Sharpening

[will139]

I have tried smart sharpen, hi-pass sharpening, and Topaz Sharpening AI and just recently became unhappy with them. I have read that if you have adjustment layers in your work flow that sharpening cannot sharpen them so it make it necessary to use a stamped layer to sharpen. Can I get some some help and how other people sharpen and any tips to assist me. I always use noise reduction as a first step when sending to PS from ACR. I use Topaz Denoise. I did use Topaz Denoise AI but don't like it all that well. I always shoot in RAW with a Canon 90D or Canon 6D Mark ll. Lenses most often used are a Canon 100-400mm f4.5-5.6 or a Canon 500mm f4. I regularly use a 1.4ext. Any help will be appreciated.

 

[DRwyoming]

Any help will be appreciated.

There are a lot of approaches, but FWIW, here's what I generally do:

- Remove noise early as you posted above.
- Do my major processing in LR or PS which may include some subtle sharpening during raw conversion but not trying for final output sharpening yet
- Do any targeted editing in PS using layers only if needed
- Resize the image for output size, if using layers in PS flatten or merge (for print work I'll keep a layered .psd or layered tiff for future editing)
- Sharpen the image at final output size and again if it was a layered PS file I'll flatten it prior to output sharpening

That flow varies depending on specific tools used and the image as well as its intended output use. For instance for web posts like what I post on these forums if I use Topaz DeNoise during my process I usually don't need any additional sharpening at output. That throws a wrench in the 'remove noise early - sharpen late' workflow philosophy but Denoise does add quite a bit of sharpening on its own.

But you're right, an adjustment layer on top of the layer stack cannot be sharpened by itself in PS though using adjustment layers and then either flattening the image or merging the image into a new layer results in an pixel layer that includes all the adjustments and can be sharpened. If you think there may be future adjustments to any of the adjustment layers (e.g. fine tuning contrast or Hue/Saturation after a test print) then a merged layer is a pretty good way to go or you can do as I often do and save a layered working copy for any future edits.

 

[will139]

There are a lot of approaches, but FWIW, here's what I generally do:

- Remove noise early as you posted above.
- Do my major processing in LR or PS which may include some subtle sharpening during raw conversion but not trying for final output sharpening yet
- Do any targeted editing in PS using layers only if needed
- Resize the image for output size, if using layers in PS flatten or merge (for print work I'll keep a layered .psd or layered tiff for future editing)
- Sharpen the image at final output size and again if it was a layered PS file I'll flatten it prior to output sharpening

That flow varies depending on specific tools used and the image as well as its intended output use. For instance for web posts like what I post on these forums if I use Topaz DeNoise during my process I usually don't need any additional sharpening at output. That throws a wrench in the 'remove noise early - sharpen late' workflow philosophy but Denoise does add quite a bit of sharpening on its own.

But you're right, an adjustment layer on top of the layer stack cannot be sharpened by itself in PS though using adjustment layers and then either flattening the image or merging the image into a new layer results in an pixel layer that includes all the adjustments and can be sharpened. If you think there may be future adjustments to any of the adjustment layers (e.g. fine tuning contrast or Hue/Saturation after a test print) then a merged layer is a pretty good way to go or you can do as I often do and save a layered working copy for any future edits.

If you have made adjustments and flattened the image so that sharpening can be applied do you duplicate the layer you are about to sharpen or does that then count as an adjustment?

 

[DRwyoming]

If you have made adjustments and flattened the image so that sharpening can be applied do you duplicate the layer you are about to sharpen or does that then count as an adjustment?

I will generally sharpen on a duplicate pixel layer assuming I've flattened the image prior to sharpening. There's several reasons for that:

- By sharpening on a duplicate pixel layer I can go a tad overboard on sharpening and then use layer opacity to tune it back to taste. The ability to toggle the layer visibility (eyeball icon) on and off makes it really easy to see the impact of sharpening and tune it back to taste. This is one of the best ways I know of to avoid over sharpening as that toggling makes it very easy to see halos and other sharpening artifacts.

- Sharpening on a dupe pixel layer makes it very easy to add a layer mask and apply the sharpening selectively if needed or perhaps to just tune down the sharpening a tad in places like out of focus backgrounds or skies.

- When sharpening in PS I'll usually switch that top sharpened layer into Luminosity blending mode which tends to result in a bit cleaner sharpened image and that becomes very easy to do when sharpening on a dupe pixel layer as all you have to do is change the blending mode of that top layer with the blending mode pull down menu. I wouldn't bother with this for something like a web post but for a large wall print it can be a useful trick.

 

[gordon_g]

I do the exact same workflow as Dave... @DRwyoming that an interesting thought on the blending on the sharpening....

