Login

Henry Olonga releases many new products!!!

Officially Licensed 3rd Party Developer Libraries
Free 3rd Party Programs

Re: Henry Olonga releases many new products!!!

Postby Henry Olonga » Sun Mar 23, 2014 2:34 am

Not quite. To clarify - everything is done within the DAW session. I have changed all the bit depth options to 64 bit. So I will ( from now on anyway ) mix, record, render and import any files in 64 bit 44.1 khz if I am appealing to maximum fidelity. I had bought a new fast HDD just for this project so space is a non issue.
When I talk about a bounce down it is within a project to a new track. For example a piano vsti instrument track bounced down from midi. Or a bass with nebula processing - all happening now at 64 bit. The best benefit is to softsynths imho.
If I import a loop I will import at 64 bit.
What I was hinting at is that the different bit depths may have been leading to some very low level changes to the sound that I could sense and hear. I am no math guru but my ears tell me it is so. What it is, is that the 64 bit workflow preserves the sound as I hear it - not improves it. If I render to a lower bit depth it appears to degrade. So it is more about preservation than improvement. The mix engine works at 64 bit but my renders back then were to a lower bit depth @ 32 bit and then played back at the higher bit depth again. Most noticeable when I bounce down through nebula to 32 bit ( with the old setting. ). The sound always felt somewhat missing something. I don't mean to get into a debate about this as this is quite subjective. What I am saying is perhaps it is better math to stick to one bit depth throughout. I always felt it was a none issue but now. . . . hmmmmm it got my attention. And also it may be that 64 bit files potentially sound better played back in 64 mix engine even if they were acquired at a lower bit depth. Do your own tests on that one. So whether it is 64 or 32 bit, sticking to one throughout may be wise. But we can all try this easily if you have DAW with a 64 bit engine like Sonar.

Open a blank project with the 64 bit mix engine engaged - load a track of a harmonically complex instrument like a piano or distorted guitar - add a few instances of nebula with some easily audible presets - render the stem to 32 bit - compare the realtime playback against the render with high quality monitoring or headphones. Repeat and this time render to 64 bit. Compare to realtime and 32 bit render. Experts may well say that they should sound the same but I would like to here what you think. . . .it's one of those things that many folk never really test.
I really hope this is easy to follow and not confusing. . . . .in simple terms . . . I am the happiest I have been with ITB mixing when I changed all bit depth settings to 64 bit with the 64 bit mix engine engaged.
User avatar
Henry Olonga
Expert
Expert
 
Posts: 572
Joined: Wed Mar 31, 2010 1:38 am
Location: South west England

Re: Henry Olonga releases many new products!!!

Postby jorismak » Sun Mar 23, 2014 3:05 am

Vst (2.x) works internally at Floating point 32bit as far as I know (there is a 64bit-extension but a lot of plugs don't work with it).

IMHO if you're input start at 24bit (which is integer), you're (more than) OK working in 32bit floating point. Floatingpoint keeps things like overs intact (it can keep the detail if you clip for instance, since it can represent a value over 1.0 no problem). It can keep all the detail from your 24-bit integer recorded inputs.

A 'mistake' I see often is that bounces get written to 24-bit integer because the input was 24-bit when they started. The moment you start doing _anything_ with it in a DAW (panning / volume / plugin / grouping / bussing) you're working in 32- (or 64-) bit floating-point, you you're bounces should be in that too.

The sound coming 'more open', 'more depth' when switching to 64 bit doesn't sound right to me at all :). Maybe there are problems somewhere or things get rounded wrong. I've done projects before that the final render to dithered 16-bit integer are _bitwise_ exactly the same when working in 64-bit or 32-bit (no bounces in the project, was for test). But who knows what plugins react to 64-bit in a special way.

But the thing is, you like what it does with the sound and it gives you confidence in your workflow, and that's all that matters in the end :).
jorismak
Member
Member
 
Posts: 344
Joined: Fri Nov 16, 2012 4:49 am

Re: Henry Olonga releases many new products!!!

Postby Henry Olonga » Mon Mar 24, 2014 11:13 am

If I were to pose a question it would sound like this.

'Does using multiple lower bit depth settings in a Sonar project sound inferior to using a 64 bit pipeline throughout? In my humble opinion the answer is yes. Your answer may differ but I have tested it on a full project and it made me say - ummmmmm now we're talking - why did I not do this before?

