Sharpening process for different file types (raw to dng to jpeg)

[Brownbear]

I would just like to check up on my sharpening process for full size images and the file type.

1. I process an image of a bird in Lightroom, applying selective sharpening the bird only, and then export this file to Adobe Photoshop CC as a high quality DNG file

2. I open up the DNG in Adobe Photoshop in ACR to zap dust spots (does a better job than LR). Then I open this file Adobe Photoshop CC proper (out of raw) and apply an unsharp mask to the bird image, and save it as a high quality jpeg.

It just occurred to me that perhaps the new jpeg also needs sharpening, simply because of the change in file type. And that perhaps I shouldn't sharpen the DNG file at all. Or do both file types need to be sharpened?

This confusion came about because I recently switched from exporting the file from Lightroom as a Jpeg to a DNG file, thinking that the quality of the image will be higher if I do the sharpening on the DNG file instead of a jpeg file.

Thank you.

[NorthernFocus]

Colin is probably best to speak to this. But unless you are resizing the image, resizing before jpeg conversion is preferable. Try not to make any more changes post jpeg conversion if possible. Every time you save a jpeg you alter more data. If you are going to resize, do that first in TIFF or PSD format, do final sharpening etc, and save as jpeg.

[Colin Southern]

Hi Christina,

I think you've got a bit of double-handling going on there. If it were me, I'd forget about sharpening in both LR and ACR. I'd be making my initial global adjustments in ACR then opening in PS where I'd zap the dust bunnies - apply capture sharpening (0.3px @ 300%) then content/creative sharpening (usually global) often in the region of 4px @ 40%. Once the image is down-sampled for internet display (typically to around 1200 x 800px) then I'd apply output sharpening of typically 0.3px @ 50%. Usually I don't need to sharpen selectively, but it does depend on the image.

[Brownbear]

Thank you Dan and Colin,

Dan...

So that means any cropping is also best performed on the raw image due to lost pixels? And that if you sharpen the PSD or DNG file, with no resizing, ie intent to print full size that there is no need to sharpen the jpeg.

Colin...

Yes, and it is because for some reason it seems easier to apply selective sharpening in LR with the brush then selecting around a bird in Photoshop. That said, I have tried your sharpening technique a couple of times, and I will try it again.

Most of my images are of birds on water or flying against a blue sky, and these are the ones that I have challenges with processing. I would think that the best approach for these types of images using your method would be

1. Capture sharpening .3 pixels x 300% but only on the bird, not water, sky or clouds because the latter shouldn't be sharpened? (noise, artifacts and just not necessary)

2. Content/Creative sharpening of say 4 pixels @ 40% but again just on the bird (same reasoning as in #1)

3. .3 px at 50% for posting on the web, and likely selective if it is a bird against the sky etc..

However, if I am sending out the jpeg image for printing full size, 4000 x 6000 pixels.... Do I need to sharpen the image that I just processed after I change it to a jpeg from a DNG (from Lightroom) or a PSD file (from Photoshop after final edits)... this is what is confusing me...


Thank you.

[NorthernFocus]

Christina, the final round of sharpening should be on the final image size but BEFORE converting to jpeg. No need to sharpen again just due to jpeg conversion. So:
- all other PP
- crop/resize
- final sharpening
- convert/save to jpeg

Reading your above post, most definitely DO NOT export from LR as a jpeg if you have additional PP to do at all. Export from LR as tiff or psd and only convert to jpeg for saving the final version. The only reason to save as a jpeg at all is if you need to for posting to web or submitting for printing etc. to someplace that requires jpeg format. The sole value of jpeg is reduced file size. The downside is that you lose data/detail so any additional PP is not a good idea.

[Brownbear]

Thanks a million!

