Transient loss

Tips & tricks, working results, technical support
User avatar
Tim Petherick
Expert
Expert
Posts: 1774
Joined: Sat Apr 17, 2010 4:07 pm
Location: Bath , Uk

Re: Transient loss

Post by Tim Petherick » Thu Mar 01, 2012 3:10 am

Well i was going to start a new thread on this ,but worried that it might get confusing to new users.

nebula regular to me sounds better than reverb version for eqs, tapes and pre sort of sampling ,i don't know why this is.

In general you want to stick to timed rather than freqd for higher quality. That also could mean switching off split mode because the first part of the sample is played by timed then the rest is freqd. For example 2.2ms on timed then the rest on freqd.

If you wanted to play the same length sample in timed as freqd , say the original was 50ms in
freqd, you could edit the xml to go up to 50ms

If your working at 96k your computer will probably

fry! :lol:

thus why nebula has the split feature to bring down the cpu.

A high quality Rendering mode would be good as just
mentioned.
oh and the change to evf 17 should not effect cpu .

If you look at cdsoundmasters Q8 and tube booster on the evf page , they use evf 17, so its not just for compressors!!
Last edited by Tim Petherick on Thu Mar 01, 2012 4:13 am, edited 1 time in total.

ericus
Member
Member
Posts: 243
Joined: Sun Mar 28, 2010 9:37 pm

Re: Transient loss

Post by ericus » Thu Mar 01, 2012 3:55 am

Thank you for all those tips!

Does this mean we need to these changes on the programs themselves or we create a copy of the Nebula.dll and modify the associated xml file?

User avatar
Tim Petherick
Expert
Expert
Posts: 1774
Joined: Sat Apr 17, 2010 4:07 pm
Location: Bath , Uk

Re: Transient loss

Post by Tim Petherick » Thu Mar 01, 2012 4:06 am

In order to get up to 50ms in timed mode you need to edit the xml thats all. it will look like this-
<LTIMED> 30000 </LTIMED> change it to <LTIMED> 50000 </LTIMED>

for the actual timed arrow tweak etc, you need to edit each program, its not a global tweak.
if you edit programs you'll lose the original settings of the program, so only edit if your sure.

I only do these tweaks for myself and wanted to share them.

User avatar
enriquesilveti
Expert
Expert
Posts: 2861
Joined: Sun Mar 28, 2010 9:00 pm
Location: Lodi | Madrid | Buenos Aires
Contact:

Re: Transient loss

Post by enriquesilveti » Thu Mar 01, 2012 10:47 am

evf 17 in evf page.
YES. EVF 17 is the last EVF mode developed. If you have a nuclear PC you can use more and more TIMED engine, we add FREQD engine just to play long kernels from EQs/Reverbs in current PCs. You can SAVE AS programs (N2P files only) from each emulation program with different name before edit in NAT or in Nebula. Maybe Giancarlo can make a patch for auto edit in NAT, request to him...

Envelope follower in Nebula
An envelope detector is sometimes referred to as an envelope follower in musical environments. It is still used to detect the amplitude variations of an incoming signal to produce a control signal that resembles those variations.
Nebula use an envelope follower to translate incoming amplitude audio signal into a value that help to select the correct kernel that will process the incoming signal with the correct filter but this filter has a length that is more than one sample that is the reason that Nebula or any signal modeling processor can emulate correctly a peak limiter with out introduce audible distortion.
Enrique Silveti.
Acustica Audio customer and technical support.

