Login

2 notes about a gearslutz thread

Tips & tricks, working results, technical support

Re: 2 notes about a gearslutz thread

Postby TranscendingMusic » Mon May 31, 2010 3:45 am

But really the main point is:
1) limitations of merely using one static tone
2) using FFT for analysis
3) that putting that much focus on aliasing, especially while using limited means of analyzing, does not convey the whole story of what's going on. Case in point, they observed more aliasing in UAD. But they were, as G said, low level, but aside from that many of us confer that UAD compression sounds nice; the most important factor.
mixing | mastering
Win 10 x64 | Sonar Platinum x64 | 3930K(OC)
User avatar
TranscendingMusic
Expert
Expert
 
Posts: 1132
Joined: Sat Mar 27, 2010 7:01 am
Location: USA

Re: 2 notes about a gearslutz thread

Postby Stolle » Mon May 31, 2010 12:36 pm

But what that shows is that there is MORE to be found. Not that what has been found isn't right. I find that thread very inspirational and it got me measuring my plugins and my hardware. I find that analog hardware never gives you aliasing while plugins most often do. For plugins trying to emulate analog hardware this must be seen as a flaw. If it can be removed then why not bother?

And yes... of course the common sense rule, that if it sounds good it's good, is still true. I find that the UAD 4k sounds phenomenal despite aliasing. I find that Nebula sounds phenomenal despite aliasing (and Nebula shows much higher levels than any other plugins if the test-tone is of a high enough frequency). Would I like them even more if they didn't exhibit aliasing? I don't know really. I suppose I'd better set up a blind test for myself using two different samplerates with Nebula.

But again, if it can be removed then why not do it? It makes it less like hardware right?
Stolle
User Level VI
User Level VI
 
Posts: 68
Joined: Mon Mar 29, 2010 3:27 am

Re: 2 notes about a gearslutz thread

Postby TranscendingMusic » Mon May 31, 2010 4:45 pm

Stolle wrote:But what that shows is that there is MORE to be found. Not that what has been found isn't right. I find that thread very inspirational and it got me measuring my plugins and my hardware. I find that analog hardware never gives you aliasing while plugins most often do. For plugins trying to emulate analog hardware this must be seen as a flaw. If it can be removed then why not bother?

And yes... of course the common sense rule, that if it sounds good it's good, is still true. I find that the UAD 4k sounds phenomenal despite aliasing. I find that Nebula sounds phenomenal despite aliasing (and Nebula shows much higher levels than any other plugins if the test-tone is of a high enough frequency). Would I like them even more if they didn't exhibit aliasing? I don't know really. I suppose I'd better set up a blind test for myself using two different samplerates with Nebula.

But again, if it can be removed then why not do it? It makes it less like hardware right?


I couldn't agree more, that from a fundamental level, doing these measurements can be a learning experience for those that are more or less not aware of it. But I see that what had happened is, for those who let's say would already know, to even bring up these shortcomings or better yet to even mention what Nebula can do, sort of makes one isolated. Simply because they tend to gloss over it. That's really the issue I suppose. It's the attitudes that would prevent somebody like Giancarlo from not going there any more. Bob Katz, I found out, does not go to Gearslutz anymore! For similar reasons.

And don't get me wrong, I am on board with the idea of removing aliasing which sure enough can be implemented with oversampling (possibly leading to other issues).

By the way when you said this: "It makes it less like hardware right?" you meant "more like hardware"...correct?
mixing | mastering
Win 10 x64 | Sonar Platinum x64 | 3930K(OC)
User avatar
TranscendingMusic
Expert
Expert
 
Posts: 1132
Joined: Sat Mar 27, 2010 7:01 am
Location: USA

Re: 2 notes about a gearslutz thread

Postby Stolle » Mon May 31, 2010 10:09 pm

TranscendingMusic wrote:By the way when you said this: "It makes it less like hardware right?" you meant "more like hardware"...correct?

Cool, we seem to agree about this. I actually meant less like hardware but I see now that I made that statement rather unclear (English is my second language). That is, aliasing makes it less like hardware (at least in theory, perhaps it's really so low as to not be audible). Removing it would make it more like hardware. Sorry for the confusion.
Stolle
User Level VI
User Level VI
 
Posts: 68
Joined: Mon Mar 29, 2010 3:27 am

Re: 2 notes about a gearslutz thread

Postby TranscendingMusic » Mon May 31, 2010 10:44 pm

Stolle wrote:
TranscendingMusic wrote:By the way when you said this: "It makes it less like hardware right?" you meant "more like hardware"...correct?

Cool, we seem to agree about this. I actually meant less like hardware but I see now that I made that statement rather unclear (English is my second language). That is, aliasing makes it less like hardware (at least in theory, perhaps it's really so low as to not be audible). Removing it would make it more like hardware. Sorry for the confusion.


Oh I see what you were saying, okay I understand now.

