Processes that don't participate in the layer concept

It seems to me that some of the older processes (algorithmic, not AI) do not use the layers concept at all. I expected the noise reduction process, for example, to create a layer with the clean material and another layer with the noise that was removed, but that does not seem to be the case.

There could be situations where it would be helpful to repatriate some of the “noise” with the program material. I am just wondering if there is at least a long-term intention for all processes within SpectraLayers to actually use layers.

The classification is pretty clear :

  • all actions listed in the “Unmix” menu will create new layers, unmixing non-destructively
  • all actions listed in the “Process” menu will only act on the current layer, not creating any new layer
    Both the Unmix menu and the Process menu contains algorithmic and AI actions.

I am suggesting it would be very useful (and philosophically consistent) for ALL the processes to use the layers approach.

I understand what you mean, but most processes in the Process menu cannot be decomposed in 2 or more layers : think for instance about Generate Silence/Tone/Noise, Gain, Normalize, Channels Remixer, EQ Match, Imprint Cast/Mold, Ambience Match, Reverb Match…

It can work with some of the restoration processes though, maybe in the future this could be considered as a future feature. Right now, what you can do is duplicate the layer you want to process, process the first with the effect, and process the second with the reverse of the effect (most processes have a reverse option).

Point taken, although I’m thinking it could be possible (and potentially useful) in the case of EQ match, ambience match and reverb match. Maybe I over-simplify, but wouldn’t the second layer simply be the difference between the input and output of those processes?

Yes, in theory you could apply most processes by creating a differential layer. However this could be a little puzzling to most users who just want to apply an effect, or denoise something, and move on.

But I’m certainly seeing the potential for advanced users, I’ll think about it as an extra option in future versions.

1 Like

True. Perhaps a checkbox if you want to create the second layer. Personally, I would find it very informative to be able to explore the bits that are removed.

When trying this process using the Reverb Reduction Process as an example, I could not get it to null. Given that its using the newer v10 AI algorithms behind the scenes, it could explain why.

The video uses the older technique of creating multiple layers and merging which passes the null test.

1 Like

In this particular case, I agree with @Robin_Lobel

As far as terminology goes, when you process something (like food) you’re not also simultaneously extracting all the ingredients (at the same time). Same idealogy would apply here.

1 Like

I am not persuaded. It is really a question of whether producing the layer of left-outs, which is really SL’s claim to fame separating it from other products, is helpful. I believe it would be helpful in some cases.

If our cooking devices gave us an accounting of all the vitamins they destroyed, the fiber they rendered less effective, and probiotics they killed, that would be in improvement, and might change the way we process our food.

Hi cparmerlee I might not understand what you’re after, but wouldn’t simply merging up an inverted duplicate of the processed layer to a duplicate of the original layer do what you need ?

Yes, 5 steps you could accomplish that. It would be nice to simply check a box and have SL create that layer automatically.

Very experienced users might find that very simple to do manually. But for new users, this would be a nice touch to make the tool a bit more accessible.