Understanding Perceived Sharpness

[feppe]

I just came across this thread: http://www.fredmiranda.com/forum/topic/833110/0. The OP says it seems "photographers with Nikon cameras and lenses more easily create the unique special look (beautiful sharp, crisp, fresh images with dreamy look, and colors which pop!)". I've heard comments like this by others. This is similar to what I've noticed.

So that's what I've been doing wrong all these years - time to sell my Canon and Mamiya gear!

In all seriousness, nobody is a great photographer due to their gear. Some subject matter requires specialized gear and certain audiences expect specific quality, but in general a camera is a camera, and the greats produce good work whether it's with a Holga or a P65+.

[Jeremy Payne]

The OP says it seems "photographers with Nikon cameras and lenses more easily create the unique special look (beautiful sharp, crisp, fresh images with dreamy look, and colors which pop!)". I've heard comments like this by others. This is similar to what I've noticed.

Nonsense.  Pure nonsense.

[Playdo]

In all seriousness, nobody is a great photographer due to their gear. Some subject matter requires specialized gear and certain audiences expect specific quality, but in general a camera is a camera, and the greats produce good work whether it's with a Holga or a P65+.

Similar to saying paint is paint, and the greats produce good work whether it's with oil or acrylic. True, they do produce good work, but the results are not the same. For me this isn't about if one piece of gear is better than another it's about understanding certain characteristics of a photograph.

[Slough]

Nonsense.  Pure nonsense.

I don't think it is nonsense, in the sense that there are differences between brands, and people have a subjective appraisal of a given image quality. I am sure there are people who favour Canon over Nikon for various reasons, some real, some not, and maybe most people do not care.

I use Nikon, and they are renowned for a 'contrasty' image. I have a book on macro photography by Paul Harcourt Davies, and I can quite easily recognise the images taken with a Tamron 90mm F2.8 macro lens, and a Nikon 105mm F2.8 AFD lens. The Nikon have more 'snap'. I have owned both lenses, and I also see the same difference in my images, which is why I sold the Tamron. I mention the book as it is something others might have on their bookshelves, hence they can see if they agree or not. Now it so happens that of the lenses that I owned, the Tamron had higher resolution (by a fair margin according to my limited tests). So that means that contrast at the small scale was higher in the Tamron. But on the larger scale, the Tamron produced flatter images. That is not a fault, but a characteristic. I happen to prefer the Nikon look but I am sure others will disagree. I also have a Sigma 400mm F5.6 APO Macro lens, which has good resolution even wide open (on DX), but a slightly flat image quality with a warm cast. The lens is decent, but I do not like the image characteristics, and in this case I think the lens is not up to Nikon (or Canon) standards.

I have not used Canon, so I cannot comment. I also hear people making comments about the Leica and Zeiss look, and I do not discount those comments. The design of a lens is all about compromise, and achieving the desired image quality. And part of that is how the MTF varies with spatial frequency.

If you have access to high end binoculars or spotting scopes, you might be surprised at the differences in contrast and colour rendition. You need to swap between two models to see the differences, as the eye (or rather the brain) quickly adapts and corrects for any colour cast.

[Playdo]

the Tamron had higher resolution (by a fair margin according to my limited tests). So that means that contrast at the small scale was higher in the Tamron. But on the larger scale, the Tamron produced flatter images.

And part of that is how the MTF varies with spatial frequency.

Hi Slough,

So it is mainly due to the proprties of a lens? Could you explain this a bit more as I'd have thought if the 'contrast on the small scale' was higher then the image would have more snap.

Do you notice these properties on Nikon lenses in general? I'm also unsure on what you mean about 'how the MTF varies with spatial frequency'.

Thanks

[Jeremy Payne]

I don't think it is nonsense

I never said there were no differences between lenses or sensors or whatnot ... but ...

This statement is pure nonsense ... pure imagination ...

"Nikon cameras and lenses more easily create the unique special look (beautiful sharp, crisp, fresh images with dreamy look, and colors which pop!"

PS - I own, shoot and prefer Nikon DSLRs

[Slough]

I never said there were no differences between lenses or sensors or whatnot ... but ...

This statement is pure nonsense ... pure imagination ...

"Nikon cameras and lenses more easily create the unique special look (beautiful sharp, crisp, fresh images with dreamy look, and colors which pop!"

PS - I own, shoot and prefer Nikon DSLRs

Okay, fair enough. It is a trifle exaggerated.  

[Slough]

Hi Slough,

So it is mainly due to the proprties of a lens? Could you explain this a bit more as I'd have thought if the 'contrast on the small scale' was higher then the image would have more snap.

