Inspiring

Answered

working with smart objects, unsharp mask and transparency

11 replies
98 views
Translate

I created a file many years ago before I stopped working with Smart Objects. At the time, I wanted to sharpen sections of the image. I now need to enlarge the canvas slightly to augment the tile border, and in the process of doing that, I noticed that my Smart Objects have transparency. AI says “the Unsharp Mask filter … operates strictly on the color and luminance channels of an image and does not utilize, modify, or factor in layer transparency or alpha channels during its calculation.” In the first screenshot below, the finished image with all its usable layers is shown. In the second, the bottom layer only is toggled on and reveals an out of focus section in the left upper corner which is replaced by the Layer in the 3rd screenshot (2166). What I don’t understand is what is going on in screen shots 4 & 5, which show semi-transparent pixels. Those areas seem to correspond with what is showing on the respective masks for those objects. If transparency is not used for masking, what function might it serve here? Was it possibly just user error on my part?

1

2

3

4

5

c.pfaffenbichlerCorrect answer

Community Expert

It seems you painted on (or filled areas in) the Layer Mask with a color other than White and Black or with lowered Opacity and/or Flow.

Applying a Smart Filter to a Smart Object does not affect the Layer Mask.

Just to make sure: Is there a Density-setting other than 100% for the Mask in the Properties Panel?

davescm

Community Expert

@Barton5C39 A few observations.

1. Unsharp mask. The effect is applied as if the layer mask (or layer transparency setting) was not there, then the mask is applied. It means that if you later change the layer mask ( or alter the layer transparency) then the unsharp mask applies evenly across the layer. You can test this for yourself by creating a layer with a gradient mask, then using unsharp mask on the layer. Now disable the mask and you will see unsharp mak has been used across the entire layer including the parts hidden by the mask.

2. You cannot judge unsharp masking (or any other edge sharpening) when zoomed out to 6%. Always use 100% zoom to check sharpening. The reason being that the preview you are seeing when zoomed out combines pixels on each layer before blending the layers. That in itself will soften the visual effect of any sharpening on the preview. When you come to output (print/export etc) then the image is flattened on an individual pixel basis.

3. I rarely apply sharpening to individual layers unless the images making a composite are mismatched. For output sharpening, if using a small number of layers I put all layers in a smart object and apply sharpening to that. If there are many layers, then for final output sharpening, I add a layer at the top and use merge visible (Shift+Ctrl+Alt+E) then sharpen that single merged layer. You also can change the blend mode of that layer if it works for you. For example try luminosity (or switch to difference to see exactly where the sharpening has been applied).

4. As for unsharp mask vs other algorithms, try them and see what works for you. I still use unsharp mask for print output - it can look over-sharpened on screen at 100% but great on the printed output.

Dave

Translate

B

Barton5C39Author

Inspiring

Thanks for your comments. I think what you are saying in #1 is that, in effect, “Unsharp mask” is a somewhat misleading term, as there is no masking of the sharpening until (and if) you apply a mask, fill the mask with Black and then reveal areas of sharpening by painting White - or by applying a gradient, as you state - is that correct?

In #3, does “output sharpening” just refer to the slight sharpening one does to an image before printing - which I believe is customary? I try really hard to capture my images as sharply as I can - and don’t usually do this (output sharpening) - but maybe I should.

Also in #3, for your approach “using a small number of layers” are you merging those layers before “putting them in” a Smart Object (presumably so because I don’t think you can have a Smart Object with “multiple” layers), and if so, are you making a copy of those layers first for the Smart Object?

For the second version (with “many layers”) what’s the difference if you are also merging those layers (or a copy of them) into a single merged layer? Is it that you are not making them a Smart Object? In this case, do you have a single merged layer at top that comes from making copies of the individual layers beneath?

I have never used the Blend Mode or Luminosity much as they seemed somewhat involved and I wanted a good online tutorial which I had trouble finding - I probably should get a handle on these techniques.