 

[pochito]

I will generally sharpen on a duplicate pixel layer assuming I've flattened the image prior to sharpening. There's several reasons for that:

I have a question.
Is there a problem with duplicating the pixel layer for sharpening even if there are adjustment layers on top?
Say:
Background/Copy of Backround/Adj Layer 1/ Adj Layer2/...
And then applying Sharpening to Copy of Backround.
Thanks

 

[gordon_g]

I have a question.
Is there a problem with duplicating the pixel layer for sharpening even if there are adjustment layers on top?
Say:
Background/Copy of Backround/Adj Layer 1/ Adj Layer2/...
And then applying Sharpening to Copy of Backround.
Thanks

Problem.... no but it’s best to apply the sharpening after all other adjustments.... it usually applies a better sharpening. In your process above you are applying the sharpening before the two adjustment layers. Depending on what adjustment you can sometimes get some funny sharpening caused by your adjustments.

 

[will139]

Problem.... no but it’s best to apply the sharpening after all other adjustments.... it usually applies a better sharpening. In your process above you are applying the sharpening before the two adjustment layers. Depending on what adjustment you can sometimes get some funny sharpening caused by your adjustments.

What sharpening process do you use? Hi-pass, smart sharpen et al??

 

[gordon_g]

What sharpening process do you use? Hi-pass, smart sharpen et al??

I’ve used high pass a lot.... but honestly I use topaz sharpen mostly now... although I will be trying Dave‘s blending trick.... I also have and still use noiseware in some images... between the three options Im pretty happy.

 

[will139]

I’ve used high pass a lot.... but honestly I use topaz sharpen mostly now... although I will be trying Dave‘s blending trick.... I also have and still use noiseware in some images... between the three options Im pretty happy.
[/QUOT
I have Topaz sharpen and used it regularly for awhile and then for some reason I went to compare it with smart sharpen and hi-pass and preferred them to the Topaz.

 

[DRwyoming]

I have a question.
Is there a problem with duplicating the pixel layer for sharpening even if there are adjustment layers on top?
Say:
Background/Copy of Backround/Adj Layer 1/ Adj Layer2/...
And then applying Sharpening to Copy of Backround.
Thanks

I second what Gordon posted, it will work but your results may be inconsistent.

For instance if you sharpen on a background or even a dupe background layer but then place adjustment layers above like Curves, Levels, blank adjustment layers with various blending modes, etc., you can end up changing the brightness, contrast or even colors of the sharpening contours which changes their impact and not always for the better. Say you sharpen the image and then later decide parts of the image lack midtone contrast so you add an S curve layer. You won't just increase the contrast of the overall image, you'll also add contrast to the sharpening contours that work on the basis of adding contrast to edges. IOW, you'll add more contrast to those edge contours which really is the same thing as over sharpening in the first place.

So yeah you could sharpen the background layer or a dupe pixel layer above the background layer even though there are adjustment layers above it but any adjustment layers will apply adjustments to the sharpened layer. Maybe that works out fine in your image and if it's done as the last step in your processing you'll see any adjustments to the sharpened layer real time so perhaps you just sharpen a bit less to compensate but it's a bit unusual from a workflow perspective.

 

[gordon_g]

I have Topaz sharpen and used it regularly for awhile and then for some reason I went to compare it with smart sharpen and hi-pass and preferred them to the Topaz.

I see some people reporting that besides you, but I have personally never seen this is my own workflow... I ALWAYS bring down the recommended settings that Topaz defaults to. and I often am using something then just the default sharping profile. it works for me about 95% of the time just perfectly. and just like anything else it not a cure all either.

 

[pochito]

I second what Gordon posted, it will work but your results may be inconsistent.

....

Thank very much you both. I understand.
Usually the last thing I do is sharpening for the output.
No problem with flattening the image before sharpening. I was just curious.

When using AI Sharpen there is the possibility of "stabilizing".
If needed, I would not "stabilize" the image at the end, after doing some work. I'd like to see if it is worth the effort.
So I would create a background duplicate, apply "stabilize" to the dupe, adjust the result with a mask (I personally feel more comfortable with the PS than with the Topaz adjustements), flatten and then continue making further adjustments.
Do you see a problem in that?

 

[DRwyoming]

Thank very much you both. I understand.
Usually the last thing I do is sharpening for the output.
No problem with flattening the image before sharpening. I was just curious.

When using AI Sharpen there is the possibility of "stabilizing".
If needed, I would not "stabilize" the image at the end, after doing some work. I'd like to see if it is worth the effort.
So I would create a background duplicate, apply "stabilize" to the dupe, adjust the result with a mask (I personally feel more comfortable with the PS than with the Topaz adjustements), flatten and then continue making further adjustments.
Do you see a problem in that?

