Lens sharpness vs post-production sharpness

[Abitconfused]

I see, for example, that the Nikon 20mm f/1.8 has twice the sharpness rating on the DXO website as the Sigma 14-24. Fine. But why isn't lens sharpness given in terms of post-production? I suspect that a very sharp lens benefits less from post-production sharpening than a lens with less than spectacular DNA. So is image sharpness a function of lens manufacture or post-production expertise or both? (I don't need to hear that only amateurs are concerned about sharpness. I am a sharpness addict and happy with that).

[xpatUSA]

I see, for example, that the Nikon 20mm f/1.8 has twice the sharpness rating on the DXO website as the Sigma 14-24. Fine.

Not a great fan of DXO's magic ratings for anything. Their "Mpix" for cameras being a case in point. But that gripe has little to do with your question.

So what is "sharpness" and how is it measured?

This gentleman knows a bit about that, it's well worth a read or two:

http://www.falklumo.com/lumolabs/art...ess/index.html

But why isn't lens sharpness given in terms of post-production?

Because there are too many variables involved, too many to list. Yesterday I was comparing "sharpness" between three different editors - not surprisingly they were all different. And it doesn't take much of a tweak on a slider to completely alter an MTF number or an edge response.

I suspect that a very sharp lens benefits less from post-production sharpening than a lens with less than spectacular DNA.

True, but kind of back-asswards, that, don't you think? Shouldn't we think instead of the final image as benefiting from a really sharp lens like your 18-35mm Art?

So, is image sharpness a function of lens manufacture or post-production expertise or both?

Both, IMHO.

(I don't need to hear that only amateurs are concerned about sharpness. I am a sharpness addict and happy with that).

For me, the sharpness of a lens per se can only be measured in the image plane - which does not necessarily include the camera's sensor. By including a camera sensor, it's raw to RGB conversion and post-processing, it is no longer just the lens that is being measured but an entire system.

I too value sharpness above most other things, BTW.

[Manfred M]

Ed - it really isn't nearly as simple as that. Lens sharpens has to do with resolution i.e how many distinct lines per mm does the lens resolve down to. This is totally independent of the camera and post-production. It is, however, one of the drivers as to how sharp your image will end up.

A second driver is your camera. Cameras anti-aliasing filters (AA) blur the image slightly to reduce the risk of moiré patterns and this is constant for a specific camera model, regardless of lens resolution. You also get some loss of sharpness when an image is reconstructed from the digital data in the raw file.

We try to compensate for that when we import the image and do a bit of sharpening to counteract the aforementioned issues with the digital capture process. This is often referred to as "import sharpening" and this enhances the areas where light shades and dark shades meet. This is totally different than lens sharpness.

A second level of digital sharpening is sharpening you do during post-production, and similar to import sharpening, here we locally enhance areas of the image that are a bit soft. In portraiture, I often crisp up the eyes and lips. Sometimes eyelashes and eyebrows can use a little bit of help too. This is sometimes referred to as "in-process sharpening".

The final bit of sharpening occurs when we prepare the image for output. The size of the image we output rarely is exactly the same size as the image we worked on in post-processing. We will either upsample the image (make it larger) or downsample the image (make it smaller). Both operations affect the inherent sharpness of the image an we often need to sharpen it (again by enhancing the boundary between dark and light tones)'

If we print the image we had to sharpen accordingly. A print of a glossy paper will tend to look sharper than one printed on a matte paper. A glossy paper tends to absorb less in whereas a matte paper may bleed a bit, resulting in some softening in the shot. This has to be adjusted for, so a bit more sharpening will required to make a matte print than a glossy print. Again this sharpening deals with where light and dark tones meet in the image.

Bottom line, lens sharpness is important as this dictates how finely detail is resolved in the image. All of the sharpening in post deals with areas where light colours and dark colours touch, so something quite different from lens sharpness.

[Shadowman]

Some websites use both criteria (display properties and output), whenever considering experts opinion find out how they evaluate; it may or may not relate to your needs.

http://www.imatest.com/docs/sqf/

[DanK]

it might help to distinguish between resolution, which Manfred defined--which I think is what you mean by the sharpness of the lens, and acutance, which is the perception of sharpness. For example, this is the definition of the latter from Wikipedia:

In photography, the term "acutance" describes a subjective perception of sharpness that is related to the edge contrast of an image. Acutance is related to the amplitude of the derivative of brightness with respect to space. Due to the nature of the human visual system, an image with higher acutance appears sharper even though an increase in acutance does not increase real resolution.