MBP 11.5 (W10 1803 | OSX 10.13.6 | i7-4870 | 16 GB | 512 GB SSD)
SP4 (W10 1903 | i5-6300 | 8 GB | 256 GB SSD)
Digiface Dante | Lyra2 | Rednet AM2
VM U18 | VM W10 1803 | VM OSX 10.13
N4/NAT4 | SPX4 | RX7 | LN2C | SMAART8 | R5 | PT2019 | PX11 | NIK5

fredrikberg
User Level III
User Level III
Posts: 36
Joined: Wed Mar 31, 2010 7:21 pm

Re: Transient loss

Post by fredrikberg » Thu Mar 01, 2012 12:14 pm

babiuk wrote:I´m reading this topic and I find it highly interesting because I can see I am not using Nebula benefits at 100%.
Anyway it´s a bit confusing to me. Could you help me to understand?
When I read this:
One could set "Split mode with 10ms kernels on timed and Evf17" as render mode.
Can I assume (CPU costs apart) that this is the best nebula setup for every preset?
And... is better (again CPU costs apart) using nebula reverb all the time instead nebula?
I asked about this because I felt some transient loss, and with these tweaks it sounds SO much better.
I think the initial transient of a sound is the most important for separating things in a mix and also for really feeling a room in drum room microphones etc...

I don´t understand all the parameters in Nebula but it seems to me that when (in Kern page) you set timed to 10ms and point the arrows to Freqd you get timedplayer for the first 10ms and freqdplayer for the rest, so the critical part of the sound is processed by the most accurate player.

I felt when doing this tweak to VXNT plate reverb there was a difference in the reverb "impact", glued better with the source in some way.

Im very happy for all the help I have got.
Thank you!

Regards

Fredrik Berg

Cupwise
Expert
Expert
Posts: 1046
Joined: Tue Nov 23, 2010 2:03 am

Re: Transient loss

Post by Cupwise » Thu Mar 01, 2012 1:53 pm

fredrikberg wrote: I don´t understand all the parameters in Nebula but it seems to me that when (in Kern page) you set timed to 10ms and point the arrows to Freqd you get timedplayer for the first 10ms and freqdplayer for the rest, so the critical part of the sound is processed by the most accurate player.

Fredrik Berg
i might be misunderstanding you, but i think in the situation you described you would only be in classic mode, and only using the freqd player. if both timed and freqd are being used (split mode) there should be an arrow on both.

fredrikberg
User Level III
User Level III
Posts: 36
Joined: Wed Mar 31, 2010 7:21 pm

Re: Transient loss

Post by fredrikberg » Thu Mar 01, 2012 2:08 pm

Did I get it right now:

If you´re sure kernels are less then 10ms then you could use Classic mode with 10ms on timed and arrows pointing on Freqd

If you are not sure how long the kernels, use Split mode with timed set to 10ms.

Am I Right?

User avatar
enriquesilveti
Expert
Expert
Posts: 2861
Joined: Sun Mar 28, 2010 9:00 pm
Location: Lodi | Madrid | Buenos Aires
Contact:

Re: Transient loss

Post by enriquesilveti » Thu Mar 01, 2012 2:13 pm

fredrikberg wrote:I don´t understand all the parameters in Nebula but it seems to me that when (in Kern page) you set timed to 10ms and point the arrows to Freqd you get timedplayer for the first 10ms and freqdplayer for the rest, so the critical part of the sound is processed by the most accurate player.
http://www.acustica-audio.com/forum/ind ... =viewtopic
Enrique Silveti.
Acustica Audio customer and technical support.

MBP 11.5 (W10 1803 | OSX 10.13.6 | i7-4870 | 16 GB | 512 GB SSD)
SP4 (W10 1903 | i5-6300 | 8 GB | 256 GB SSD)
Digiface Dante | Lyra2 | Rednet AM2
VM U18 | VM W10 1803 | VM OSX 10.13
N4/NAT4 | SPX4 | RX7 | LN2C | SMAART8 | R5 | PT2019 | PX11 | NIK5

fredrikberg
User Level III
User Level III
Posts: 36
Joined: Wed Mar 31, 2010 7:21 pm

Re: Transient loss

Post by fredrikberg » Thu Mar 01, 2012 2:19 pm

Thank you!

That explains it very well.