Christina, the final round of sharpening should be on the final image size but BEFORE converting to jpeg. No need to sharpen again just due to jpeg conversion. So:
- all other PP
- crop/resize
- final sharpening
- convert/save to jpeg

Reading your above post, most definitely DO NOT export from LR as a jpeg if you have additional PP to do at all. Export from LR as tiff or psd and only convert to jpeg for saving the final version. The only reason to save as a jpeg at all is if you need to for posting to web or submitting for printing etc. to someplace that requires jpeg format. The sole value of jpeg is reduced file size. The downside is that you lose data/detail so any additional PP is not a good idea.

[dje]

Christina one thing to consider with the D7100 is that it has no anti-aliasing filter. This means that for a given lens, and assuming focus is spot on, it should be capable of producing sharper images than a camera which has an AA filter. You therefore may not need to apply as much sharpening for this camera as you would for others.

I'd suggest that you experiment a bit with the settings to see what gives the results you like.

Dave

[Brownbear]

Thank you Dan. Yes, I definitely notice a difference in sharpness, ie; much sharper than my Sony 200 and Nikon D80.

I will experiment, but I think I tend to (or used to) undersharpen when I processed... So just another thing to learn to do well.

Truly appreciated. Thank you for sharing.

Christina one thing to consider with the D7100 is that it has no anti-aliasing filter. This means that for a given lens, and assuming focus is spot on, it should be capable of producing sharper images than a camera which has an AA filter. You therefore may not need to apply as much sharpening for this camera as you would for others.

I'd suggest that you experiment a bit with the settings to see what gives the results you like.

Dave

[Colin Southern]

Yes, and it is because for some reason it seems easier to apply selective sharpening in LR with the brush then selecting around a bird in Photoshop.

There are techniques in Photoshop that make selective sharpening ridiculously easy. Probably the easiest is to just duplicate the layer - sharpen the top layer - and then just erase the portions you don't want sharpened. Should't take more than 30 seconds.

Most of my images are of birds on water or flying against a blue sky, and these are the ones that I have challenges with processing. I would think that the best approach for these types of images using your method would be

1. Capture sharpening .3 pixels x 300% but only on the bird, not water, sky or clouds because the latter shouldn't be sharpened? (noise, artifacts and just not necessary)

No - capture sharpening is usually applied to the whole image, but having said that, it only makes a difference for when you're working on the full resolution image at 100%; by the time you're displaying an image at 12.5 -> 25% you won't see it's effects, and if you down-sample the image for internet display then it's be sampled out anyway, so it's not a "biggie" if it's missed out altogether -- it just makes the image nicer to work on at high magnifications.

2. Content/Creative sharpening of say 4 pixels @ 40% but again just on the bird (same reasoning as in #1)

Normally no, but images that have areas of high-frequency detail and low-frequency detail may well require different sharpening treatment -- and bird photos are likely to fit that category. If you're talking bird and sky then my figures aren't going to have any noticeable effect on the sky anyway (unless there are clouds).

3. .3 px at 50% for posting on the web, and likely selective if it is a bird against the sky etc..

No - you wouldn't normally have to make output sharpening selective.

However, if I am sending out the jpeg image for printing full size, 4000 x 6000 pixels.... Do I need to sharpen the image that I just processed after I change it to a jpeg from a DNG (from Lightroom) or a PSD file (from Photoshop after final edits)... this is what is confusing me...

No - the conversion between file formats won't make any difference providing you're not trying to compress it excessively (setting 10 to 12 in Photoshop will be fine). The effectiveness of the sharpening is going to mostly depend on the ratio of the pixel dimensions to the printed size (generally, the higher the resolution and the smaller the size, the more any given sharpening gets "hidden") (often things that look over-sharpened on screen will print fine).

[Brownbear]

Thank you Colin. The further clarification is very helpful and truly appreciated!

I tried your method on several bird images yesterday, and yes sharpening selectively with a duplicate layer was not too hard. I must say that at full size I can see bright little speckles in the sky that disappear when I erase the unsharp mask with the brush. (Not at all noticeable in smaller sizes) So I'm going to keep practicing with your method.

Not a soul has mentioned DNG files, so I suppose that no one processes these types of files, ie; just raw, psd, or tiff before converting to a jpeg?