Translate

davescm

Community Expert

Hi,
To take the points in turn :
1. What unsharp mask does is combine the original layer content with a blurred and inverted copy of that original thereby introducing contrast along the edges of the image.This effectively increases the contrast of those edges, giving the impression of sharpening the image. It is an old photographic technique which used two exposures when printing (the first a standard exposure and the second a blurred negative exposure known as the mask) and has nothing to do with the layer masks in Photoshop. If overdone, then those ‘sharpened’ edges can show as halo’s along the edge. So it is correctly named but I can see where the name could mislead. In Photoshop it is applied to the image as if there was no layer mask on the layer and once ‘sharpened’ the layer mask then impacts the resulting image.

3. Output sharpening does indeed refer to sharpening for the final output e.g. printing.
There are usually three stages of sharpening.
Stage 1 - Capture sharpening. This is often done as part of the raw conversion and is really done to compensate for the softening effect of the anti-aliasing filter which is built over the sensor of a digital camera. A camera without an anti-aliasing filter can be prone to moire pattern effects.
Stage 2 - Creative sharpening - this sharpening (or blurring) is done for creative effect.
Stage 3 - Output sharpening - This is done prior to the output of the image and is done to compensate for the softening effect of placing ink on paper. Viewed on screen it can look over-sharpened but if done correctly i.e. enough to compensate for the printing process, it will look correct on paper.
It is worth thinking along those lines and applying sharpening as appropriate to each stage.

When output sharpening an image with just a few layers, I do not merge them. I just select them all and put them in a single smart object and sharpen that. With many layers I use the ‘merge visible’ technique. Neither way is better - both can give the same visual result. Either way, because output sharpening is specific to the particular printer, paper and print size, I tend not to save that sharpening step. That way, I sharpen each time for the specific output required.

The good thing about blend modes is that you can instantly see what they do. Luminosity blend mode just means the layer being blended only affects the luminosity of the image and the hue and saturation (i.e. colour) is taken from the layer(s) below.
https://helpx.adobe.com/uk/photoshop/desktop/repair-retouch/adjust-light-tone/blending-mode-descriptions.html

Dave

Translate

Trevor.Dennis

Community Expert

I am in the same position as @c.pfaffenbichler You appear to be over thinking it.

This a BIG file going on the 6.6% zoom ratio and it still being more than 1900 pixels wide in your screenshot. Is the image scientific in nature?

You have a bunch of Curves layers all clipped to the same Smart Object layer, and all with solid white masks so affecting the entire document area. What’s going on there?

This layer appears to be empty other than the top left corner, and has Unsharp Mask applied.

I can’t see a scroll bar for the Layers Panel in your screenshot, and the layer is the upper most pixel layer in the stack, so the Unsharp Mask will only be applied to the top left corner, and have no effect on the layers below it in the stack. I think you knew that?

I am not entirely sure this is the bottom of the stack, because you might have cropped the scroll bar when making the screen shot, but going by the .CR2 layer name I think this might be the original RAW file object layer. The two smart objects above it have masks with greyscale areas, so they will show as around 50% opacity. They will not show in the top left corner because the 2166 Smart Object layer hides that area.

OK, bottom line. If it was me I would turn off every one of the Unsharp Mask sharpening, because there are much better ways to sharpen nowadays. If everything else is as you want it, I would add a copy merged layer to the top of the stack (Shift Ctrl Alt E) and apply sharpening to that entire layer. Personally I would use Topaz Photo, or Topaz Photo AI, but I’m guessing you don’t have that? It’s a very busy image, and I suspect that might make a difference to how it is approached.

I wonder if using High Pass in two stages might work here, using low value for the fine detail, and higher value for the less busy areas, but I am not sure if you’d need to mask some of the upper High Pass layer, or just set both to Overlay and they work themselves out.

If you have the Generative Credits you might split into 2000 x 2000 pixel squares, and use Flux Pro with a prompt like this: (the prompt comes Prompt Vault from The Photoshop Training Channel)

Enhance this image with crisp , sharp focus an increased micro-detail throughout. Bring out surface textures, and material definition. Keep the pose, camera angle, composition, scale and lighting exactly the same.

