Frequency Separation

[Piotr]

Please add frequency separation!

[Andrius]

Feature request noted! In the meantime, you could use a manual method for frequency separation — is that something you've been able to try?

[Piotr]

Feature request noted! In the meantime, you could use a manual method for frequency separation — is that something you've been able to try?

No, how can I do it in Pixelmator Pro?

[Andrius]

The steps are pretty much the same as you would use in any other image editor which has these fundamental tools. Here they are:

1. Duplicate the base image and apply the Color Controls effect to it, set the Saturation to 0%.

Using the Effects tool is a better option as the adjustments there are purely 'mathematical', whereas if you used the color adjustments, they use a perceptual color model for more realistic and accurate editing results.

2. Duplicate the layer and invert it.
3. Set the blending mode of the upper layer to Linear Light and set its opacity at 50%.
4. Apply a Gaussian blur to the top layer, with a radius of around 3-4px.
5. Group the two layers together and set the blending mode of the group to Overlay.

When you edit the lowest layer (by retouching it in some way, for example), the details should be preserved pretty well, though if you have some specific workflows in mind that don't work using these steps, I could try to find some way of refining this workflow.

[Piotr]

Thank you very much for answer!

[Andrius]

No problem, were you able to put those steps to good use?

[Gloortom]

Can this be added (scripted) in the meantime before you implement it properly?
Also, do you have an ETA for this? (ie., this year or next?)

The steps are pretty much the same as you would use in any other image editor which has these fundamental tools. Here they are:

1. Duplicate the base image and apply the Color Controls effect to it, set the Saturation to 0%.

Using the Effects tool is a better option as the adjustments there are purely 'mathematical', whereas if you used the color adjustments, they use a perceptual color model for more realistic and accurate editing results.

2. Duplicate the layer and invert it.
3. Set the blending mode of the upper layer to Linear Light and set its opacity at 50%.
4. Apply a Gaussian blur to the top layer, with a radius of around 3-4px.
5. Group the two layers together and set the blending mode of the group to Overlay.

When you edit the lowest layer (by retouching it in some way, for example), the details should be preserved pretty well, though if you have some specific workflows in mind that don't work using these steps, I could try to find some way of refining this workflow.

[Han]

Please add frequencyseparation!

+1

[Han]

The steps are pretty much the same as you would use in any other image editor which has these fundamental tools. Here they are:

1. Duplicate the base image and apply the Color Controls effect to it, set the Saturation to 0%.

Using the Effects tool is a better option as the adjustments there are purely 'mathematical', whereas if you used the color adjustments, they use a perceptual color model for more realistic and accurate editing results.

2. Duplicate the layer and invert it.
3. Set the blending mode of the upper layer to Linear Light and set its opacity at 50%.
4. Apply a Gaussian blur to the top layer, with a radius of around 3-4px.
5. Group the two layers together and set the blending mode of the group to Overlay.

When you edit the lowest layer (by retouching it in some way, for example), the details should be preserved pretty well, though if you have some specific workflows in mind that don't work using these steps, I could try to find some way of refining this workflow.

Can you do a video tutorial showing how this is best done? That would be very helpful for non-pro's editors like me.

[Andrius]

Can you do a video tutorial showing how this is best done? That would be very helpful for non-pro's editors like me.

Potentially! Thanks for the idea, we'll think about it.

[Teelar]

This method doesn’t produce good results.

The resulting top layer isn’t preserving all of texture, just kind of half-way doing it.

I however have taken a Pixelmator Pro document, exported it as a PSD, had another program on my iPad do its handy frequency separation magic, bring it back with newly created separate layers, and then I was able to continue on smoothing skin in Pixelmator Pro. To be clear, there is no way to smooth skin as it stands now.

I need a high-pass top layer (in Linear), and a blurred bottom layer, so that I can smooth out skin discolorations on the bottom, and fix texture issues on top. Cheers, guys.



Edit:

Holy crap I figured it out! The trick is to not desaturate the layer first! Do everything else and now I have a perfectly frequency-separated file!

Edit #2

I made a tutorial → https://www.pixelmator.com/community/vi ... =6&t=17286

[Andrius]

Out of interest, when you were desaturating before, were you using the color adjustments or the effects to desaturate? Because that's what may have been giving you incorrect results...

[Teelar]

I was using effects, but I believe it comes down to different purposes. I think your method is good if you’re looking to get some pop of detail (?). But the two layers don’t blend into the original. I found an old Gimp tutorial and applied it to us: don’t desaturate, don’t linear light top layer.

[Andrius]

Ah, the reason I asked is the Saturation slider in the color adjustments is perception-based (as is the entire color model for the color adjustments), while this most likely requires more of the purely mathematical desaturation. But that's just a hunch, to be honest.