Thank you.

There are techniques in Photoshop that make selective sharpening ridiculously easy. Probably the easiest is to just duplicate the layer - sharpen the top layer - and then just erase the portions you don't want sharpened. Should't take more than 30 seconds.



No - capture sharpening is usually applied to the whole image, but having said that, it only makes a difference for when you're working on the full resolution image at 100%; by the time you're displaying an image at 12.5 -> 25% you won't see it's effects, and if you down-sample the image for internet display then it's be sampled out anyway, so it's not a "biggie" if it's missed out altogether -- it just makes the image nicer to work on at high magnifications.




Normally no, but images that have areas of high-frequency detail and low-frequency detail may well require different sharpening treatment -- and bird photos are likely to fit that category. If you're talking bird and sky then my figures aren't going to have any noticeable effect on the sky anyway (unless there are clouds).



No - you wouldn't normally have to make output sharpening selective.
[I]


No - the conversion between file formats won't make any difference providing you're not trying to compress it excessively (setting 10 to 12 in Photoshop will be fine). The effectiveness of the sharpening is going to mostly depend on the ratio of the pixel dimensions to the printed size (generally, the higher the resolution and the smaller the size, the more any given sharpening gets "hidden") (often things that look over-sharpened on screen will print fine).

[Colin Southern]

Thank you Colin. The further clarification is very helpful and truly appreciated!

I tried your method on several bird images yesterday, and yes sharpening selectively with a duplicate layer was not too hard. I must say that at full size I can see bright little speckles in the sky that disappear when I erase the unsharp mask with the brush. (Not at all noticeable in smaller sizes) So I'm going to keep practicing with your method.

You're very welcome.

I should have mentioned that USM threshold also comes into it, depending on ISO. If you're shooting at a high ISO then you may need to set a threshold in the 4 to 7 region, which frankly, pretty much wipes out the normal 0.3 @ 300% USM for capture sharpening. Best guess is that's probably what you're seeing the effect of (capture sharping on high ISO files producing artifacts).

If you're wanting a quick and dirty technique for removing stuff like that (stuff that's there anyway, not just produced from sharpening) then duplicate the layer (ctrl J) then use the dust and scratches filter set to 1 or (occasionally) 2 - then erase as per normal to cut through to bits that shouldn't be softened. PS: when you're erasing, the size and softness of the brush make a huge difference, and you can vary the opacity of the bottom layer to see where you have / haven't erased.

Not a soul has mentioned DNG files, so I suppose that no one processes these types of files, ie; just raw, psd, or tiff before converting to a jpeg?

I work exclusively with DNG, but they're only useful to "edit" at the "global" stage (setting overall levels and colours) - I'd certainly never muck around doing anything that's better catered for with a pixel-based editor like Photoshop.

My personal opinion is that if one has Bridge / ACR / Photoshop combo then LR becomes obsolete (to be honest, LR drove me nuts - especially it's interface. I find bridge / ACR much more efficient).

[Manfred M]

My personal opinion is that if one has Bridge / ACR / Photoshop combo then LR becomes obsolete (to be honest, LR drove me nuts - especially it's interface. I find bridge / ACR much more efficient).

Let me second that; ACR and the Lightroom Develop module are technicallly identical; except for the user interface. Like Colin, I find the one in ACR far more user friendly and usable.

[Brownbear]

Colin,

Thank you for taking the time to explain in further detail (including DNGs), and for the quick and dirty trick for removing stuff like that. Sounds simple enough to do, and your directions make it so. Yes, they are images photographed at a higher ISO's. I will try it.

Manfred,

Thank you for sharing. I'm just most comfortable with the ACR in LR but I will try it out the ACR in Photoshop CC for a while and see how it goes. They do look to have very similar functions. It will save me a few steps and file format mix-ups.

[Shadowman]

Has anyone tried with success Octave Sharpening?

http://www.varis.com/StepByStep/sharpen/Sharpen.html