KVR Audio

kuniklo · Post by **kuniklo** » Mon Oct 02, 2006 8:02 pm

Lately I've been working with just a hardware sequencer and a Virus. Can't do everything I can do on the computer, but I have to say I'm enjoying myself a lot more without chasing these rainbows. Music software is more and more these days about breaking the tools and features I *do* need with buggy implementations of new features I *don't* need. There's a backlash coming, I think.

Sascha Franck · Post by **Sascha Franck** » Mon Oct 02, 2006 9:29 pm

Josmoker wrote: Why do you think Steinberg has to care about those? (btw which DX based host is there left?) I would think that VST3 makes things a bit easier for VST-based host developers once support has been added for it.

I'm not sure whether things will become easier indeed. Also, as said, just think about it: To actually make use of that CPU saving feature, each and every plugin programmer has to update his/her plugins. And this very feature clearly *is* something that can be done straight in the host, as other hosts are perfectly proving.
But anyways, my main complaint goes about Cubase 4 apparently not guaranteeing older plugin compatibility anymore. In that article they're actually saying that older plugins may not work properly anymore - which I find ridiculous.
In the end, what Steinberg seems to be trying is forcing programmers to (more or less instantly) update their plugins, so their new CPU saving feature would shine in full glory.

And yes, you're right, I'm complaining about Logics flaws just as much.
Still got my new Macbook today though - and so far the performance under Logic seems stunning.
But jasonsantiago, don't start to feel too happy about that yet - I will still have to use Cubase on quite some occasions. So better ignore me or whatever... so fwiw, I'm bashing whatever I want, if I feel it might be appropriate. And in this case I do.

derek · Post by **derek** » Tue Oct 03, 2006 2:22 am

the sad thing is, cubase and nuendo were able to do the plugin power saving thing all along,
dropout free, and dead easy. most people (including some in hamburg ;-) always thought that ti wouldnt be possible without changing the engine because turning off a plugin produces a dropout (because of PDC reinventing itself i assume) - yet, simply bypassing the plugin doesnt produce a dropout nor a CPU spike (none thats visible on the systems i use fwiw). of course the latency of that channel then isnt properly compensated, but then, at that point theres just silence on the channel - i could deal with non-latency-compensated digital silence :-)

so that basic functionality was there. and as far as "watching the buffers" goes (200 vs 200x256 example), hm...is that really necessary? is there no other way for a host to find out whether theres audio present on a track? i mean, for audio one could just use the audio object boundaries. i actually begged for a "object boundaries to bypass automation" macro in the steinberg forums again and again... and in cases where i had to finish a project that was just a little over my CPUs capabilities but where i needed control on all channels and thus couldnt freeze or bounce (dont you just love those), i actually got myself performance back by automating some bypasses manually.

oh well...i guess the point is, they wanted one clean, definite solution that works for everything (the above only works for audio, and would also work for the new instrument channels, but would not work for the classic midi-channel-routed-to-VSTI approach).

ill still take that (with all due respect to sascha :-) over logic and its "sometimes it works sometimes it dont, and sometimes i will simply cut off delay taps for no apparent reason because thats how i feel like right now). to me the people that did this best are the samplitude guys, of course they do come from a completely different background with their object FX and all...

Sascha Franck · Post by **Sascha Franck** » Tue Oct 03, 2006 7:02 am

derek wrote: ill still take that (with all due respect to sascha over logic and its "sometimes it works sometimes it dont, and sometimes i will simply cut off delay taps for no apparent reason because thats how i feel like right now). to me the people that did this best are the samplitude guys, of course they do come from a completely different background with their object FX and all...

I totally agree with you on both statements.
The Logic issue proves that you can f**k things up easily (fortunately this seems to be fixed in 7.2, but of course I wasn't able to test much yet), so yeah, there's chances the b0rg would f**k it up as well (just slightly different, of course...) while the Samplitude issue proves that you can as well do it with teutonic accuracy.
Either way, it's doable, and apparently it's doable without forcing the host to do too much work (otherwise the procedure would fail in Samplitude as well).

stag · Post by **stag** » Tue Oct 03, 2006 8:57 am

So now you're software engineers?! you seem to be talking like:
"Any car can run on Diesel, see my Mercedes does it, even the a BMW. Why does not this Honda run on Diesel? it's dead easy, any car can do that,"

Once i asked for a mono button on a well known software sampler, i was kindly answered by one of its devs. he told me i wasn't the first to do that request, so he made me realize that a thing that seems so trivial and easy to acomplish from the user point of view, may take a lot of hours of work to untangle the code and then writting it back all over again as new.

Just like with the cars, so simple!! as long as we trade engines... If you think you could teach Steinberg Team to code why don't you do that?? instead of whine about it.

