Sandra Vixen wrote:
Hacking wavelet denoise here, not sure if anyone here uses that but I find it to work really well for reducing noise. It's a GIMP plugin.

Lately I've been using it to smooth skin on MUA MIA shoots, and I find it to produce fairly good results

I tried Photoshop and OpticsPro as well with similar results

I'd like to hear other editors opinions on using noise filters to smooth skin?

Thoughs? Opinions?

Sandra Vixen wrote:
I'd like to hear other editors opinions on using noise filters to smooth skin?

Thoughs? Opinions?

Usually the necessity for noise reduction means the shot was not properly lit/exposed, so the best way to approach it is to use the camera sensor in the ranges where it has high signal to noise ratio.

When physical factors don't allow that a way to reduce noise without use of filters is to take multiple shots of the same subject then average between all shots. This obviously works for static scenes.

As for filtering: personally I almost never do it as it destroys the image structure. If that is so necessary it is best to do it before white balancing the raw data. In any case I don't think noise filtering should not be used for smoothing skin.

Sandra Vixen wrote:
Hacking wavelet denoise here, not sure if anyone here uses that but I find it to work really well for reducing noise. It's a GIMP plugin.

Lately I've been using it to smooth skin on MUA MIA shoots, and I find it to produce fairly good results

I tried Photoshop and OpticsPro as well with similar results

I'd like to hear other editors opinions on using noise filters to smooth skin?

Thoughs? Opinions?

Sorry. I misread the topic header. My explanation was related to noise reduction and not to noise reduction to soften the skin. I do not use any denoise to soften the skin, but I work with frequency separation. Never used noise reduction to smooth skin.

Bernard Wolf wrote:
Noise reduction usually softens or blurs the image and in realistic professional level skin retouching blurring is a no, no.
But a quick skin smoothing can be attained in Camera Raw if you use the adjustment brush and brush in a minus clarity on the skin without going too strong. I know it contradicts what I just said about blurring but for a quick smoothing it's not too bad. None blurring is the preferred pro level with D&B and frequency separation retouching.

One thing to be aware of is that Clarity only affects the luminosity of the image. Negative values on skin will result in popping of the most saturated areas, such as freckles or redness for example.

Andrey Bautin wrote:
One thing to be aware of is that Clarity only affects the luminosity of the image. Negative values on skin will result in popping of the most saturated areas, such as freckles or redness for example.

If you go back into Camera Raw as a filter on a layer you can put the layer in luminosity mode and the color will not be affected.

Bernard Wolf wrote:

If you go back into Camera Raw as a filter on a layer you can put the layer in luminosity mode and the color will not be affected.

There's no reason to. My point was that Clarity is already affecting only luma and hence the color changes. A lot of people teach the opposite but it's the broken link between chroma and luma that alters color.
Luminosity blending mode doesn't let you keep the color of the image, it keeps color channels which doesn't always translate to what you'd expect in terms of perception.

Just a thing to keep in mind.

Andrey Bautin wrote:
My point was that Clarity is already affecting only luma and hence the color changes. A lot of people teach the opposite but it's the broken link between chroma and luma that alters color.

I don't know what your info is based on but a simple test proves this wrong:

1. Download a full color gradient [link expired]
2. Open in PS
3. Duplicate image
4. Convert the duplicate to LAB mode
5. Switch back to the original
6. Apply Filter->Camera Raw filter boosting Clarity to 100%
7. Convert this image to LAB mode too
8. Compare the a/b channels of both images

Result: they are different which means that Clarity filter affects chromaticity too. The deformed color cube also demonstrates that:



As a whole this "clarity" thing is one of the easiest ways to destroy an image. A serious professional should never even touch that.

anchev wrote:

I don't know what your info is based on but a simple test proves this wrong:

1. Download a full color gradient https://files.anchev.net/srgb-hsl-gradient-16bit.png
2. Open in PS
3. Duplicate image
4. Convert the duplicate to LAB mode
5. Switch back to the original
6. Apply Filter->Camera Raw filter boosting Clarity to 100%
7. Convert this image to LAB mode too
8. Compare the a/b channels of both images

Result: they are different which means that Clarity filter affects chromaticity too. The deformed color cube also demonstrates that:



As a whole this "clarity" thing is one of the easiest ways to destroy an image. A serious professional should never even touch that.