As far as theory vs practice goes - this thing we do is mainly art and it is all about nuance and aesthetic and tiny benefits from increasingly expensive tools. For some, they ignore the law of dimishing returns and will pay £40 000 for an A*I console. The UAD guys will have you believe it sounds identical - so why bother? Well something matters to them that they will get a loan to buy the thing - gulp. I have only one A*I channel strip and wish I could afford 16. I get why hardware is amazing. I bought my first sable paint brush set that was awefully expensive, could have bought a set ten times cheaper. They're just paintbrushes, but I asked the shopkeeper, what is the set that will give me the best results effortlessly, you know the one that all the big guys buy from you? He said this sable set and that was that. Never felt I wanted better after that and of course the big guys can do better than I can with cheaper tools. Not the arrow but the Indian right? But what if this Indian needs all the help he can get because he is not a professional archer? If what I am suggesting would cost a huge amount of money then it's worth fighting over. But with HDDs costing next to nothing if I am even half onto something this is a simple change that may be considered a cheap but substantial upgrade.

Projects won't sound rubbish in 32 bit floating point of course. But if you don't have a reference because the same project has never been bounced at 64 bit then you may never know and will never miss it.

Not really fishing for a debate as I admit it is subjective. I suggest to do a bounce through a Nebula chain in the method I described and report back to me as that is my main focus. Remember that this is track by track not the whole mix. But you need to do this in a 64 bit engine. If it sounds identical to you then - nothing to see here really. I have read about the theory enough but my ears tell me different. And I do hope my ears are dependable. . . . a bit like the endless do daws sound different debate on various forums. For some it is a matter of pan laws. For others it is worth switching to an unfamiliar DAW even when everyone says - they sound the same. This is just a little tip for those who wish to extract a tiny subjective increase in fidelity for free and with only one hassle - hdd space. Not sure about how Reaper will test as the engine does realtime resampling wheras Sonar does not. It adds another variable to things as I see it.

Please understand that I am writing this because I believe that it has made a major difference to the feel of my projects - enough for me to consider offloading my hardware. Now for me that is a big deal as I have UAD, powercore and the best of the best plugins and never thought this way until I 'made the change'. It would make sense that if Nebula works internally at 64 bit there may be a benefit to all I am saying. It is also possible I had too much breakfast and it's fogging my mind LOL, in which case - carry on as you were.
Anyway, I hope it is of use to someone.

jorismak wrote:Vst (2.x) works internally at Floating point 32bit as far as I know (there is a 64bit-extension but a lot of plugs don't work with it).

IMHO if you're input start at 24bit (which is integer), you're (more than) OK working in 32bit floating point. Floatingpoint keeps things like overs intact (it can keep the detail if you clip for instance, since it can represent a value over 1.0 no problem). It can keep all the detail from your 24-bit integer recorded inputs.

A 'mistake' I see often is that bounces get written to 24-bit integer because the input was 24-bit when they started. The moment you start doing _anything_ with it in a DAW (panning / volume / plugin / grouping / bussing) you're working in 32- (or 64-) bit floating-point, you you're bounces should be in that too.

The sound coming 'more open', 'more depth' when switching to 64 bit doesn't sound right to me at all :). Maybe there are problems somewhere or things get rounded wrong. I've done projects before that the final render to dithered 16-bit integer are _bitwise_ exactly the same when working in 64-bit or 32-bit (no bounces in the project, was for test). But who knows what plugins react to 64-bit in a special way.

But the thing is, you like what it does with the sound and it gives you confidence in your workflow, and that's all that matters in the end :).
User avatar
Henry Olonga
Expert
Expert
 
Posts: 572
Joined: Wed Mar 31, 2010 1:38 am
Location: South west England

Re: Henry Olonga releases many new products!!!

Postby Henry Olonga » Mon Mar 24, 2014 3:32 pm

Here's another thought. When we go from 24 bit to 16 bit we have to add dither right? But when you are mixing in the 64 bit engine and bounce down to 32 bit fp with an in the box render, I assume that there is no dither and there may be truncation and some errors may creep in no matter how small. Then we effectively go back to 64 bit again on playback. I am just wondering whether that may have something to do with it. Honestly I don't know how it works internally as I don't know how the bakers at cakewalk code. But I have a new found appreciation for the sound I am getting out of Sonar when I keep the bit depth the same - in my case at 64 bit. :D
User avatar
Henry Olonga
Expert
Expert
 
Posts: 572
Joined: Wed Mar 31, 2010 1:38 am
Location: South west England

Re: Henry Olonga releases many new products!!!

Postby giancarlo » Mon Mar 24, 2014 6:08 pm

jorismak wrote:Vst (2.x) works internally at Floating point 32bit as far as I know (there is a 64bit-extension but a lot of plugs don't work with it).

