesn't really work." Why so hostile? bro is a casual friendly referral and obviously not intended as some deep symbolic gesture. It was simply an olive branch to relieve tensions and leave some room for the argument. why you'd need to feel compelled to retort it, is beyond me

carry on gentleman

->Back in: Relax, if this came across as hostility, it wasn't meant to be, I just don't like the bro-mate-dude thing so much and my original statement wan't hostile at all, I was just questioning your "method" of re-sampling Diversion waveforms which you have been advertising on this forum for weeks now, all good with me, no beef, no hostility, love and peace (and no smilies, I've decided to avoid smilies).

[recursive one]

Picked Diversion up in the market place. I had already read about the CPU consumption of the synth, but thought i'd give it a ride anyway, because i couldn't imagine it was THAT severe. Unfortunately, it is... barely usable for me. Even Spire before the 1.1 update took half the CPU Diversion takes for similar patches. And compared to version 1.1, it rather takes 10 times the CPU. Add 2 oscillators, 8x unison, 2x oversampling, which is the lowest option, the analog lp1, and some pad style envelopes with long release, and it maxes out my CPU, with one instance. Well, that's really much too much for the sound quality presented. Really hate to rant, because i really like the filters, and also the architecture seems very interesting, but, meh. How are you supposed to use this thing in tracks with up to 50-60 tracks, and you have to bounce every single one of them, right from the beginning. Quite unacceptable for me. I read in another thread that there won't be an update which will improve this, because the dev thinks that this kind of required processing power is necessary for the sound of Diversion. Well, IMO, there's at least a handful of softsynths which sound as good, or better, and don't slaughter your CPU for breakfast. I thought Diva was heavy, but this is insane compared.

Forgive me this devastating resumé, but i really expected this to be hard on the CPU, but i didn't expect it to have such a hefty foot print. Why does it have to take 10 times as much as Spire? /me is puzzled.

Yes, it's a CPU hog, indeed. But you may compensate for that by doing (rather by not doing) two things.
1. In most cases you don't need 8x oversampling, 2x is enough. It reduces CPU consumption quite considerably.
2. Diversion only has global unison, which multiplies the whole signal path and frankly, big heavy detuned unison sounds, like pads or hypersaws, is not it's strongest point. It is not intended to compete with Spire in this specific area. If I use unison, it is usually 2 or 3 voices, I almost never need 8 of them. On my i7 it is is acceptable CPU-wise.

As for bouncing, I don't know which DAW you are using, but for me it is not a problem. I use Studio One, where you can bounce a track and then convert it back into a synth channel by pressing one knob. I was also reluctant about such workflow before I got a Virus. You may run only one instance of the Virus plugin at one time, this means that you have to bounce it all the time. I made my last track only with the Virus (about 40 channels iirc), and surprisingly found out that when you get used to this, it enhances worflow rather than slows it down - instead of contstanly tweaking stuff to get a perfect sound (which you can never acheive anyway lol) you just press "convert to audio" and move on. With Diversion it should be even more convenient because it loads and bounces much faster.

[chk071]

Good points. I will surely not give up on it, even though the first disapointment was quite big, because such a complex synth seems predestined to do pads. Those filters though, yummy.

[recursive one]

There is a free "collaboration" soundset fro Diversion. you may download it from the Dmitry's site. There are some "trancey" pads in it, which sound huge using only 2x unison. E.g., the preset "mythical pad", wich is one of the best trance pads I've ever heard, gives between 4 and 7 CPU load here at 4-note chords, pretty acceptable for pads imo.

I have CPU spikes when loading presets'though, seems that the synth does some precalculations.

[fischkopf]

Well, the synth's CPU demand doesn't change over time, so as time passes more people will have the hardware to really push this synth to it's limits.

[aMUSEd]

I didn't feel the need to state the obvious. If they were keys or leads that would likely extend into the higher octaves, I'd completely agree. But, with pads which are usually played within a few octaves, it is doable. I mean, what is the alternative -- having full blown unison with maxed out voices which would be hard to fit into a busy track? In this case, I'd say the benefits outweigh the costs. About waveforms not blending/looping smoothly, when was the last time you checked out diversion? This seems to be a non issue, I'd never had problems with this -- it blends perfectly for the most part. Stream longer segments into the oscillators if need be.

I use Diversion on a daily basis, so I am fully aware of what it can do and what not, "seems to be a non-issue" sounds pretty vague, but just go ahead, if this works for you -> fine, imo people need to know about the facts, it's not all about opinions.

I gave a suggestion on how to save cpu, a process that works and is therefore a fact. You saying that higher octaves would alias so therefore that option is not valid, is your opinion bro I only asked when the last time you used diversion since I've never had any show stopping problems using that method, figured maybe it was an older version issue that has since been fixed.

For me the show stopping problem of this approach is it turns a synth patch with oscs generating sounds in realtime into a sampler with one static sample mapped across the keyboard. Even leaving aside the aliasing issue, although I may *possibly* use such a patch as a low cpu way to sketch some music then substitute it for the original patch at high quality for bouncing down, even then my reservation is I would have no way to edit the oscs further except by changing the sample itself, since the oscs are no longer generating the sound in realtime I can't change it dynamically as I would with most patches. For me that's a big issue as if I do buy someone else's patches I very rarely use them as they are, they are usually just starting points for creating a sound that is more unique.

[Lejurai]

When do we see 2.0?

[urlwolf]

I see that LFOs don't have a delay parameter. Is there any way to make a delay?
Maybe an MSEG that is flat and then jumps, assigned to LFO amount? a bit convoluted, right?

[ere2learn]

I just use an envelope on the LFO amount.I think it allows for more freedom than just having a delay amount

[Lejurai]

I just use an envelope on the LFO amount

Same here.

[urlwolf]

Looking at the basic saw in a scope, it feels rounded, even though there's no LP filtering that I can see. This is worrying. Why is that? I started an with an ini file, selected basic saw.

I also tried to replicate a saw by using a sine with FM self-feedback to maximum. It still looks rounded up on the scope. I suspect there's some LP filtering that I'm not seeing. All filters are off AFAIK. Anyone can shed light? I would not want to use a synth where the saw is not a saw

[ere2learn]

Something up,check the RM(ring modulation) on the side,

I get this,on a new init

[ere2learn]

Looks like you have some drive on somewhere with ASYM 2