Funny thing though, and still important to keep in mind, that low levels and small amounts of aliasing as seen through these analysis, could in fact even look similar to the hardware analysis counterpart. When factoring in THD, noise, and a/c harmonics vs. small and low level of aliasing, it all comes back to then what each tool sounds like - good or not - at the end of the day.
mixing | mastering
Win 10 x64 | Sonar Platinum x64 | 3930K(OC)
User avatar
TranscendingMusic
Expert
Expert
 
Posts: 1132
Joined: Sat Mar 27, 2010 7:01 am
Location: USA

Re: 2 notes about a gearslutz thread

Postby Stolle » Tue Jun 01, 2010 1:01 am

You are right about the noise aspect of hardware. I for one have a hum at 50hz and octaves of that which looks like aliasing in my hardware. Still fun to see the results of running a sine through hardware. I examined all my distortion pedals. From this I gather that Nebula would need around 22 kernels to capture a Fuzz face. Lets round it off to 20. That would mean double the processing power compared to the programs we have today. Considering the power of today's processors that doesn't seem impossible. Unfortunately I remember seeing Gian saying something that there are other reasons for NAT not being able to capture more than 10 kernels as of yet. Something about noise.
Stolle
User Level VI
User Level VI
 
Posts: 68
Joined: Mon Mar 29, 2010 3:27 am

Re: 2 notes about a gearslutz thread

Postby TranscendingMusic » Tue Jun 01, 2010 1:25 am

Stolle wrote:You are right about the noise aspect of hardware. I for one have a hum at 50hz and octaves of that which looks like aliasing in my hardware. Still fun to see the results of running a sine through hardware. I examined all my distortion pedals. From this I gather that Nebula would need around 22 kernels to capture a Fuzz face. Lets round it off to 20. That would mean double the processing power compared to the programs we have today. Considering the power of today's processors that doesn't seem impossible. Unfortunately I remember seeing Gian saying something that there are other reasons for NAT not being able to capture more than 10 kernels as of yet. Something about noise.


Yes it's definitely fun and still, again partially useful to run these types of tones for testing, agreed. And that's a very good example of an analogous(no pun intended) element to aliasing.
That's exactly right about the Kernel implementation and noise. The technological side of it is a perpetual condition. At the rate we are going with computing power, it will eventually be a non-issue. But making the sampling process better so to speak in order to distinguish between noise and distortion Kernels will be an amazing step towards implementing more Kernels. It is inevitable however!
mixing | mastering
Win 10 x64 | Sonar Platinum x64 | 3930K(OC)
User avatar
TranscendingMusic
Expert
Expert
 
Posts: 1132
Joined: Sat Mar 27, 2010 7:01 am
Location: USA

Re: 2 notes about a gearslutz thread

Postby giancarlo » Tue Jul 20, 2010 11:26 am

today I gave a glance to the topic:
http://www.gearslutz.com/board/new-prod ... os-32.html


We released 4kg to heaven, a free preset for nebula3 free (so all users) in 2007. Nebula was downloaded hundreds of thousands of times, and that particular preset was used A LOT. It's an S*L console sampled in spain. 4kg to heaven requires around 10 megabytes, we were criticized for the cpu load by kvr-anians, but nowadays you could load tons of instances using that preset.

Alexb is creating complex models using a brute force approach, sampling at 96Khz, and taking in account a lot of variables, for example forcing incredible values for kernel lengths. It means alexb consoles need memory, and they need cpu cycles. Here it's not a matter of workflow. If you discard constraints, you could get something in realtime for all your channels. It's incredible that we are speaking about that, since a console emulation should not require a realtime processing, you should just render tracks, but this should be a different topic. Even the bus-group channel should not matter so much, since they are not relevant for the final result (I think the improvement is very little, once you have processed all input tracks). Even crosstalk argument is not so relevant, since it's very small in professional analog consoles. So a daw implementing summing doesn't lead to different result than processing individual tracks.

Here it's matter of what's closer and which is the cost for it, and I'm speaking about dynamic behaviour and dynamic harmonic distortion. If workflow is your issue and you don't care the sound you could ASK to alexb a release of CLC with a limited set of samples and a smaller kernel length. It's possible and the resulting model should be a good compromise if you plan to use it in realtime.

I'm sure a lot of you, guys, will not perceive the difference, especially processing a busy rock mix.
User avatar
giancarlo
Founder
Founder
 
Posts: 9171
Joined: Mon Sep 21, 2009 10:40 pm
Location: Italy

Re: 2 notes about a gearslutz thread

Postby mertayy » Tue Jul 20, 2010 11:56 am

Someone should one day write how to use Nebula and the logic behind it in the most simple terms so discussions can be avoided. We had a big Nebula thread once there and it was useful for many, it would be nice to start a new one.

I mean I have no interest in Stevens products but such arguments is not good for anyone in that thread or somewhere else.
mertayy
Member
Member
 
Posts: 244
Joined: Sun Mar 28, 2010 6:23 pm

Re: 2 notes about a gearslutz thread

Postby giancarlo » Tue Jul 20, 2010 4:11 pm

well, you can't stop people... a forum is a free place. If a developer announces a product, other people will start comparing it with similar products. This is the purpose of a forum: when we announced nebula in 2007 ALL threads were crowded with comparison with other products, I was not surprised about that.
What is not correct is that someone writes something negative about a product in a topic where you can't reply because it's not your topic.
User avatar
giancarlo
Founder
Founder
 
Posts: 9171
Joined: Mon Sep 21, 2009 10:40 pm
Location: Italy

PreviousNext

Return to Working with Nebula

Who is online

Users browsing this forum: No registered users and 1 guest