What you are doing in PP is mostly changing acutance. For example, sharpening is increasing the change in brightness at edges.

[tao2]

So is image sharpness a function of lens manufacture or post-production expertise or both?

Like Ted, ah take DxO tests/ findings, when it comes tae Nikon lenses especially, with a large spoonful of salt. Over many years, they have shown a remarkable affinity with them tae the exclusion of all others; and their tests have been tailored tae suit Nikon. Now, with my bias also out of the way...

The 3 or 4 sharpest lenses ah own are all 30-40 years old. The outstanding one is an Industar 61 LZ MF, built in 1978, shoots macro(ish), focusing down tae 33cms (the book says 0.9m). Its resolving power, centre/edge is a huge 40/27 lines/mm . It outperforms any of the other pile of lenses in the cupboard and a great many others beside. IMO it's the best FSU lens ever, a Tessar design, great (radioactive ) glass, all metal build. It produces excellent images, on film or digital. It's the lens every time for me.

Ah'd compare image sharpness through a good lens, tae a scalpel in expert hands. Image sharpening in software like a butcher with a cleaver.

[NorthernFocus]

Now I understand your user name, Ed

[Abitconfused]

I like the idea of acutance as it incorporates illusion into the equation as we may expect for any art form (art being a lie that is also the truth). Selective sharpening, selective saturation, and selective contrast enhancements may do more for the product than pixel resolution. The medium is to convey the meaning of a subject of interest not to dazzle oneself with our own capabilities. Yet I am a slave to technology and that chromosome is never sated. Should I not buy the Nikon 20mm f/1.8 as I have a Sigma 12-24mm? Are a fool and his money soon parted? That is the question!

[Shadowman]

I like the idea of acutance as it incorporates illusion into the equation as we may expect for any art form (art being a lie that is also the truth). Selective sharpening, selective saturation, and selective contrast enhancements may do more for the product than pixel resolution. The medium is to convey the meaning of a subject of interest not to dazzle oneself with our own capabilities. Yet I am a slave to technology and that chromosome is never sated. Should I not buy the Nikon 20mm f/1.8 as I have a Sigma 12-24mm? Are a fool and his money soon parted? That is the question!

Just considering Sigma alone, you already have quite a few Sigma lenses; couldn't you apply the same analysis to those lenses that you are considering for the potential new lens? If your current lineup of Sigma lenses don't meet expectations and are rated similar to the potential new lens then you'll have your answer.

[Manfred M]

Ed - while the Nikkor lens might be sharper than your Sigma, that doesn't necessarily mean you will get sharper images. A bit of camera shake, a bit of missed focus are all going to impact the sharpness of your image more than the lens.

The biggest improvement I made in sharpness (especially for landscapes) was investing in a decent tripod and head. The bad news is that the new lens is going to be a lot less expensive...

[xpatUSA]

. . . . Yet I am a slave to technology and that chromosome is never sated.

Should I not buy the Nikon 20mm f/1.8 as I have a Sigma 12-24mm?

Ed, I believe you can answer your own question. I would take a look here at both lenses:

http://www.slrgear.com/reviews/index.php

If it's sharpness you're after, I can't say that the Sigma 12-24mm is all that impressive, according to them:



That blur units thingy is pretty good - you can use the sliders to set zoom and aperture and thereby see how their sample of one lens behaves

[Abitconfused]

Oddly I am more tempted by the Sigma than the Nikon at 20mm. The Sigma may not be quite as sharp but it seems more professional and solid. Some photographers advise that aftermarket lenses may be disowned by evolving camera bodies but I doubt that as both camera and lens firmware can be updated. Thoughts? P.S. This may well be my last lens purchase in this price range as I have about every FOV covered and further acquisitions offer deminishing returns.

[Manfred M]

Some photographers advise that aftermarket lenses may be disowned by evolving camera bodies but I doubt that as both camera and lens firmware can be updated.

