1) Fill a NEW layer with 50% Gray.
2) Change this Layer's Mode to SOFT LIGHT.
3) With a paint brush (3-5px, 80% Hardness) Paint the Dark pixels with WHITE.

1) Fill a NEW layer with 50% Gray.
2) Change this Layer's Mode to SOFT LIGHT.
3) With a small paint brush (3-5px, 80% Hardness) Paint the Light pixels with Black.

When you set your layer to Soft Light, filling it with 50% gray does nothing other than waste the time it takes to fill it with 50% gray.  You can just as easily paint white or black on a blank layer to dodge and burn.

A slice of oblivion wrote:
When you set your layer to Soft Light, filling it with 50% gray does nothing other than waste the time it takes to fill it with 50% gray.  You can just as easily paint white or black on a blank layer to dodge and burn.

If you use the Alt/Option-click on the New Layer icon, it brings up the dialog which lets you set blending mode, opacity, and fill all at once, as well as assign a name. It doesn't really take much longer to click on a checkbox.

Also, by filling with blend-neutral color, you can use the burn/dodge tools on the overlay/soft-light layer if you prefer that to painting. Filled with gray vs transparent also gives slightly different results when blurring.

I generally don't fill with gray because of the way I work--I paint on the layers and don't typically blur--but there are legitimate reasons for it for some workflows.

A slice of oblivion wrote:
This is a great post, but there is one detail that I disagree with.


When you set your layer to Soft Light, filling it with 50% gray does nothing other than waste the time it takes to fill it with 50% gray.  You can just as easily paint white or black on a blank layer to dodge and burn.

Christos wrote:
I will agree to this point. I do it this way for reasons outside of the scope of this thread.

There are some advanced coloring techniques that I use with this layer and channel calculations -- when I do NOT use it in Soft Light Mode.

But you are correct, for just spot texture correction the 50% fill is not needed.

Ok that was what confused me.  Your work is amazing and it didn't make sense why you would add an extra unnecessary step.

If there were extra things to be done with the 50% gray layer then I understand why you would include it.  Sorry if that sounded rude.  I honestly didn't mean it to.

OLJ studio wrote:
We do know that sometimes retouchers “transplant” texture from “texture libraries” to get some “ideal” look

It would be really interesting to get tech guide how to create those textures from images with “prefect’ skin and how to apply those textures.

You shoot a model with truly flawless skin with side lighting to really ramp up the texture of the pores.  Use the shoulder, cheek, stomach, or any smooth skin that has pores.

Apply pore texture to skin as a displacement map and adjust opacity of that layer to taste.

The tough part is finding the model with flawless skin to shoot in the first place.

Christos wrote:
The professional retoucher's technique is to "Lighten" the DARK sides and to "Darken" the LIGHT sides -- which produces and more consistent skin texture. Often they work at the pixel level with a 2px brush. VERY Meticulous, VERY time consuming. For most of us, a 4px or 5px brush would suffice.

Thanks, great post!

My retouching technique depends heavily on healing and cloning rather than pixel-level dodging and burning, but it sounds like you are effectively lowering the local contrast on the skin, while not smearing the texture like a blur would do.

If the intention is to paint *every* dark pixel lighter and *every* light pixel darker, the following technique may be a useful shortcut:

- Duplicate base skin layers into a merged layer
- Invert
- Desaturate (we only want to deal with the L channel)
- Switch to Soft Light mode
- Adjust levels to center the midtones (to balance your changes so they are neutral overall)
- If you have a strong lighting ratio, compressing the output levels might minimize any unintended impact on intentional shadow and highlight areas while still smoothing the smaller highlight and shadows inside pores.
- Adjust opacity to taste, and maybe even add a *tiny* blur (0.5-2 pixels) to simulate something similar to what a brush stroke would have given you.
- Use a mask to only apply this areas of the skin where it is desired.

Setting the mode to Luminosity rather than Soft Light with a *very* low opacity (probably under 15%) might help if Soft Light causes a saturation shift.

I played with a photo here quickly and got something that certainly looked better than Ye Olde Gaussian Blur, but I'm not sure how similar it would be in practice to the brush technique you've explained here.

But there again, I've never seen the technique I just described explained anywhere, which probably means it sucks for most photos, or there are 10 easier ways to get the same effect.

Richard Tallent wrote:

Thanks, great post!

My retouching technique depends heavily on healing and cloning rather than pixel-level dodging and burning, but it sounds like you are effectively lowering the local contrast on the skin, while not smearing the texture like a blur would do.

If the intention is to paint *every* dark pixel lighter and *every* light pixel darker, the following technique may be a useful shortcut:

- Duplicate base skin layers into a merged layer
- Invert
- Desaturate (we only want to deal with the L channel)
- Switch to Soft Light mode
- Adjust levels to center the midtones (to balance your changes so they are neutral overall)
- If you have a strong lighting ratio, compressing the output levels might minimize any unintended impact on intentional shadow and highlight areas while still smoothing the smaller highlight and shadows inside pores.
- Adjust opacity to taste, and maybe even add a *tiny* blur (0.5-2 pixels) to simulate something similar to what a brush stroke would have given you.
- Use a mask to only apply this areas of the skin where it is desired.

