Lens sharpness vs post-production sharpness

[NorthernFocus]

Strangely enough, I rarely look at MTF charts...

Not so strange. Well maybe for an engineer it is a bit strange. If so then I'm strange too because I don't look at the charts either. I do look at some of the sites that do testing. DxO is OK for primes but pretty useless for zoom lenses as they base there findings on the optimum spot in the range. LenScore and Photozone are pretty good. Some argue that these test sites aren't valid due to test methodologies etc. But from a practical standpoint we're choosing among available options and the test labs are useful for relative comparisons. So they do factor into my decision making process.

I've gone full circle on OEM vs third party lenses. In film days I shot almost all third party lenses because it's all I could afford. After going digital my finances were in better shape and I went exclusively Nikon. In recent years Nikon has effectively surrendered the APS-C format market and third party vendors have definitely taken advantage in that space. Sigma has really upped there game with the "Art" series and that is reflected in the price that they command in the market. They've also removed one of the main barriers to entry by coming out with the lens dock to allow future firmware updates as new camera bodies are released. I have no hesitation at all purchasing third party, particularly Sigma.

[Abitconfused]

With increased sophistication in engineering we must someday reach the point where any increase in sharpness, however derived, cannot be detected by the consumer.



I will throw this into the discussion. The Nikon 105 f/2.8 VR has a DXO sharpness rating of 21 significantly less sharp than the Nikon 24mm f/1.8 at 30. But, thanks certainty in part to the VR, it was very functional even at ISO 1000. So I agree that sharpness is situational, the argument is more entertaining than practical, and, that the color yellow is not like other colors.

[NorthernFocus]

With increased sophistication in engineering we must someday reach the point where any increase in sharpness, however derived, cannot be detected by the consumer....

That is only true if nothing else in the equation changes. But as sensor tech improves better optics will allow higher quality images with smaller sensors. Imagine imagery out of a camera the size of a P/S with image quality comparable to a FF DSLR with a prime lens costing thousands of dollars. So optics will never get "too good".

[george013]

In general I buy a lens as a physical object. Myy pp skils I must buy somewhere else.

What I don't understand on the DxO reviews is that they claim the test results with different camera's. For Nikon only they have 26 camera's. I just can't believe they're used all individual for the lens tests.

When looking after a lens test I mostly use Photozone and in the chapter sharpness I pay attention not only to the MTF but also to the differences between centrer, border and extreme.

George

[Black Pearl]

I will throw this into the discussion. The Nikon 105 f/2.8 VR has a DXO sharpness rating of 21 significantly less sharp than the Nikon 24mm f/1.8 at 30. But, thanks certainty in part to the VR, it was very functional even at ISO 1000. So I agree that sharpness is situational, the argument is more entertaining than practical, and, that the color yellow is not like other colors.

You have to remember that a macro lens is often optimised for close focus performance and most less test are done at far distances. Unless DXo states this or explains their tests the info could well be useless.

[xpatUSA]

With increased sophistication in engineering we must someday reach the point where any increase in sharpness, however derived, cannot be detected by the consumer.

Nice shot that, Ed. But, when I printed it out on my Apple ImageWriter, it went blurry and lost all it's color - and it doesn't look too good on my old VGA monitor either (pardon the English humor, just kidding).

Here's some good reading to make you more than just a bit confused:

http://www.bobatkins.com/photography.../mtf/mtf2.html

Atkins writes in a clear and concise manner without going too deeply into stuff . .

. . and beats DXO's peculiar metrics every time

[chauncey]

Those MTF table references, at least to me, indicate that the straightest and highest lines indicate
the best lens. My two Canon favorites, 300 f/2.8 and 180 macro, have those virtually straight lines.

What might I be overlooking?

[Manfred M]

What might I be overlooking?

A straight line could also indicate a terrible lens. Data close to a value of "1", horizontally along the top of the graph is a "perfect" lens, which doesn't exist. Having a lens with an MTF that is close to "1" from centre to edge, even if it varies a bit would be a good lens.