We don't know exactly how luminosity is calculated in ACR, there are multiple methods so the difference could come from there.
Compare let's say saturation in ACR vs LAB vs Photoshop adjustment on color blending you'll see a clear difference. Much larger than that of clarity you showed.
So there are different methods used.

Andrey Bautin wrote:
We don't know exactly how luminosity is calculated in ACR, there are multiple methods so the difference could come from there.

Yes, we don't know, that's why we should not claim things without checking and that's why I checked your statement. Not because I intend to use "clarity" but out of curiosity.

Compare let's say saturation in ACR vs LAB vs Photoshop adjustment on color blending you'll see a clear difference. Much larger than that of clarity you showed.
So there are different methods used.

I have no idea what you are comparing, how and in what context. I just checked what you said previously. One thing is sure: Adobe's tools are far from what they promise to be:

https://www.modelmayhem.com/forums/post/968723

As a whole this "clarity" thing is one of the easiest ways to destroy an image. A serious professional should never even touch that.

Yess. Agree. Clarity is not a good way to have a greater perception of sharpness or softness.
The clarity filter also influences the chromaticity of the image and not only the brightness, acting above all on the half tones.

Adobe's tools are far from what they promise to be:

https://www.modelmayhem.com/forums/post/968723

I saw the topic you're talking about. But your conclusions are incorrect and the pictures you attached too. The reason is you  wrong to create the gradient and then you have incorrect results. I attached the real RGB gradient with correct values, made in Photoshop. You can see the values are correct and linear. This is an RGB. Sorry but is not Photoshop to do a mistake.

Bernard Wolf wrote:
Noise reduction usually softens or blurs the image and in realistic professional level skin retouching blurring is a no, no.

That's not always true -

http://www.oleg-ti.com/?mod=content&id=4

...Not every professional shot is a beauty shot. And you don't blur the image as a whole because you clone, blur, add a black alpha alpha layer, then paint in the blurred skin where you need it. If you're shooting 20 outfits in a day with 5 angles for each, you don't waste budget dodging and burning faces that will only be 5-20% of the total image area.

And even if people here were to use it a retouching method for headshots, done properly - with some grain added after the blur - then it looks better than about 95% of MM shots.

fireshoot wrote:
I saw the topic you're talking about. But your conclusions are incorrect and the pictures you attached too. The reason is you  wrong to create the gradient and then you have incorrect results. I attached the real RGB gradient with correct values, made in Photoshop. You can see the values are correct and linear. This is an RGB. Sorry, If you're wrong to pull the conclusions, then you can not blame photoshop.

What do you mean "you're wrong"? There are certain steps described which anyone can follow and check the result for oneself. What exactly is "wrong" in facts (because I show facts, not conclusions)?

If you have any comments about that topic - comment in that thread to avoid diluting current one with off-topic.

anchev wrote:

What do you mean "you're wrong"? There are certain steps described which anyone can follow and check the result for oneself. What exactly is "wrong" in facts (because I show facts, not conclusions)?

If you have any comments about that topic - comment in that thread to avoid diluting current one with off-topic.

Ok you wrote in the other topic is wrong and I show you the reason.

Sandra Vixen wrote:
Hacking wavelet denoise here, not sure if anyone here uses that but I find it to work really well for reducing noise. It's a GIMP plugin.

Lately I've been using it to smooth skin on MUA MIA shoots, and I find it to produce fairly good results

You'd be much better using blur and grain

http://www.oleg-ti.com/?mod=content&id=4

...Or retouching a clone of the image using your current method and then painting in the retouched skin only where needed.

Also, if you are using Gimp, download GMIC and look at the skin smoothing methods available there.

thiswayup wrote:
You'd be much better using blur and grain

Blurring destroys texture. Adding *noise* (which is quite different from film *grain*) to simulate texture does not restore it. Not only this is not "much better" than anything but it is a disaster to the essence of the photographic process and the blotchy image in this article is a proof of that. Don't do that. Learn how to process images correctly. Just like with geometry - there are no fast or royal roads to retouching and being lazy/"clever"/quick is not the way to do it right. Filters have their proper place but surely not like this.