And that is exactly the crux of the issue and the problem is that this doesn't happen. Nikon does not have third party lens information in their camera firmware (but they ensure compatibility their own lens line). The third party lens manufacturers have no (financial) incentive to update the firmware in their legacy lenses to run on new bodies; they could but why would they?

If you buy a third party lens, this is a risk that you take as it may be unusable on a new camera body. On the other hand, you could be lucky and it might work.

[Black Pearl]

Oddly I am more tempted by the Sigma than the Nikon at 20mm. The Sigma may not be quite as sharp but it seems more professional and solid. Some photographers advise that aftermarket lenses may be disowned by evolving camera bodies but I doubt that as both camera and lens firmware can be updated. Thoughts? P.S. This may well be my last lens purchase in this price range as I have about every FOV covered and further acquisitions offer deminishing returns.

I very much recommend you try one of these before you purchase one. I sold them right from launch, used them on various bodies (the first time I got to see it on a FF body was with a Kodak DCS-14n) and have had a fair bit of feedback from photographers who have had them and its not what you'd call a stella lens. Yes at the time it came out it was impressive for its width and yes they are reasonably well made you have to stop them down to optimum aperture to get any sort of useful image (at wider apertures they are grim) and even then on a modern high pixel body you're going to be fighting it for results every time you use it.

[Brev00]

I am confused. You say you seek ultimate sharpness but you prefer the Sigma for more innocuous considerations like professional quality. I believe the Nikon has the special coating that adds a little zip of extra contrast to the images particularly shooting into bright light. From your post, it seems like you are after something subjective, a feeling of sharpness, more than what we find in mtf charts. A prime like the well reviewed Nikon 20 will outperform your zoom on all measures except focal range flexibility. Get the Nikon and enjoy! (my review is based solely on what I have read).

[Abitconfused]

The Nikon is looking better and better. I don't feel like too much of a spoiled brat for wanting a wide angle prime. My expectations for the Sigma are/were based on my experience with the Sigma 50 ART on my Nikon D810. It is frighteningly sharp even handheld. My fingers are trembling over the shopping cart button even as we speak...

[Rob Ekins]

I worked out a long time ago that 'Sharpness' is a state of mind, buy what you can afford and work with it..

[James G]

I have to be honest and admit that while I'm thoroughly enjoying this thread. I'm finding it both informative and educational about technical aspects of lens technology and design that I've never really considered.

BUT, it does remind me somewhat of a time when it was absolutely fundamental to understand how many angels could stand on the head of a pin!

My rather simplistic approach to the 'original' question is that I suspect lens sharpness probably becomes significant for images that explore the extremes of certain types of capture. I suppose, but am not expert enough to be sure that this would be significant for microscopy and high magnification macro's, or astro-photography for instance.
For the kind of photography I generally engage in though, final image acuity is more likely to be of relevance and the impact of lens sharpness at capture, while important, is trumped by post processing.

My general approach to buying new lenses, has been to get hold of it, attach it to my camera and shoot of some real world captures before making the decision. Oh yes, I also take the real world experience of people I trust into account. ..... Usually my CiCs brethren!!!

[chauncey]

I rely on MTF tables to measure lenses acuity.

[Manfred M]

Strangely enough, I rarely look at MTF charts. I usually approach lens purchases from a "need" basis, so price, focal length / focal length range and maximum aperture are usually what drives a purchase decision. I generally go with Nikon lenses and only go third party if Nikon does not make the lens I am after. That often means I am only considering a single lens model, so looking at the MTF chart is not going to influence my buying decision anyways. I will check the reviews and Nikon has put out some lenses that were so-so optically, but I would not have bought these anyways.

Some of the older Sigma and Tamron lenses were built to a price point (a nice way of saying optical / mechanical performance were not outstanding), but more recently Sigma, with their Art line, has been putting out some stunning lenses. They have also out some lenses with sub-optimal performance; the one you are looking at as well as the hyper-expensive (and physically huge) f/2.8 200 - 500mm lens seem to be more marketing exercises to say these are shorter / longer / faster that the competitions.

Unless you are shooting from a tripod, the actual lens sharpness can be a bit of a moot point (less so in the focal length range you are considering), as motion blur (in spite of stabilization) and missed focus are going to affect image sharpness far more than lens performance.