Sharp or not?

[Grace Scalzo]

I have seen many suggestions in the avian forum suggesting more sharpening on an image that to my eyes does not look sharp to begin with, so I would like to discuss that a bit. In my workflow, after evaluating images for other merits, I look at the possible keepers in Light room and enlarge them to 1:1. I then examine them for sharpness and hit the x if they do not look critically sharp, which really is quite evident at 1:1. So my basic question is, to what extent will applying sharpening in post produce an acceptable result in post processing? My own answer to that is "never" altho there are some images that I have in my collection that I have kept for other reasons (Rarity, a memory, a reminder to myself that I messed up!)

[Axel Hildebrandt]

Very good point! I have noticed that images that I have processed on my 24'' monitor look a little less sharp on my 15'' laptop. Since there are these slight variations, I might sharpen a little more, just in case. When a posted image does not look very sharp, I assume that it might be a similar issue and some people simply prefer their images to be less sharp.

I normally don't keep an image unless it is sharp at 100% just as you described. If a bird is large in the frame and not sharp, you could probably get away with more sharpening after resizing but this is where people have different standards. If it is a rare capture or the only image of a species, I can see the point of sharpening more to give it the appearance of sharpness but in doubt disclosure is the best policy as far as I'm concerned.

[John Chardine]

Grace- In my experience, images that start out sharp at 100%, sharpen up wonderfully in post-processing, images that are not, sometimes do, sometimes don't. I occasionally test-process RAW images that are not as sharp as I would like them at 100%, and the results are often IMO very good, at least for the web and smaller prints up to 8x10". Dumping criteria are personal rules and depend as you suggest on things like rarity, but also on what you have in the can already for that species or whatever. I definitely loosen my criteria when I have just made a series of images of a species not photographed by me before, and if the keepers are close to good at 100%, as mentioned, the results after sharpening can be great.

[Grace Scalzo]

Thanks for your replies, John and Axel. I'm in the process of evaluating and deleting alot of images. Maybe I need to look just a bit longer at some of the ones that I deem to not be sharp before trashing them, again, based on some criteria.

[David Thomasson]

Grace, maybe it would help if you posted a 100% crop of an image that you regard as marginal -- or maybe a link to the full image -- and see what others can do with it. Results, of course, often depend on how you go about sharpening, and there are many ways to skin that cat. A raw file would be preferable for this.

[John Chardine]

I agree David, this would be useful.

[Grace Scalzo]

Well, I could find one (well, many) that I consider marginal. I delete these for the most part. My bigger question had to do with why further sharpening is suggested on some images in avian that are clearly not sharp from the get-go. I'm not talking about the image that is sharp but just needs a little more for web presentation, or one where the sharpening method could be improved, but again, those that are soft. To me, they lack detail, look muddy, and no amount of sharpening will help.

[David Thomasson]

Well, I could find one (well, many) that I consider marginal. I delete these for the most part. My bigger question had to do with why further sharpening is suggested on some images in avian that are clearly not sharp from the get-go. I'm not talking about the image that is sharp but just needs a little more for web presentation, or one where the sharpening method could be improved, but again, those that are soft. To me, they lack detail, look muddy, and no amount of sharpening will help.

I had your original question in mind: "to what extent will applying sharpening in post produce an acceptable result in post processing?" Of course, it depends on the image, but I thought it might help answer your question if you had an image or two that seemed marginal. Raw images are naturally a bit soft, which is why a little capture sharpening in ACR is almost always necessary. Beyond that, I've worked with some images that started off looking rather soft but turned out very nicely with a couple of rounds of sharpening -- once in ACR, and a different method inside Photoshop.

[Grace Scalzo]

David, you really have answered my question. I have always thought of sharp as an absolute. I enlarge it, look at it, and you can just see a really superbly sharp image. But you have explained that there may be other ways of salvaging.....I've got a couple in mind, will post them in here in a couple days. (Knee surgery tomorrow, following, I plan to spend some time doing a real hard edit as I will not be in the field, hence the question)

[arash_hazeghi]

I have seen many suggestions in the avian forum suggesting more sharpening on an image that to my eyes does not look sharp to begin with, so I would like to discuss that a bit. In my workflow, after evaluating images for other merits, I look at the possible keepers in Light room and enlarge them to 1:1. I then examine them for sharpness and hit the x if they do not look critically sharp, which really is quite evident at 1:1. So my basic question is, to what extent will applying sharpening in post produce an acceptable result in post processing? My own answer to that is "never" altho there are some images that I have in my collection that I have kept for other reasons (Rarity, a memory, a reminder to myself that I messed up!)

Grace,
You can definitely sharpen up and get away with photos that are not critically sharp at 100% if the subject is large in the frame. Well, we have to define "critically sharp" first which is subjective to some extent. But with a 12+ mpixel camera even photos that look somewhat soft at 100% can be look deceivingly sharp when down-sampled to 1024 pixels or smaller and sharpened using advanced techniques. I personally do not enjoy photographs that are not tack sharp at 100% and trash them and that is why I am very critical about AF performance as well, because I don't take slightly soft images but many people do and some are to my surprise well established photographers ;) different people have different standards and as long as the final result is good it works! same is true about noise. This is why when small 1024 pixel samples are posted to demonstrate sharpness of a lens,AF performance or noise, I tend to dismiss them. With advanced image processing you can do lots of tricks ;)

[John Chardine]

The reason this (post-sharpening) is suggested Grace is very simple- often a little more sharpening takes an image to the next level. This does not work with all images but with some it does. You have to learn to recognise which images it works with and which images it doesn't.

