Beginner's Tips For Infinite Linear Oversampling?

Discussion in 'REAKTOR' started by mike_jm, Sep 7, 2017.

  1. mike_jm

    mike_jm Member

    Messages:
    130
    After trying 8x oversampling on TT amp and being blown away by the quality improvement but also the horrible crushing CPU burden (an extra 20% of my entire quad core computing power per instance), I'm sold on learning Infinite Linear Oversampling.

    I saw this also from Efflam's thread which sell it pretty well also:

    [​IMG]
    https://www.native-instruments.com/...tive-method-to-oversample-waveshapers.309111/

    Let's say you have a simple project like this where you want to oversample a simple primary module like the Saturator:

    [​IMG]

    What do I need to calculate/measure, what formula do I need, how/where do I calculate it on Wolfram, and where do I plug it into the Core macros in the Reaktor project?

    https://www.native-instruments.com/en/reaktor-community/reaktor-user-library/entry/show/10381/

    If it's too hard to explain no worries. You guys have helped me enough already. But I'd really like to learn. Thanks.
     
  2. colB

    colB NI Product Owner

    Messages:
    1,835
    It's all in Efflam's upload. There is a layman's explanation in a pdf in the zip file, and the description in the UL page has a link to a thorough academic paper on the subject.
     
  3. salamanderanagram

    salamanderanagram NI Product Owner

    Messages:
    3,203
    you need to find a saturation function, then have wolfram alpha integrate it for you, for example

    y = tanh x

    instead you want to type in

    integrate y = tanh x

    then you have the equations you need to implement for efflam's tutorial. alternately, i have several videos going through the process in full if you want me to dig them up.
     
  4. mike_jm

    mike_jm Member

    Messages:
    130
    I would love to see your tutorials. Just to be clear, though, this is what I'm wondering - this would not be directly compatible with a primary module like "Saturator", right? ie. It has to be implemented directly within core with a matching core saturation process?

    So this would work most directly with something like too-B. But to apply to something like TT amp which uses the saturation primary module, you'd have to replace the entire saturation module with core (like too-B) to get the right integration to match the saturation you're applying?

    I'd love to see your vids on the subject - I googled but couldn't find them.

    ColB, thanks,I will read the article again. It went a bit over my head the first time. Did you say you were experimenting with a tanh saturator already with this function? If so, might you be willing to share? I'd love to hear it. Should be like too-B with more efficiency. A few of those in serial sequence should give a very good tube saturation which is I think what your amp sim needs for the low gain stuff.

    You mentioned you were not happy with your prior amp sim result on low-mid settings.

    Try running two or three too-B's in sequence already and you'll find it sounds very nice on low gain stuff. With some damping filters per "tube" like TT amp offers and tube compression following the preamp saturation like TT offers, you could likely get a nice effect for low CPU cost.
     
    Last edited: Sep 7, 2017
  5. salamanderanagram

    salamanderanagram NI Product Owner

    Messages:
    3,203
    here's one, i think there's at least one more.

    https://www.reaktortutorials.com/media/anti-aliased-waveshaper/61160

    i like this particular waveshaper function because you can put it into the feedback path of a ZDF filter and still solve the filter equation - it's the only non-linear waveshaper function that i know of with this particular property.

    yeah, you can't use this with any 'black box" modules that you don't know the actual equations for.
     
    Last edited: Sep 7, 2017
    • Like Like x 1