IMHO if you're input start at 24bit (which is integer), you're (more than) OK working in 32bit floating point. Floatingpoint keeps things like overs intact (it can keep the detail if you clip for instance, since it can represent a value over 1.0 no problem). It can keep all the detail from your 24-bit integer recorded inputs.

A 'mistake' I see often is that bounces get written to 24-bit integer because the input was 24-bit when they started. The moment you start doing _anything_ with it in a DAW (panning / volume / plugin / grouping / bussing) you're working in 32- (or 64-) bit floating-point, you you're bounces should be in that too.

The sound coming 'more open', 'more depth' when switching to 64 bit doesn't sound right to me at all :). Maybe there are problems somewhere or things get rounded wrong. I've done projects before that the final render to dithered 16-bit integer are _bitwise_ exactly the same when working in 64-bit or 32-bit (no bounces in the project, was for test). But who knows what plugins react to 64-bit in a special way.

But the thing is, you like what it does with the sound and it gives you confidence in your workflow, and that's all that matters in the end :).



nebula is working in 64 bits mode. Sampling is not important, you could sample at 24, 32 or 64 bits. What's important is the internal precision of kernels.
Again, nebula could be used as 32 or 64 bits plugin; all plugins based on juce (the highest number in the market) will be 32 bits only.. I think think you will hear the difference really; but kernels/irs should be 64 bits or higher.
User avatar
giancarlo
Founder
Founder
 
Posts: 9160
Joined: Mon Sep 21, 2009 10:40 pm
Location: Italy

Re: Henry Olonga releases many new products!!!

Postby Shibata » Tue Apr 08, 2014 1:07 am

Henry Olonga
Sorry bro
I did mixing test with 64 and 32bit math on
SONAR X3 Producer (x64) and Reaper 4.591 (x64) with Nebula Server 1.3.681 (x64)
And in all cases I received zero with Null test after.
So, it is placebo. No chance against physics)
User avatar
Shibata
User Level VII
User Level VII
 
Posts: 77
Joined: Tue Jul 09, 2013 8:06 pm
Location: Odessa Ukraine

Re: Henry Olonga releases many new products!!!

Postby pleplo » Tue Apr 08, 2014 8:56 am

:o
User avatar
pleplo
User Level XI
User Level XI
 
Posts: 181
Joined: Tue Mar 13, 2012 10:08 pm
Location: Montréal, Canada

Re: Henry Olonga releases many new products!!!

Postby jorismak » Tue Apr 08, 2014 10:02 am

nebula uses 64bit math internally, yes. That has nothing to do with how your DAW is set and it doesn't matter if it's the 32bit/64bit version of the plugin :).

Nebula's internal precision is always the same (well, maybe the Cuda settings are different, but let's leave them out for now). 32bit/64bit in Sonar means what format the VST plugins will receive. Since most of you record at 24bit or 32bitFP, what Sonar does is scaling it up and passing 64bit FP through all the VST's and it's internal mixer. In theory this means LOADS of headroom and precision. Like I said, how wonder if it actually gains something from 32bit FP.

But as said earlier, the most important thing is that the person using it all is happy with it. If it requires 64bit - placebo or not - then do it. If it gives the artist / producer confidence in the sound quality and the work, then you'll hear that back in the performances and that's worth more than a bit of extra headroom in the science :).

Kinda the same with bands wanting to do stuff on analog tape and without computers in sight (Foo Fighters for example). Is it 'better' quality nice? Hell no. Does it mean they have to buckle down and work four times as hard and you'll notice that because of the concentration the performances are better. YES.

Slash loves analog tape for his guitar tracks, 'something in the sound'. Do you actually notice it in the final mix? Nope. Does he want a complete signal chain that way? Nope, he likes the sound for his tracks.. if it goes into ProTools after the tape machine and gets mixed with autotuned performances of others, he doesn't care :). Because he likes the sound of guitar-on-tape, he has more fun in the recording and that means better albums. That's all that matters :).
jorismak
Member
Member
 
Posts: 344
Joined: Fri Nov 16, 2012 4:49 am

Re: Henry Olonga releases many new products!!!

Postby Shibata » Tue Apr 08, 2014 1:46 pm

Yes, I totally agree that the process conditions affect the final result. But we are talking about 64 bit internal mixing engine of Sonar and Reaper, which do not affect the conditions of the process, but like Henry said, the sound after mixing with Nebula - get better.

So is not, because: Sonar and Reaper have the ability to customize three parameters
64 bit record
64 bit render
64 bit float point mixing math

I took two tracks of drums and guitar and mix it with AlexB N**e 88RS with the same settings on it.
With 32 and 64 bit engine. And also with render settings^ 32 32, 64 64 and so on.
Insert the results, sweep the phase and get the zero.