Do you notice these properties on Nikon lenses in general? I'm also unsure on what you mean about 'how the MTF varies with spatial frequency'.

Thanks

I am no expert, but I will explain what I understand. But do search for the explanation of MTF on this site by MR, as it is clear and accurate, and will probably help you understand the issues. If you want to see how not to explain MTF, then search for an explanation by someone called Ken Rockwell. The word inept comes to mind.

MTF is a measure of contrast at a given spatial frequency. Simply put it records contrast along a line across the image plane. (I'm ignoring tangential and sagital issues.) If you photograph a series of black and white lines, the image should record an exact copy i.e. perfect black and perfect white, with sharp transitions. But in practice some contrast is lost. The black is not quite black, the white not quite white, and the transition is not sudden, but gradual. The MTF plot represents how close to perfect the contrast is along that line across the image plane. Clearly the one variable here is the spacing of the black and white lines. That is what I mean by spatial frequency. You can plot the MTF for very closely spaced lines, and for widely spaced lines. If you look at MTF plots from manufacturers such as Canon, you see that they usually present MTF curves for two spatial frequencies (and they also provide sagittal and tangential plots, but let's ignore that issue).

Regarding snap, I believe that the snap of an image is due to the contrast at mid spatial frequencies. TV camera lenses used to be optimised for contrast rather than resolution for the obvious reason that TV images had low resolution. Resolution is due to contrast on the very small scale. If you think about it, the resolution limit is where the contrast goes to zero (or at least to the point where you cannot separate lines). So to get high resolution you need to maintain contrast at small scales.

In reality to represent a lens accurately (at a given subject distance) you would need to plot MTF at a whole range of spatial frequencies, but people seem to think that 2 is enough. There are in fact many problems with MTF plots, which most people ignore, preferring to take them as gospel. They tell you nothing about IQ at infinity and close focus, or tendency to flare and ghost, for example.

[ErikKaffehr]

Hi,

The issue with flare is important and actually hard to measure. Flare depends on many factors. Antireflex coatings obviously, but also on blackness of mounting tubes in the lens, mirror chamber etc, internal baffling in the lens.

Placebo certainly plays a role.

In my view there is probably not that much of difference between lens families if we shoot raw in a correctly controlled color managed environmeent. On the other hand, correct color is not very usable, it would be flat. Therefore color is always adjusted using an "S" curve enhancing midtone contrast. Most pictures on the net may probably come from in camera JPEGS, with usually something like half a dozen different camera settings producing different colors. Add to this that most images may be in sRGB and most screens are not calibrated. Very few web-viewers handle color profiles correct anyway.

So there are many variables:

- In camera settings, say Nikon 6 Canon 6 that's just 36 combinations
- sRGB/Adobe RGB
- Raw processor if not in Camera JPEGs are used
- Screen calibration and settings (many modern screens are actually close to Adobe RGB, but color is normally assumed to be sRGB)
- Does viewer handle color profiles correctly?

[attachment=18257:ColorSam...anon_20D.jpg]

All this images were made from the same crop from a picture shot by a friend with a Canon 20D. Lightroom tries to reproduce different in camera setting when converting raw. You see that all images are quite different.
Lightroom doesn't have a lot of color profiles for the equipment I use (Sony).

Best regards
Erik

I am no expert, but I will explain what I understand. But do search for the explanation of MTF on this site by MR, as it is clear and accurate, and will probably help you understand the issues. If you want to see how not to explain MTF, then search for an explanation by someone called Ken Rockwell. The word inept comes to mind.

MTF is a measure of contrast at a given spatial frequency. Simply put it records contrast along a line across the image plane. (I'm ignoring tangential and sagital issues.) If you photograph a series of black and white lines, the image should record an exact copy i.e. perfect black and perfect white, with sharp transitions. But in practice some contrast is lost. The black is not quite black, the white not quite white, and the transition is not sudden, but gradual. The MTF plot represents how close to perfect the contrast is along that line across the image plane. Clearly the one variable here is the spacing of the black and white lines. That is what I mean by spatial frequency. You can plot the MTF for very closely spaced lines, and for widely spaced lines. If you look at MTF plots from manufacturers such as Canon, you see that they usually present MTF curves for two spatial frequencies (and they also provide sagittal and tangential plots, but let's ignore that issue).

Regarding snap, I believe that the snap of an image is due to the contrast at mid spatial frequencies. TV camera lenses used to be optimised for contrast rather than resolution for the obvious reason that TV images had low resolution. Resolution is due to contrast on the very small scale. If you think about it, the resolution limit is where the contrast goes to zero (or at least to the point where you cannot separate lines). So to get high resolution you need to maintain contrast at small scales.