See now you did the rigth thing and pick up the perfect sequencer, or the closest to perfection, in your opinion. That's nice, that's coherent. I hope it fits you good!

mbncp · Post by **mbncp** » Tue Oct 03, 2006 10:19 am

derek wrote:so that basic functionality was there. and as far as "watching the buffers" goes (200 vs 200x256 example), hm...is that really necessary? is there no other way for a host to find out whether theres audio present on a track? i mean, for audio one could just use the audio object boundaries. i actually begged for a "object boundaries to bypass automation" macro in the steinberg forums again and again... and in cases where i had to finish a project that was just a little over my CPUs capabilities but where i needed control on all channels and thus couldnt freeze or bounce (dont you just love those), i actually got myself performance back by automating some bypasses manually.

Sure you don't have to scan the whole buffer, but in any case it is still more than querying a single value. Also, I don't like the idea that the host takes the decision to bypass a plug.
And in case of a plug with 64 I/O, if only one channel is used, the host can't bypass that plug anymore, and the plug still has to check all of them.
And currently we are speaking of a very limited amount of channels, how many plugs have 200 channels or more. A sampler with 1000+ channels could pass each voice to a single channel, allowing per voice fx/handling.

Now I don't mind conservative limited minded people

arturmeinild · Post by **arturmeinild** » Tue Oct 03, 2006 10:34 am

- Can I still use my existing plug-ins and instruments?
Yes. Existing VST 2.4 plug-ins and instruments are fully compatible with Cubase 4. Any plug-in or instrument older than VST 2.4 must be tested and potentially updated.

So what this means is that plugin developers will actually have to develop 2 versions of all plugins for now. VST 2.3 versions for 32 bit hosts (Cubase SX/SL 3 and earlier) and VST 3 versions for 64 bit hosts. Is this correct, or are there other possibilites, with VST 2.4 for example?

ObsoleteAcc99 · Post by **ObsoleteAcc99** » Tue Oct 03, 2006 1:29 pm

[DELETED]

Josmoker · Post by **Josmoker** » Tue Oct 03, 2006 9:43 pm

So what this means is that plugin developers will actually have to develop 2 versions of all plugins for now. VST 2.3 versions for 32 bit hosts (Cubase SX/SL 3 and earlier) and VST 3 versions for 64 bit hosts. Is this correct, or are there other possibilites, with VST 2.4 for example?

Nope, certainly doesn't have anything to do with 32 bit vs. 64 bit hosts. Plus they probably expect that new developments are going to be done in VST3.

Cheers, Jo

aldoofbanga · Post by **aldoofbanga** » Tue Oct 03, 2006 11:44 pm

From reports most VST plugins (including pre 2.4) are still working in C4

ObsoleteAcc99 · Post by **ObsoleteAcc99** » Wed Oct 04, 2006 2:29 am

[DELETED]

derek · Post by **derek** » Wed Oct 04, 2006 2:35 am

stag wrote:So now you're software engineers?! you seem to be talking like:
"Any car can run on Diesel, see my Mercedes does it, even the a BMW. Why does not this Honda run on Diesel? it's dead easy, any car can do that,"

well, if honda would claim its impossible to make a car run on diesel and there were several competitors that showed that it is possible, that would be a correct observation dont you think? ;-)

anyway, no reason for a heated debate over this. im a bit sad that steinberg didnt introduce plugin performance management that works for all plugins, but im still happy that they introduced a solution that leads to a performance managing cubase and nuendo at some point in the future.

derek · Post by **derek** » Wed Oct 04, 2006 2:44 am

mbncp wrote: Sure you don't have to scan the whole buffer, but in any case it is still more than querying a single value.

knowing that this is really just theorizing after the fact, i have to ask, is it really? cant a host know "audio object present on audio track yes/no"?

and fwiw, the thing i suggested (a macro that copies audio object boundaries to the bypass automation of that track) would still work and still help with old plugins.and you could only do it when you needed it. thats actually a feature i still would love to see.

mbncp wrote: Also, I don't like the idea that the host takes the decision to bypass a plug.

well, in *any* possible scenario i think performance management should also always give you a button to turn it off. for the rare cases where it causes more trouble, like your 64 i/o plugin...

mbncp wrote: And in case of a plug with 64 I/O, if only one channel is used, the host can't bypass that plug anymore, and the plug still has to check all of them.

just out of curiosity - i cant imagine any real life scenario (or any existing plugin) like that. thats not because i dont believe you but because i just dont know any, and because im a nerd, i want you to tell me about that now :-)

mbncp wrote: And currently we are speaking of a very limited amount of channels, how many plugs have 200 channels or more. A sampler with 1000+ channels could pass each voice to a single channel, allowing per voice fx/handling.

i think most people here talk strictly about effect plugins. btw, there is a sampler that already offers what you say - vsampler offers up to three individual VST effects *per sample*. and with performance management too!

mbncp · Post by **mbncp** » Wed Oct 04, 2006 9:25 am

Sorry, but why playing tricks, guessing, user input request, when all this can be totally transparent and far more efficient.

And I hope that in the future samplers will have a bunch of I/O, so I can assign a separate channel for each drum and percussion instrument, then I can create my own little mixer (64 I/O) to deal with them, my way.

Sascha Franck · Post by **Sascha Franck** » Wed Oct 04, 2006 9:29 am

Can we hear some of your drumworks, mbncp? I mean, what would it be, requiring more than 32 individual outs?

Cubae 4, older plugins, performance improvements...