There might be an issue with AI sharpening if this a scientific image, because you couldn’t be 100% sure that new content was not introduced.

@davescm Any ideas on the best way to sharpen this?

Translate

B

Barton5C39Author

Inspiring

@Trevor.Dennis thank you. To answer your questions,

This is not a scientific file. I make large, composite pictures of densely textured surfaces that occur organically and that I frame as artworks akin to ones produced by human hand. I am often queried about the size of my files, which are meant to be printed, with their extensive detail and texture intact. I sense skepticism about my file sizes, and I’m not sure why. I generally don’t have a problem working with them.
The adjustment layers clipped to the top SO apply to the portion of the image that was most poorly in focus. That SO is only that section of the image. I don’t remember, but I may have used different captures of that section in that SO, which I then needed to both apply a different value of sharpening to, as well as other adjustments, to make it match the adjoining areas. The other SO’s below this one have diffrent values of sharpening applied, which were not as badly out of focus.
You are correct, the .CR2 layer is the original layer. @c.pfaffenbichler has confirmed that I may have applied grayscale when painting to reveal the sharpening. I am still trying to understand the import of his comment “Applying a Smart Filter to a Smart Object does not affect the Layer Mask” and how that affected painting with what I thought was White. To the extent I applied sharpening, applied a Mask, filled it with Black, and then intended to paint with White to reveal the areas I wanted to be sharpened, if the color was not White but grayscale, that presumably would result in a diminished level of sharpening than intended, is that correct?
Thank you for your suggestion as to how to potentially reapply sharpening to this file. I am aware that sharpening methods have evolved since I worked with this image back in 2019 or so. It’s my first time hearing the terms Topaz Photo, Generative Credits, Prompts Vault and Flux Pro - so I’ll do a little research to get an idea about these tools.
About a year ago I had a question about Sharpening and got an excellent response from @Conrad_C detailing the evolution and current state of Sharpening workflow in Photoshop, Camera Raw and Lightroom. He praised High Pass but said he was advising beginners working in PS to use Smart Sharpen for results that improve upon Unsharp Mask and are generally quite good. Interestingly, in that same thread, @D Fosse said “Just use Unsharp Mask. High pass sharpening is just a manually built Unsharp Mask. Same result, much more work.” So I guess one should experiment with different methods and find what works best for them.

As always, I am enormously grateful for the time people take in this forum to share their knowledge and opinions.

Translate

c.pfaffenbichler

Community Expert

I don't understand what the problem is supposed to be.

If you created a layer mask with gray values the masked smart object's visibility will be determined by those (and its own transparency).

Edit: so if you want to hide the semitransparent areas you can edit the layer mask, with curves for example or by painting or filling.

Translate

B

Barton5C39Author

Inspiring

Thank you. The workflow I adopted years ago when creating this file was:

1.   Turn the layer into a SO
2.   Duplicate the SO
3.   Apply the Unsharp mask filter
4.   Add a Layer Mask.
5.   Fill with Black.
6.   Paint with White to reveal areas needing sharpening.

I don’t remember then using grayscale, nor do I believe I should have. Hence my confusion about semi-transparency. When I zoom into 100%, and turn off the Unsharp Mask for a given SO, I can see the sharpening. So it appears that my workflow worked - but I don’t understand why grayscale would show when toggling on the SO’s individually, as in my screenshots. Just curious about what is happening here.

Translate

c.pfaffenbichler

Correct answer

Community Expert

It seems you painted on (or filled areas in) the Layer Mask with a color other than White and Black or with lowered Opacity and/or Flow.

Applying a Smart Filter to a Smart Object does not affect the Layer Mask.

Just to make sure: Is there a Density-setting other than 100% for the Mask in the Properties Panel?

Translate

Sign up

To post, reply, or follow discussions, please sign in with your Adobe ID.

Sign in to Adobe Community

To post, reply, or follow discussions, please sign in with your Adobe ID.