Setting the mode to Luminosity rather than Soft Light with a *very* low opacity (probably under 15%) might help if Soft Light causes a saturation shift.

I played with a photo here quickly and got something that certainly looked better than Ye Olde Gaussian Blur, but I'm not sure how similar it would be in practice to the brush technique you've explained here.

But there again, I've never seen the technique I just described explained anywhere, which probably means it sucks for most photos, or there are 10 easier ways to get the same effect.

I will have to try that in Photoshop so many different ways to achieve the same result it becomes about which method you become most comfortable with and gives you the results you desire. Its all pixel math in the end.......

Danny Griffin wrote:
Awesome post. Thank you.

One question though for curiosity's sake. How long did it take you to edit your main image from start to finish?

Danny

Julian Marsalis wrote:

I will have to try that in Photoshop so many different ways to achieve the same result it becomes about which method you become most comfortable with and gives you the results you desire. Its all pixel math in the end.......

Christos wrote:
The object here is really not to paint EVERY dark or light pixel. Just those that fall into the too dark or too light category...and even then you must consider the specular highlights and shadows on the models face. These must be approached individually.

I favor second method of using a clone brush with multiple passes so I can selectively bring down the darks and lights accordingly.

The method you suggest would be good for creating a tonal balance in a specified area, but would do little to combat large pores, and acne directly without affecting the pixels around target area.

Remember, I am speaking to those who want to understand the methodology behind how the pro retouchers do it. Pros do have other methods like skin maps and such..

Yep I agree I just like trying different methods to be honest as long as they are not destructive to the image it becomes a learning experience to see what improvements can be made in how I do like to do it. The key for me is always using layer masks to control the degree of any change I make.

Having models with flawless skin and a great Mua isn't happening for me so I work with what I can and hopefully make my client happy new ways to approach old problems is always good in my book....

Christos wrote:
So what happens if I MUST Blur?


Consequently, this is also how I sharpen my images at the very end. I NEVER use the sharpening tools. I always use a HIGH PASS layer at 2 or 3 pixels.

-Christos

Hey Christos, great info about pixel level D&B.  I'll be spending my free time practicing. Thanks.  Can you elaborate on your sharpening method?  Why High Pass method over USM?

Hey Christos, great info about pixel level D&B.  I'll be spending my free time practicing. Thanks.  Can you elaborate on your sharpening method?  Why High Pass method over USM?

Christos wrote:
USM....is suitable for  quick sharpening on a Internet ready file. But know this... While USM sharpens the edges and gives a good contrast, it will actually increase the noise value throughout the image. It is also permanent.

USM's Threshold setting is quite effective at reducing the increase of noise. Use of Smart filters also makes its use not permanent, as would duping the layer before sharpening--which is what High Pass sharpening requires anyhow.

Christos wrote:
A High Pass layer is not permanent, does not create additional noise and can be adjusted with the opacity setting. It also works at both 8bith and 16bit modes. Also by Changing this layer to Overlay, Soft Light and Hard Light, you may be able to target your sharpening to effect you want.

Copying the layer and applying a USM to that layer is arguably more effective than using a High Pass layer for sharpening purposes; it can be masked, is equally not permanent, can be adjusted with the opacity and blend-if settings, and can use different blending modes to change the effect

For band-pass operations--mid-frequency enhancement, for example--the High Pass filter can give substantially enhanced results, but for most sharpening operations, it's less effective than the USM filter.

Kevin_Connery wrote:

Christos wrote:
USM....is suitable for  quick sharpening on a Internet ready file. But know this... While USM sharpens the edges and gives a good contrast, it will actually increase the noise value throughout the image. It is also permanent.

USM's Threshold setting is quite effective at reducing the increase of noise. Use of Smart filters also makes its use not permanent, as would duping the layer before sharpening--which is what High Pass sharpening requires anyhow.


Copying the layer and applying a USM to that layer is arguably more effective than using a High Pass layer for sharpening purposes; it can be masked, is equally not permanent, can be adjusted with the opacity and blend-if settings, and can use different blending modes to change the effect

For band-pass operations--mid-frequency enhancement, for example--the High Pass filter can give substantially enhanced results, but for most sharpening operations, it's less effective than the USM filter.

Christos wrote:
I have observed that the degree of sharpening is a personal one, and I will leave it at that.

No worries. It's a rare thing in Photoshop that there's Only One Way To Do something; pretty much everything has a trade-off somewhere.

Christos wrote:
Method 1:

1) Fill a NEW layer with 50% Gray.
2) Change this Layer's Mode to SOFT LIGHT.
3) With a paint brush (3-5px, 80% Hardness) Paint the Dark pixels with WHITE.

1) Fill a NEW layer with 50% Gray.
2) Change this Layer's Mode to SOFT LIGHT.
3) With a small paint brush (3-5px, 80% Hardness) Paint the Light pixels with Black.

When you are done, adjust the Opacity levels to suit your taste.... Or you can set the Opacity of your brush to begin with (No less than 40%).

What is the benefit in doing this 50% gray step?

Kevlar Vest Girl wrote:

What is the benefit in doing this 50% gray step?