Cupwise
Expert
Expert
Posts: 1046
Joined: Tue Nov 23, 2010 2:03 am

Re: Transient loss

Post by Cupwise » Thu Mar 01, 2012 3:04 pm

if you want to use timed only, the arrow should be on timed, if you want freqd it would be on freqd, for split there would be an arrow on both.

changing the kern length on timed will have no effect if the arrow is on freqd.

fredrikberg
User Level III
User Level III
Posts: 36
Joined: Wed Mar 31, 2010 7:21 pm

Re: Transient loss

Post by fredrikberg » Thu Mar 01, 2012 3:16 pm

Of course, thats how it is set on my tweaked presets, I just messed it up in my mind trying to explain.

Raynorshyn
User Level IV
User Level IV
Posts: 41
Joined: Wed Mar 31, 2010 5:15 am

Re: Transient loss

Post by Raynorshyn » Thu Mar 01, 2012 3:41 pm

Ok... I'm still a hair confused here:

Timp says "point the arrows so they are going towards freqd player lengths.

Then increase the timed kernals in order to complete the sample lengths"

Then Cupwise says "if you want to use timed only, the arrow should be on timed"

In Classic mode, or hybrid, which way should the arrow point to use the timed engine on the kernel?

User avatar
misterambient
User Level V
User Level V
Posts: 56
Joined: Sat Mar 05, 2011 3:11 am
Location: The Netherlandz
Contact:

Re: Transient loss

Post by misterambient » Thu Mar 01, 2012 4:23 pm

Thanks for all the valuable info. I'll try to summarize:

For best results:

Nebula for tapes, preamp, eq and compressors
Nebula Reverb for consoles, TB+, reverb, delay

Use split mode with 10ms kernels on timed and Evf17.

Evf17 can be set globally.

For the actual timed arrow tweak etc, you need to edit each program, and use SAVE AS (N2P files only) before edit.

These optimizations apply to Nebula and Nebula Reverb.

Work in 96kHz because of the slight src quality reduction.

fredrikberg
User Level III
User Level III
Posts: 36
Joined: Wed Mar 31, 2010 7:21 pm

Re: Transient loss

Post by fredrikberg » Thu Mar 01, 2012 4:44 pm

Raynorshyn wrote:Ok... I'm still a hair confused here:

Timp says "point the arrows so they are going towards freqd player lengths.

Then increase the timed kernals in order to complete the sample lengths"

Then Cupwise says "if you want to use timed only, the arrow should be on timed"

In Classic mode, or hybrid, which way should the arrow point to use the timed engine on the kernel?
The arrow should point to Freqd for best result

Timed > Freqd

Try switch to Economic mode and you see the arrows turning to Timed

Timed < Freqd


Also, Giancarlo told me in a mail that Nebula Reverb will always sound the best because of longer tails and no truncation, And it´s the only one that support multithreading.

User avatar
enriquesilveti
Expert
Expert
Posts: 2861
Joined: Sun Mar 28, 2010 9:00 pm
Location: Lodi | Madrid | Buenos Aires
Contact:

Re: Transient loss

Post by enriquesilveti » Thu Mar 01, 2012 5:34 pm

Also, Giancarlo told me in a mail that Nebula Reverb will always sound the best because of longer tails and no truncation, And it´s the only one that support multithreading.
If you set Nebula audio engine for an individual emulation preset (KERN page) to play less than the complete kernel length you will get noise due truncation, long kernel length in TIMED are not possible yet (Do you have a x32 CPU with 96 GB of RAM computer?) , so you can use split mode (SPLITH, H=kernel=harmonic) in the first harmonic called fundamental, in that case you use the best of both world, direct convolution for transients and FFT convolution for the tail and that help user to load 10-60 instances of Nebula in theirs audio sessions.
Enrique Silveti.
Acustica Audio customer and technical support.

MBP 11.5 (W10 1803 | OSX 10.13.6 | i7-4870 | 16 GB | 512 GB SSD)
SP4 (W10 1903 | i5-6300 | 8 GB | 256 GB SSD)
Digiface Dante | Lyra2 | Rednet AM2
VM U18 | VM W10 1803 | VM OSX 10.13
N4/NAT4 | SPX4 | RX7 | LN2C | SMAART8 | R5 | PT2019 | PX11 | NIK5

Post Reply