The left side of the chart shows you performance at the centre of the frame and the right side what is happening at the edge of the frame (where real good lenses do tend to drop off). The charts tend to show you what happens when the lens is wide open and also what happens at the "sweet spot" of around f/8.

On zoom lenses, the manufacturer will often choose a single focal length, so performance at other focal lengths could be quite bad, buy you wouldn't know it from the MTF chart.

[Brev00]

One other thing I have found is that when I increase the level of sharpness in my captures due to camera upgrades, lens purchases, technique refinements, etc., I also spend more time in pp to bring out that sharpness to its best potential. When my images are more wonky, sharpening tends to decrease the quality of the image. So, I am not sure I will get to a point where my gear is so sharp I don't need any pp. Maybe the reverse.

[NorthernFocus]

One other thing I have found is that when I increase the level of sharpness in my captures due to camera upgrades, lens purchases, technique refinements, etc., I also spend more time in pp to bring out that sharpness to its best potential. When my images are more wonky, sharpening tends to decrease the quality of the image. So, I am not sure I will get to a point where my gear is so sharp I don't need any pp. Maybe the reverse.

Interesting how differently we all do things. I'm the opposite. With my two high rez camera bodies that don't have AA filters, I often turn sharpening off in the adjustments panel in LR and rely on the default "sharpen for web" when converting to jpegs for web display. If I'm going to print then I do sharpen appropriately.

On the other hand, due to the above, it is very frustrating when I spend time sharpening virtually every image captured with a body that cost several times as much

[xpatUSA]

By coincidence this very day, I discovered that one of my lenses is noticeably soft at a particular zoom setting. A pity, because I was trying to record some long-overdue trim work on the front of my residence in excruciatingly fine detail.

Quite frankly, it was bloody irritating and all the more so because the camera was not the highest res in my collection. That is to say, at 2640x1760px (4.7MP), you should be able to put the world's worst lens on it and just not see any aberrations, diffraction, blah-di-blah. Would that it were so

To check, I put that lens on my highest res camera at the same lens settings and repeated the shot . . . yep, still soft.

Back to the other camera, I tried a less than well-respected "super-zoom" (18-200mm) at the same focal length as Mr. Softy. It came much mo' better! So, I've just ordered a replacement - a Sigma 17-70mm Contemporary which has pretty good reviews.

This long story is to demonstrate that, for me anyway, a sharp lens is important enough to buy a new one rather than continue to mess with sharpening excessively in post.

[Manfred M]

This long story is to demonstrate that, for me anyway, a sharp lens is important enough to buy a new one rather than continue to mess with sharpening excessively in post.

Moral of the lesson - you can't sharpen something that is soft to begin with. All the sharpening in the world will not help an image that was not properly focused or has a touch of motion blur.

[NorthernFocus]

Moral of the lesson - you can't sharpen something that is soft to begin with. All the sharpening in the world will not help an image that was not properly focused or has a touch of motion blur.

So true. All of the various sharpening tools in spite of their marketing claims can't correct an OOF image. The only real cure I know of is down sampling. I do have many images that aren't of sufficient sharpness to produce a decent sized print that look fine when reduced to 700px or so for web publishing.

[Brev00]

Ted,

I hope you enjoy the Sigma! I have been using the model before the C for over five years. According to reviews, the C is sharper on the long side which is nice since it is at 70mm that you can take full advantage of its close focusing and max magnification. A very fun lens that can go from landscapes to selective focus on a dime.

[xpatUSA]

Ted,

I hope you enjoy the Sigma! I have been using the model before the C for over five years. According to reviews, the C is sharper on the long side which is nice since it is at 70mm that you can take full advantage of its close focusing and max magnification. A very fun lens that can go from landscapes to selective focus on a dime.

Thanks Larry,

I had the original model for a long while until it fell apart on me one day when I fumbled trying to mount it on a camera. The original got good reviews in the sharpness department too so, if the 'C' is better than that, I'll be a happy camper.

[Brev00]

I look forward to reading your impressions.