In reality to represent a lens accurately (at a given subject distance) you would need to plot MTF at a whole range of spatial frequencies, but people seem to think that 2 is enough. There are in fact many problems with MTF plots, which most people ignore, preferring to take them as gospel. They tell you nothing about IQ at infinity and close focus, or tendency to flare and ghost, for example.

[ErikKaffehr]

Hi,

MTF says how contrast is transferred from subject to sensor for items of different size. Large items can be reproduced with full contrast but smaller items loose contrast. The item size is called linear frequency, because the terminology comes from signal processing theory. At small frequencies MTF is essentially 100% at resolution limit it is close to zero. Ideally MTF would fall essentially almost linearly with increasing frequency. MTF is limited by lens aberrations and diffraction. Aberration goes down stopping down and diffraction goes up. Diffraction is a property of light, not optics, so there are no lenses not affected by diffraction (at least not with present technology).

In practice MTF ids often presented as three different frequencies 10/20/40 lp/mm. If we look at an A4 print and compare to old 135 format what we call "full format" now, we would need to enlarge the image 8-times for A4. So:

10 lp/mm -> 20 items/mm on film -> item size 8/20 -> 0.4 mm

10 lp/mm -> 0.4 mm
20 lp/mm -> 0.2 mm
40 lp/mm -> 0.1 mm

Another way to see MTF is how small details will have a certain MTF like 50%, this is the way Imatest works.

Why don't you check this? http://www.normankoren.com/Tutorials/MTF1A.html

Best regards
Erik

Hi Slough,

So it is mainly due to the proprties of a lens? Could you explain this a bit more as I'd have thought if the 'contrast on the small scale' was higher then the image would have more snap.

Do you notice these properties on Nikon lenses in general? I'm also unsure on what you mean about 'how the MTF varies with spatial frequency'.

Thanks

[Slough]

That is a good point about JPG. As stated, each manufacturer will apply tone curves to the RAW data when creating a JPG image. And of coruse many people use settings such as enhanced saturation. That means that each camera will have its own look according to the users preferences, and the camera makers choices. Peronally I shoot RAW with neutral settings, as do many here I am sure.

[Playdo]

Hi,

Thanks very much for the info. So, correct me if I'm wrong;

1. The 'snap' is partly due to contrast on the small scale (referred to as micro-contrast?) at mid-spacial frequencies.
2. At capture this is largely a property of the lens.
3. In post processing, sharpening technique (ie unsharp mask etc) mimics this kind of contrast.
4. A tonal curve controls the overall contrast of an image.

Erik: As mentioned previously I really notice it in the shots taken by the D3 http://www.flickr.com/photos/tvchicklet/se...57600313161578/. I also see it a lot more in D700 shots. Recently I think that Nikon's increased the vibrancy/saturation in some of their presets (even if shooting RAW I think people are post processing to a similar style of these presets).

What I wasn't sure of is if the crispness was due to this 'overall' contrast (due to the increased vibrancy/saturation) or whether it was due to smaller contrast (a property of the lens/sharpening technique/pixel density).

[Jeremy Payne]

The 'snap' is partly due to contrast on the small scale (referred to as micro-contrast?) at mid-spacial frequencies.

Sure ... but let's be clear ... there is no generalized and statistically significant difference between "Nikon" and "Canon" in this regard.  Any difference you think you can divine from examining images is likely the result of differences in RAW processing and not the result of differences between "Nikon" and "Canon" optics.

I promise you that you (and others) would fail a blind image test based on your hypothesis.

You seem convinced of something that is a illusory, and the sooner you disabuse yourself of this shibboleth, the better.

[Slough]

Sure ... but let's be clear ... there is no generalized and statistically significant difference between "Nikon" and "Canon" in this regard.  Any difference you think you can divine from examining images is likely the result of differences in RAW processing and not the result of differences between "Nikon" and "Canon" optics.

I take it that you own/use both Nikon and Canon, and that you have performed these tests? As I've said before, not having used Canon, I cannot comment.

As I said, I see very real differences between the rendition of the Tamron 90mm F2.8 macro lens, and Nikon micro lenses, and also some of my old Nikon lenses have less 'snap'. I once owned the 28mm-105mm AFD zoom which is very popular, but I never liked the 'flat' and rather cool rendition. I even see this in online samples, which is curious as online JPGs are small, and processed/manipulated by RAW conversion software (sometimes in camera). Improved coatings, use of ED glass, and better designs will change the IQ.