[arash_hazeghi]

Another way to tell quickly if the soft image can be saved with sharpening is to use nearest neighbor interpolation in Photoshop, just crop the photo to your liking then down-sample to 1024 pixels clicking on "nearest neighbor" if the result is sharp and has good details then you can go back and start formal processing, at the end applying smart sharpen with a small radius and large amount (e.x. r=0.5 amount=80) will sharpen up things nicely. If the down-sampled image looks murky however there isn't much you can do...

[Roger Clark]

All,

From the multiple points made so far there are several things to consider.

The modulation transfer function (the apparent sharpness) varies between monitors. This is especially true of LCD versus CRT monitors. If, for example, you view an image on a CRT monitor that was prepared on an LCD, it will likely look soft.

Similarly, the blur filter strength varies between cameras, so sharpness out of a camera, even with the same lens at perfect focus will vary from body to body.

Most of the "sharpening" tools that people discuss are actually not real sharpening algorithms. For example, unsharp mask is really an edge contrast enhancement and does not actually sharpen. High edge contrast gives the (false) illusion of sharpness. We had a big thread about this a while back and it is in educational resources now:
Important Sharpening Information! (http://www.birdphotographers.net/forums/showthread.php?18534-Important-Sharpening-Information%21)

http://www.birdphotographers.net/forums/showthread.php?18534-Important-Sharpening-Information!

So, with real sharpening tools, slightly unsharp images can be made quite sharp.

Finally, I would say the subject trumps sharpness. A great subject in a great pose/interaction, trumps perfect sharpness. While it is nice to have sharp images, and I strive for them, I strive more for interesting subjects doing interesting things. I'll close with a link to a 3-megapixel image that was printed full page in the fall 2004 Natures Best (page 21 if you have that issue). The image is not sharp, yet placed in the international contest, has sold in galleries as a 16x18-inch print and is hanging in many homes and businesses:
http://www.clarkvision.com/galleries/gallery.bird/web/c01.14.2003.img_5113.egret-flight.f-600.html

There are many such examples of seemingly unsharp images placing/winning in photo contests and on display and selling in galleries. The key is the subject, not the sharpness.

Roger

[Grace Scalzo]

Thanks everyone, I am learning a ton from this thread!! Appreciate your explanations and help.

[John Chardine]

Roger- You mention "real sharpening tools" and suggest the USM isn't one. Is there a real sharpening tool that is available in Photoshop or as an add-in? I am aware of the sharpening algorithm that astronomers use (can't remember the name off-hand) but mortals would like something that ran in software they were familiar with like Ps.

[Roger Clark]

Hi John,

My favorite sharpening tool is called Richardson-Lucy deconvolution. It is an iterative computational method that takes a few minutes on a typical image. I do not know of a plugin for photoshop (I too really wish there was one). I use ImagesPlus (around $220 now, http://www.mlunsold.com/). It is also available in Iris (free) (http://www.astrosurf.com/~buil/ although the site is down for the moment).

ImagesPlus (probably iris too) do not do color management. I write a 16-bit tiff file from photoshop, read it into imagesplus, run the deconvolution (sometimes some different versions as discussed in the sharpening thread), write the results as 16-bit tiff, then read them into photoshop and assign the color space. I then overlay them on the original and select what parts of which deconvolution to go on the final image (described in the sharpening thread).
Often, like with USM, a simple deconvolution works for the whole image. But if you want to fix depth of field limits, or motion blur, it takes multiple runs and different parameters work best for different parts of the image.

Roger

[Grace Scalzo]

Roger, what's the difference between the free one and the $220 one?

[Roger Clark]

Roger, what's the difference between the free one and the $220 one?

Grace,
I have not used the free one so I do not know. ImagesPlus is a commercial product with a full gui and is relatively simple to use. Some things are a little funky compared to photoshop, and some things work better, and then there are tools that photoshop does not have. Iris is a capable program and I see many use it but I have not seen the interface. The implementation of Richardson-Lucy should be similar/same in both programs as far as the technical computations. ImagesPlus shows each iteration so you can decide to stop when the image is to your liking and the image is converted to 32-bit floating point and computations are done with 32-bit floating point math.

Roger

[William Malacarne]

Roger

Is this the one.

http://www.astrosurf.com/buil/us/iris/iris.htm
IRIS astronomical images processing software ver 5.59

Thanks
Bill

[John Chardine]

Thanks Roger. The link to the previous thread on this subject was very useful as well.

[Roger Clark]

Roger

Is this the one.

http://www.astrosurf.com/buil/us/iris/iris.htm
IRIS astronomical images processing software ver 5.59

Thanks
Bill


Yes. Note the Saturn image on that page.

Roger

[Robert Park]

I just wanted to post the Richardson-Lucy deconvolution is also in Raw Developer and it does wonders when you do upsizing of the raw when outputting the file.
Sorry it is MAC only though
http://www.iridientdigital.com/products/

[Axel Hildebrandt]

I just wanted to post the Richardson-Lucy deconvolution is also in Raw Developer and it does wonders when you do upsizing of the raw when outputting the file.
Sorry it is MAC only though
http://www.iridientdigital.com/products/

Thanks, Robert!