Login

Transient loss

Tips & tricks, working results, technical support

Re: Transient loss

Postby edbilleaud » Sat Mar 31, 2012 4:47 pm

voidar wrote:I did some null tests.

On my setup these are more successfull (closer null) using the FREQD kernel mode. TIMED leaves a lot of high-end while FREQD leaves some low-end mud

Just out of curiousity I sampled a channel-bus instance of Slate VCC and then ran Nebula Reverb with this program against the VST.

Strangely cancellation was best with no oversampling on the VCC side while I am pretty sure the program was rendered at 8x oversampling.


Pardon my denseness, but I don't understand. Nulled against what? And, what did you do with VCC exactly? Sample it into Nebula?
edbilleaud
User Level VIII
User Level VIII
 
Posts: 89
Joined: Fri Sep 03, 2010 8:38 pm

Re: Transient loss

Postby MOjO_FET » Mon Apr 02, 2012 2:20 pm

Hi, I want to add that since reading this thread I went back to tweaking Nebula inside, which I previously never really understood.

And, it does make a difference to me! Like now I get the analogue mojo I know from the hardware. The sound gets real character I can shape and sculpt in a creative way I could not previously achieve with algos.

I don't even care that much about the CPU consumption. Although it means freezing almost every track. But the sound makes me smile. I have a i7-2630QM 2GHz, 16GB DDR3 RAM with a gForce GT555M with 4GB's laptop which can run around 4-6 instances @ 48K 24Bit in Cubase 5.5 on timed for clean odd and even between 10 to 30 ms. I'm not sure yet whether there is a difference between 10 or 30 or even 50 ms. It's probably program dependant.
It would be nice if this whole topic got more explained and I mean every aspect of it so anybody can understand it. I think exactly this tweak makes Nebula what we always wanted it to be.
Michael
User avatar
MOjO_FET
User Level V
User Level V
 
Posts: 52
Joined: Tue Mar 30, 2010 11:48 am

Re: Transient loss

Postby voidar » Mon Apr 02, 2012 11:23 pm

edbilleaud wrote:Pardon my denseness, but I don't understand. Nulled against what? And, what did you do with VCC exactly? Sample it into Nebula?


Yes, I sampled a VCC setting using NAT and then proceeded with null-testing the program to the VST with a complex and dynamic signal.
voidar
User Level III
User Level III
 
Posts: 33
Joined: Thu Dec 23, 2010 11:20 am

Re: Transient loss

Postby yr » Tue Apr 03, 2012 1:00 pm

There seem to be more artifacts when all kernels are switched to "timed" (which can explain the null-test results). Here are some images of the harmonic distortion with the different options:

Timed clean
timed clean.PNG
timed clean.PNG (130.09 KiB) Viewed 1509 times


Timed clean&even
timed clean&even.PNG
timed clean&even.PNG (131.41 KiB) Viewed 1509 times


Timed clean even&odd
timed clean even&odd.PNG
timed clean even&odd.PNG (135.51 KiB) Viewed 1509 times
Reuven | post-production & sound-design | scenography |
website | nebula presets
User avatar
yr
Beta Tester
Beta Tester
 
Posts: 438
Joined: Mon Mar 29, 2010 9:04 am
Location: Amsterdam

Re: Transient loss

Postby voidar » Tue Apr 03, 2012 2:02 pm

That "thorn"-like result is what I allways got under the scope when using the non-reverb version.
voidar
User Level III
User Level III
 
Posts: 33
Joined: Thu Dec 23, 2010 11:20 am

Re: Transient loss

Postby darren » Tue Apr 03, 2012 4:10 pm

mitch wrote:Hi all! First of all, a huge thank you to all for having talked about this tweak!!! It is not so complicated at the end and it really is making things much more real!
Before the tweak, I found myself using more Nebulas to achieve a certain sound, because things where getting a bit "muddy" when using many instances on a source.

In my experience on the cpu side is that:
-I bought an i5/16gig/SSD computer especially for Nebula, and switched to Reaper.
-Before the timeD-10ms/EFV17 tweak, I was able to run around 40-50 instances at the same time maximum; but strangely the crashes occured when the cpu meter showed only 40%cpu load on Reaper.
-now, I can run the same amount of Nebulas before overload, but the cpu meter shows 80% load, which is more logical, AND IT SOUNDS BETTER :D

For me Nebula is very heavy for a plugin anyway,not so cheap (if you count the libraries, the pc, Reaper)
and not so easy to understand in the beginning.
Then the install/licence scheme is not perfect neither.

I have choosen it and even bought an expensive Pc because I love the sound, and I think a big part of the userbase has the same motivation. So having a "normal" mode with the best possible sound without tweak (at least with 10ms for timeD) sould be the basis, and then the economic mode should switch the timeD/freqD setting for low cpu consumption.
For some people tweaking the plugin while having no idea how to is frightening, and they won't read all the posts of the forum...