[Jeremy Payne]

I take it that you own/use both Nikon and Canon, and that you have performed these tests? As I've said before, not having used Canon, I cannot comment.

As I said, I see very real differences between the rendition of the Tamron 90mm F2.8 macro lens, and Nikon micro lenses, and also some of my old Nikon lenses have less 'snap'. I once owned the 28mm-105mm AFD zoom which is very popular, but I never liked the 'flat' and rather cool rendition. I even see this in online samples, which is curious as online JPGs are small, and processed/manipulated by RAW conversion software (sometimes in camera). Improved coatings, use of ED glass, and better designs will change the IQ.

Ok, fair enough.  I have not performed extensive tests.

BUT ... I have seen enough images shot with enough lenses and cameras over the last 30 years to know in my gut that there is nothing shared across ALL nikon equipment that is somehow recognizably "nikon" and that it is this ephemeral aspect of the equipment that makes it "better" at producing "good" images than any other serious system.

There are differences between lenses and cameras ... but I'll stick by my 'nonsense' characterization until I see some evidence of this 'shared' aspect that cuts across all Nikon-branded equipment.

I'm not making the extraordinary claim that there is some Nikon magic that is better than the Canon magic ... I'm just the skeptic.  

Like I said, there are differences is equipment ... but to then infer from that that 'Nikon' is better than 'Canon' simply because not all cameras and lenses are alike is nonsense.

[Playdo]

there is nothing shared across ALL nikon equipment that is somehow recognizably "nikon"

No one mentioned it being across all Nikon equipment. I stated that I see something in the majority of Nikon photos that I see and then went on to say I've noticed it a lot in D3 and D700 shots. I don't appreciate you insulting my perception, or others'.

infer from that that 'Nikon' is better than 'Canon'

Was it mentioned/inferred that one was better than another? (that's a rhetorical question by the way).

[Jeremy Payne]

I don't appreciate you insulting my perception, or others'.

Get over it ... I'm sorry you feel insulted ... I'm just trying to be direct with you.  Most people are too polite to tell you when you are full of it.  

Was it mentioned/inferred that one was better than another? (that's a rhetorical question by the way).

rhetorical, smergorical ... You wrote this:

"... 'photographers with Nikon cameras and lenses more easily create the unique special look (beautiful sharp, crisp, fresh images with dreamy look, and colors which pop!)'. I've heard comments like this by others. This is similar to what I've noticed."

I said I thought this was nonsense and I stand by that.

[ErikKaffehr]

Hi,

In essential, yes sharpening affects mid spatial frequencies. I agree with your interpretation by and large.

Regarding color rendition my guess is that Nikon may have more vibrant colors. It may also be that at least some Nikon models can have sharper color filter grid array characteristics than corresponding Canon models, but I would not overemphasize these differences.

The effect of sharpening would be not visible in small pictures (like anything that fits on the computer screen). HD video is just two Megapixels ;-)

Best regards
Erik

Hi,

Thanks very much for the info. So, correct me if I'm wrong;

1. The 'snap' is partly due to contrast on the small scale (referred to as micro-contrast?) at mid-spacial frequencies.
2. At capture this is largely a property of the lens.
3. In post processing, sharpening technique (ie unsharp mask etc) mimics this kind of contrast.
4. A tonal curve controls the overall contrast of an image.

Erik: As mentioned previously I really notice it in the shots taken by the D3 http://www.flickr.com/photos/tvchicklet/se...57600313161578/. I also see it a lot more in D700 shots. Recently I think that Nikon's increased the vibrancy/saturation in some of their presets (even if shooting RAW I think people are post processing to a similar style of these presets).

What I wasn't sure of is if the crispness was due to this 'overall' contrast (due to the increased vibrancy/saturation) or whether it was due to smaller contrast (a property of the lens/sharpening technique/pixel density).

[Playdo]

I'm just trying to be direct with you.  Most people are too polite to tell you when you are full of it.

I'll help you out then Jeremy; you're full of it.

rhetorical, smergorical ... You wrote this:

"... 'photographers with Nikon cameras and lenses more easily create the unique special look (beautiful sharp, crisp, fresh images with dreamy look, and colors which pop!)'. I've heard comments like this by others. This is similar to what I've noticed."

I said I thought this was nonsense and I stand by that.

It was a reference to another post and the opinions of some other photographers. I was referring to the 'crisp and colours which pop'; two factors that I see quite evidently. I know you can't believe it Jeremy but your perception/differentiation of imagery isn't universal. 30 years eh? You'd think you'd know that certain equipment would more easily produce certain results by now. Anyway, just letting you know, I'll be sure to skip over your future posts.

[Playdo]

Thank you for the info Eric