I've been using the Topaz products for only a few months now but in my experience they really throw a wrench in the 'sharpen late' workflow philosophy. IOW, after running a Stabilize Sharpen I often don't need more sharpening at the end of the workflow. But if I've got some subtle motion blur that Stabilize can fix I'll certainly do it and do it fairly early but then I might not sharpen again at the end of the workflow.

Similar story with running Topaz DeNoise, it provides a lot of sharpening and for its noise reducing benefits I'll run it early in the workflow but often won't add any additional sharpening at the end of the workflow.

These tools are great and I've been using them a lot but it does change my approach to sharpening as the last step as I often don't need or want any more sharpening than they give me.

 

[gordon_g]

Thank very much you both. I understand.
Usually the last thing I do is sharpening for the output.
No problem with flattening the image before sharpening. I was just curious.

When using AI Sharpen there is the possibility of "stabilizing".
If needed, I would not "stabilize" the image at the end, after doing some work. I'd like to see if it is worth the effort.
So I would create a background duplicate, apply "stabilize" to the dupe, adjust the result with a mask (I personally feel more comfortable with the PS than with the Topaz adjustements), flatten and then continue making further adjustments.
Do you see a problem in that?

the stabilize Feature is just a form of sharpening... do it at the very end.

 

[gordon_g]

I've been using the Topaz products for only a few months now but in my experience they really throw a wrench in the 'sharpen late' workflow philosophy. IOW, after running a Stabilize Sharpen I often don't need more sharpening at the end of the workflow. But if I've got some subtle motion blur that Stabilize can fix I'll certainly do it and do it fairly early but then I might not sharpen again at the end of the workflow.

Similar story with running Topaz DeNoise, it provides a lot of sharpening and for its noise reducing benefits I'll run it early in the workflow but often won't add any additional sharpening at the end of the workflow.

These tools are great and I've been using them a lot but it does change my approach to sharpening as the last step as I often don't need or want any more sharpening than they give me.

Dave I actually turn all the sharping off (not completely off but I usually end up about 80% less when using denoise... make my changes and then sharpen at the end.

 

[bleirer]

Adjustment layers will sharpen, if you are talking about brightness/contrast, levels, curves, etc. They will apply to everything below. If you are talking about regular layers, sharpening only applies to the one layer you are on. So you could select the layers you want included and then either right click and choose 'convert to smart object' and then sharpen the smart object, or select the layers and use ctrl alt e which will make a merged layer that can then be sharpened directly or to make sharpening non-destructive convert it to a smart object.

 

[will139]

I will generally sharpen on a duplicate pixel layer assuming I've flattened the image prior to sharpening. There's several reasons for that:

- By sharpening on a duplicate pixel layer I can go a tad overboard on sharpening and then use layer opacity to tune it back to taste. The ability to toggle the layer visibility (eyeball icon) on and off makes it really easy to see the impact of sharpening and tune it back to taste. This is one of the best ways I know of to avoid over sharpening as that toggling makes it very easy to see halos and other sharpening artifacts.

- Sharpening on a dupe pixel layer makes it very easy to add a layer mask and apply the sharpening selectively if needed or perhaps to just tune down the sharpening a tad in places like out of focus backgrounds or skies.

- When sharpening in PS I'll usually switch that top sharpened layer into Luminosity blending mode which tends to result in a bit cleaner sharpened image and that becomes very easy to do when sharpening on a dupe pixel layer as all you have to do is change the blending mode of that top layer with the blending mode pull down menu. I wouldn't bother with this for something like a web post but for a large wall print it can be a useful trick.

So if you do the luminosity trick you do it before sharpening or after?

 

[DRwyoming]

So if you do the luminosity trick you do it before sharpening or after?

Well it really doesn't matter which sequence you use. IOW, you could spawn a dupe layer then change its blending mode to Luminosity and then sharpen that layer or do it in the reverse order and it wouldn't change anything. Changing the blending mode first and then sharpening with something like the Unsharp Mask might make it easier to see the eventual sharpening effect but the sequence doesn't really matter much.

One note, if you use something like High Pass sharpening you either don't use the luminosity blending trick or it takes more steps. IOW, High Pass sharpening depends on using a contrast oriented blending mode (i.e. Overlay, Soft Light or Hard Light) so you can't also choose Luminosity blending for that layer. But you can spawn two dupe pixel layers, run the high pass filter on the top one, change it's blending mode to one of those listed above, fade the top layer's opacity to taste, merge those two layers and then change the blending mode of that merged layer to luminosity mode.

The differences between Normal and Luminosity mode for sharpening is usually pretty subtle so it's something I reserve for larger wall prints and don't really bother with for smaller prints or web images. IOW, about 9 times out of 10 it's not worth the extra steps but is still nice to have in your bag of tricks.