Ayway thank you again for the tweak and for Nebula of course!!!


Mitch, could you maybe show us some screenshots of your nebula parameters, this way it's easier to make sense of what you have done and try these things out?
User avatar
darren
User Level X
User Level X
 
Posts: 104
Joined: Wed Dec 08, 2010 6:46 pm

Re: Transient loss

Postby edbilleaud » Wed Apr 04, 2012 7:35 pm

darren wrote:
Mitch, could you maybe show us some screenshots of your nebula parameters, this way it's easier to make sense of what you have done and try these things out?


YES PLEASE! Some screen shots showing EXACTLY what and how to apply all these tweaks, and maybe with some commentary, would be SO APPRECIATED! Something the "average" person could understand! Many thanks in advance!
edbilleaud
User Level VIII
User Level VIII
 
Posts: 89
Joined: Fri Sep 03, 2010 8:38 pm

Re: Transient loss

Postby mitch » Thu Apr 05, 2012 9:13 am

To Daren and Ed: No prinscreen here, I am such a computer beginner, that I even don't know how to do it :lol:
I just took the time to read carefully all the posts in this thread, and then made a sum-up of all the posts that did provide something useful, and then I made a test and here you go, not so hard (I am a beginner too!)

First of all, save all of your "original" program files on another place, just in case you screw-up anything.(these programs files are not big compared to the vector files, so it is a fast thing to do)
You don't need to rename anything, as long as you make the same tweaks on all of your programs, and you keep a backup program file untouched on a hard-drive just in case.

For both tweaks, you can just follow the post of FPoitevin on page 6 of this thread.(whom I thank for his excellent description by the way!)

There are two tweaks:
1)changing from any other mode (mainly rms) to evfs
This is a simple matter of finding where the setting is, and then clic on save for each program at the end of the process.


2)changing the algorithm of the transients for the sounds which go through a certain program, example: "2055Angels LO".
This process is in the page "kernels".

The library designers give generally other combinations than these ones, because they want to reduce the cpu consumption.
But to get a better result, we want to change each programm's Kernel settings (in milliseconds) to:
timeD: 10ms (calculation in TimeD for the first 10ms, and then calculation in FreqD for the rest of your sound).
You could put bigger timeD settings, but then it could be too high on cpu use for the result you get.
I didn't bother about the other settings (FreqD).

And again, for both tweaks, follow the post of FPOITEVIN
on page 6 that tells it all!

have a nice day!
mitch
User Level IV
User Level IV
 
Posts: 45
Joined: Thu Apr 07, 2011 1:08 pm

Re: Transient loss

Postby musicgreator » Thu Apr 05, 2012 12:13 pm

is 10ms the minimum for TIMED or do you think a little less is beneficial too?? I have 7.2ms right now.
Last edited by musicgreator on Sun Apr 08, 2012 9:38 am, edited 1 time in total.
musicgreator
User Level XI
User Level XI
 
Posts: 161
Joined: Sun Mar 28, 2010 5:16 pm

Re: Transient loss

Postby edbilleaud » Sun Apr 08, 2012 4:18 am

mitch wrote:To Daren and Ed: No prinscreen here, I am such a computer beginner, that I even don't know how to do it :lol:
I just took the time to read carefully all the posts in this thread, and then made a sum-up of all the posts that did provide something useful, and then I made a test and here you go, not so hard (I am a beginner too!)

First of all, save all of your "original" program files on another place, just in case you screw-up anything.(these programs files are not big compared to the vector files, so it is a fast thing to do)
You don't need to rename anything, as long as you make the same tweaks on all of your programs, and you keep a backup program file untouched on a hard-drive just in case.

For both tweaks, you can just follow the post of FPoitevin on page 6 of this thread.(whom I thank for his excellent description by the way!)

There are two tweaks:
1)changing from any other mode (mainly rms) to evfs
This is a simple matter of finding where the setting is, and then clic on save for each program at the end of the process.


2)changing the algorithm of the transients for the sounds which go through a certain program, example: "2055Angels LO".
This process is in the page "kernels".

The library designers give generally other combinations than these ones, because they want to reduce the cpu consumption.
But to get a better result, we want to change each programm's Kernel settings (in milliseconds) to:
timeD: 10ms (calculation in TimeD for the first 10ms, and then calculation in FreqD for the rest of your sound).
You could put bigger timeD settings, but then it could be too high on cpu use for the result you get.
I didn't bother about the other settings (FreqD).

And again, for both tweaks, follow the post of FPOITEVIN
on page 6 that tells it all!

have a nice day!


thanks Mitch, that's very kind of you.
edbilleaud
User Level VIII
User Level VIII
 
Posts: 89
Joined: Fri Sep 03, 2010 8:38 pm

PreviousNext

Return to Working with Nebula

Who is online

Users browsing this forum: daniddd and 6 guests