For additional testing, did the same thing on Cubase 6 with 32 FP and compare each other between DAW results and get zero)

This is not the difference, when you get the result by recording to tape or to digital DAW because of the procces itself - its just checkbox in the settings.
User avatar
Shibata
User Level VII
User Level VII
 
Posts: 77
Joined: Tue Jul 09, 2013 8:06 pm
Location: Odessa Ukraine

Re: Henry Olonga releases many new products!!!

Postby Henry Olonga » Thu Apr 10, 2014 11:18 am

Hey guys,

PRICE REDUCTION. EVERYTHING NEBULA BACK TO £1 EXCEPT BUNDLES. And I MEAN EVERYTHING!

Shibata. If it doesn't jive you can just ignore it. :) . My point is all about MIXING with DIFFERENT bit depths does something to the sound that subjectively may make things inferior. It is about the difference between REALTIME PLAYBACK and a bounce down of the file. My proposition is that by rendering in 64 bit the bounce down is the same as realtime. I also think that 64 bit math throughout makes sense in that rounding errors can get minimised. That is sound science. I did my tests in Sonar and said that other daws may behave differently mixing different bit depths and sample rates. Some folk can get caught up on null tests and can get quite convinced it is the end of the matter. BUT THIS IS ART and my ears tell me different - I listen to reverb tails, sibilance etc I have a very, very hi resolution monitoring chain and can hear, things that get missed by lower res systems. I can hear low level detail that most monitors miss. Some guys get caught up on the science and in this day and age any tools are usually top notch such that we are splitting hairs on tweaks like this. There's a guy on GS forum who insists that Prismsound Orpheus is not so good compared to his soundcard. Says the clocking is not so good - he tested it nulling against itself ten times and his card measured better. I had to smile. I mean Prismsound make the gear that tests everyone else's clocks and his is a professional piece of gear for sure - but in his eyes his tool was technically better because of the 'infallible' null test. But no deterring the guy, everyone who owns Orpheus got ripped off. Me - I have owned a few high end soundcards RME, Mykerinos, TC Electronic, EMU. They measure with similar specs but wow can the sound vary quite a bit to these ears and folk know this in Pro audio. We act as if small audible differences in analog are super important ( paying tens if thousands for exotic sounds ) and then do the opposite when we use digital tools - actually many digital tools do measure similar but sound different. One of our developers did a test with the Sonitus Eq and some other Eqs many more times expensive. They nulled. But they don't all sound the same do they?

But you are here on a forum where we CAN hear subtle differences and appreciate them to the point where even we invest in a less than perfect workflow because of the results. :)

I noticed you never mentioned what you HEARD - I mean that is a glaring omission in my book. What did it SOUND LIKE? And when you start to add up many tracks across a mix - what did it SOUND LIKE? Come on brother - not one or two tracks here and there. A whole mix. We are arguing about whether there are audible differences and you never said whether you HEARD differences at all - just measurements. I hope this forum is about better hearing and mixing technique. For many folks Nebula does sound better that it is worth jumping through many hoops to get there. I hope I have good hearing left. I can hear everything with my setup. A soft sneeze at the back of a concert hall. It's in the ears. I would not bother telling my customers about a technique I believe can help them get better results if I didn't think it made a difference. You know I believe I can hear how Sonar sounds different to Reaper in realtime. Some cannot. After a bounce the files null. But they sound different in realtime to MY ears. Forums argue forever on this point - whether different DAWS sound different. I think in realtime they do. Perhaps has to do with ASIO buffering implementation. That's way above my level of research . I don't seemingly care about the technical aspects - the tech specs are OFF THE CHARTS in modern pro audio - we have it that good. Nothing is rubbish out there. I just wish for the best sound and if doing something as simple as changing a few boxes can get me closer to an analog sound then I will do it. Perhaps the real placebo is believing that the null test has everything to do with what we ear in the analog domain - using our ears to hear soundwaves.

If this adds nothing to your results then you can walk on by. There may only be a few other mad guys who agree with me, but I don't see how your test adds anything - we don't mix one or two tracks and don't listen to phase scopes. I costs nothing more that HDD space. That's it. It gives you tonnes more headroom in the mix engine and to these ears subjectively gives superior results. Anyway, I think I have flogged this horse enough . . . . carry on :)
Last edited by Henry Olonga on Thu Apr 10, 2014 12:02 pm, edited 4 times in total.
User avatar
Henry Olonga
Expert
Expert
 
Posts: 572
Joined: Wed Mar 31, 2010 1:38 am
Location: South west England

PreviousNext

Return to 3rd party libraries

Who is online

Users browsing this forum: No registered users and 8 guests