 

[pochito]

Thank you everybody for the clear and useful responses

 

[Cristobal]

I've been using the Topaz products for only a few months now but in my experience they really throw a wrench in the 'sharpen late' workflow philosophy. IOW, after running a Stabilize Sharpen I often don't need more sharpening at the end of the workflow. But if I've got some subtle motion blur that Stabilize can fix I'll certainly do it and do it fairly early but then I might not sharpen again at the end of the workflow.

Similar story with running Topaz DeNoise, it provides a lot of sharpening and for its noise reducing benefits I'll run it early in the workflow but often won't add any additional sharpening at the end of the workflow.

These tools are great and I've been using them a lot but it does change my approach to sharpening as the last step as I often don't need or want any more sharpening than they give me.

Sorry, I arrive later in the discussion; I am not so familiar with post processing techniques but I understand that for most situations, the combo LR + Topaz DeNoise is sufficient for sharpening; so do you make some adjustments or not in LR before sending your image in Denoise Al , like cropping, lens ajustment and exposition?

 

[DRwyoming]

Sorry, I arrive later in the discussion; I am not so familiar with post processing techniques but I understand that for most situations, the combo LR + Topaz DeNoise is sufficient for sharpening; so do you make some adjustments or not in LR before sending your image in Denoise Al , like cropping, lens ajustment and exposition?

It really depends on the image but there's no real harm in cropping or applying lens corrections prior to running the image through DeNoise. I also tend to do exposure adjustments, especially and major shadow pulling or highlight recovery before sending the file to the Topaz tools. That's mostly because I send copies of the files to other editing tools like Topaz products or Photoshop and if I'm going to do any major shadow pulling or highlight recovery I'd prefer to do that on the original raw file.

 

[Cristobal]

It really depends on the image but there's no real harm in cropping or applying lens corrections prior to running the image through DeNoise. I also tend to do exposure adjustments, especially and major shadow pulling or highlight recovery before sending the file to the Topaz tools. That's mostly because I send copies of the files to other editing tools like Topaz products or Photoshop and if I'm going to do any major shadow pulling or highlight recovery I'd prefer to do that on the original raw file.

Thank you for your quick answer; I know now how to proceed and I will follow your method

 

[Calson]

The amount of sharpening depends on the image size and the nature of the scene or subject. So I sharpen an image more to make a 1000 x 1000 ppi image file than I do for a 3000 x 3000 ppi image file. Sharpening is the very last step after resizing a file and I alway save the original file before any resizing. If I have a 4000 x 6000 file I will reduce it to 3000 x 3000 and then sharpen and if going to an even smaller file size I start again with the 4000 x 6000 file and reduce it in size and then apply sharpening.

I have learned that I need to adjust levels and then contrast with the native file before doing anything else. Increasing contrast can greatly reduce the amount of sharpening that is needed. There are a great many ways to apply sharpening to an image and it takes time to learn which approach works best for any given image.

Back in the days of film it was much much simpler. You exposed the film and sent it to a lab and that was it. Now the photographer is the lab and nothing is automated and it provides a great many more image processing images but at the cost of ones time - no free lunch. When shooting chrome it would take an hour to produce a single Cibachrome print in my darkroom. Now I can upload the image files to a lab and have a hundred prints delivered to my door two days later which compensates for the post processing time invested.

 

[Cristobal]

The amount of sharpening depends on the image size and the nature of the scene or subject. So I sharpen an image more to make a 1000 x 1000 ppi image file than I do for a 3000 x 3000 ppi image file. Sharpening is the very last step after resizing a file and I alway save the original file before any resizing. If I have a 4000 x 6000 file I will reduce it to 3000 x 3000 and then sharpen and if going to an even smaller file size I start again with the 4000 x 6000 file and reduce it in size and then apply sharpening.

I have learned that I need to adjust levels and then contrast with the native file before doing anything else. Increasing contrast can greatly reduce the amount of sharpening that is needed. There are a great many ways to apply sharpening to an image and it takes time to learn which approach works best for any given image.

Back in the days of film it was much much simpler. You exposed the film and sent it to a lab and that was it. Now the photographer is the lab and nothing is automated and it provides a great many more image processing images but at the cost of ones time - no free lunch. When shooting chrome it would take an hour to produce a single Cibachrome print in my darkroom. Now I can upload the image files to a lab and have a hundred prints delivered to my door two days later which compensates for the post processing time invested.

I don't understand why you have to resize your image before the sharpening operation; if you don't have to crop anyway your original and you want to have the best resolution during printing, its not preferable to have the maximum Mpx image before printing?