What's new

AET velocity morphing

jesusginard

Active Member
Working on a piano library with 8 velocity layers.
Although they are all normalised and velocity controlled volume, you can still hear clearly the change from one layer to the other one.
Thinking about how to smooth the transition I realised that Kontakt has a feature for it: AET Filter. I tried it and the CPU went from 10% to 40% or more, which is a big shame since it sounds very nice!
Anyone's got a solution for this or it's just the way that it is?
 

Light and Sound

Developer
Most orchestral libs don't use velocity AET as far as I am aware (for a few reasons, though CPU usage certainly being one of them). The difference between them is normally something you try to smoothen out during the recording process, ensuring that each different layer recording is as close to the last as possible, while still progressively moving forward.

You can sometimes smoothen the effect with an eq which is also controlled by velocity, though I personally think it's better to try and ensure you get the best recordings possible first and only use EQ as a last resort.

The way velocity controls the dynamic of each sample is something else you might want to look at, for example if one sample is 1-30 and another 31-40, within your DAW create a repeating pattern of one note velocity 25, 26, 27 etc, up to 35 or so and just stick it on repeat, and tweak the volumes curves within kontakt (assuming you use them instead of handling it via script) until they sound like a natural increment.

Lastly you might find that if you do not normalize the samples, you will get better results.
 

Farkle

Senior Member
Working on a piano library with 8 velocity layers.
Although they are all normalised and velocity controlled volume, you can still hear clearly the change from one layer to the other one.
Thinking about how to smooth the transition I realised that Kontakt has a feature for it: AET Filter. I tried it and the CPU went from 10% to 40% or more, which is a big shame since it sounds very nice!
Anyone's got a solution for this or it's just the way that it is?


I've been playing with the AET filter, it's VERY nice. And, unfortunately, it's a CPU hog. It's an extremely advanced and granular FFT filter, so it chomps up your processor. I'm only using it on a couple of samples, and it sounds amazing, but I don't think we can get around the CPU's it takes to run it.

I bounce my buffer up from 128 to 256, and that streamlines it. That's probably the only way to deal with it; increase your buffer to help Kontakt out.

Mike
 

mmorgan

New Member
You will get problems if you use 48khz samples

I'm a noob so excuse my ignorance on this. Is this because Kontakt is expecting 44.1khz and if I sample at 48 it will need to dither down? I would think that Kontakt would down sample (dither) when the samples are brought in and that would be the end of it.

Just starting with sampling so this is an important thing for me...I usually record at 48khz (occasionally do film work)

Thanks
 

d.healey

Senior Member
I'm a noob so excuse my ignorance on this. Is this because Kontakt is expecting 44.1khz and if I sample at 48 it will need to dither down? I would think that Kontakt would down sample (dither) when the samples are brought in and that would be the end of it.

Just starting with sampling so this is an important thing for me...I usually record at 48khz (occasionally do film work)

Thanks
It's because of a bug that has yet to be fixed by NI
 
Top Bottom