DxO sharpness ratings of FE lenses: Zooms vs. Primes

[KenLee]

Sorry if this is a perennial question...

I've been shooting Large Format for a long time and am finally considering a digital camera. My typical print size is 16x20 inches and the A7Rii can produce those at 300dpi with no up-scaling required.

Looking at FE lenses for the A7rii on DxOMark, I notice that most of the zoom lenses have sharpness ratings below 30, while many primes have ratings well above 30. Are these numbers meaningful or reliable ? Is this due to poor edge performance of zooms, or is there more involved ?

Many thanks !

 

[GrahamWelland]

In my experience Sony lenses suffer some sample variation and others are ho-hum (my 24-70/4 & 70-200/4 come to mind). In my outfit the only ones that really work well for me are the 16-35/4, 55/1.8 and 90 macro. Plenty of threads here on the forum and elsewhere that discuss the relative lens quality and choices folks have made.

I actually use my Canon TSE lenses with Metabones adapter more often for landscape work these days.

 

[thrice]

Almost all zoom lenses have a great deal of variation across the field at different focal lengths.

It is simply the nature of the beast. Some zoom lenses (eg. 24-70 2.8 GM) are very good but you can be guaranteed (despite what some will say) that they are not tack sharp across the field at every focal length.

Sharp
Cheap
Fast

Pick 2

 

[The Ute]

Sorry if this is a perennial question...

I've been shooting Large Format for a long time and am finally considering a digital camera. My typical print size is 16x20 inches and the A7Rii can produce those at 300dpi with no up-scaling required.

Looking at FE lenses for the A7rii on DxOMark, I notice that most of the zoom lenses have sharpness ratings below 30, while many primes have ratings well above 30. Are these numbers meaningful or reliable ? Is this due to poor edge performance of zooms, or is there more involved ?

Many thanks !

I have both the 24-70GM as well as a set of primes: Batis 25, 35 2.8, 55 1.8
The 24-70 will not completely match the primes in terms of overall sharpness but it is very, very good and very versatile.
If image quality is your main consideration I'd go w a bag full of primes.

 

[jdphoto]

Sorry if this is a perennial question...

I've been shooting Large Format for a long time and am finally considering a digital camera. My typical print size is 16x20 inches and the A7Rii can produce those at 300dpi with no up-scaling required.

Looking at FE lenses for the A7rii on DxOMark, I notice that most of the zoom lenses have sharpness ratings below 30, while many primes have ratings well above 30. Are these numbers meaningful or reliable ? Is this due to poor edge performance of zooms, or is there more involved ?

Many thanks !

Ken, first wanted to say thank you for all the great info on your website. I appreciate all the effort you've put in to help fellow photographers! I've found the film scanner tips to be essential! I shoot film from 4x5 to 6x7 (Mamiya RZ ProII) and also shoot with a Sony A7r2. I find the Sony's output with good primes to be on par with MFD @ 16x20. The 55mm 1.8 rates close to the Otus and is one of the best prime lenses i've ever used for any format. The generous crops is also a benefit with the Sony. I use DXO as a reference and find them to be accurate, objective and quite useful. Thanks again.

 

[ggibson]

The DxO sharpness numbers are a summary of a huge number of variables, especially with zooms. There are measurements at each f-stop and (for zooms) a variety of focal lengths, so of course boiling it down to one number isn't going to tell the whole story. A zoom could beat a prime at a particular focal length and f-stop, but be worse across other areas of the zoom range and have a lower overall sharpness score. If you're debating between a zoom and a prime, you can dive into the sharpness measurements to compare them more directly instead of just going by the one sharpness score. For example, I have the 24-240mm lens, which is quite good at the wide end and not too far off something like the 28/2 prime, but the sharpness scores would lead you to believe the gap is much greater between these two.

But in general, it's safe to say that zooms are usually not going to be as sharp as primes (not only for FE mount too).

 

[tashley]

The DxO sharpness numbers are a summary of a huge number of variables, especially with zooms. There are measurements at each f-stop and (for zooms) a variety of focal lengths, so of course boiling it down to one number isn't going to tell the whole story. A zoom could beat a prime at a particular focal length and f-stop, but be worse across other areas of the zoom range and have a lower overall sharpness score. If you're debating between a zoom and a prime, you can dive into the sharpness measurements to compare them more directly instead of just going by the one sharpness score. For example, I have the 24-240mm lens, which is quite good at the wide end and not too far off something like the 28/2 prime, but the sharpness scores would lead you to believe the gap is much greater between these two.

But in general, it's safe to say that zooms are usually not going to be as sharp as primes (not only for FE mount too).

The key thing to remember about the zooms is that (as other above have noted) they vary across their range. However in my experience each of them has at least one focal length where at f5.6 it is essentially as good as the primes - especially at your print sizes - and thus you have the convenience of a zoom but with prime quality to hand if you need it, as long as you know where it is to be found, wrapped up in one lens. I find the wide f4, 24-70 f2.8 and 70-200 f4 are all extremely useful lenses and I use them more than the primes, even for serious work.

 

[jdphoto]

The key thing to remember about the zooms is that (as other above have noted) they vary across their range. However in my experience each of them has at least one focal length where at f5.6 it is essentially as good as the primes - especially at your print sizes - and thus you have the convenience of a zoom but with prime quality to hand if you need it, as long as you know where it is to be found, wrapped up in one lens. I find the wide f4, 24-70 f2.8 and 70-200 f4 are all extremely useful lenses and I use them more than the primes, even for serious work.

I agree that zooms offer a functionality that's not possible with primes. I shoot primes and often need a zoom to facilitate a composition or get better reach.

 

[KenLee]

Thank you for your insights.

Towards Ute's wise and concise recommendation of "a bag full of primes" and with portability and affordability in mind, I have started out purchasing a used 75mm Voigtlander Color Heliar f/2.5. It's a fraction of the price and volume of an 85mm Batis or GM, and (for me) a more desirable focal length.

This sample photo impressed me adequately: https://1.img-dpreview.com/files/p/E~forums/57060044/ee69758b49ce4ebb9e22004963b8fa3d

It would make a nice Carbon Pigment print at 16x20 inches: http://www.kennethleegallery.com/images/forum/getdpi/V75.jpg

 

[KenLee]

​

I just got the A7RII and Voigtlander 75mm 2.5 Color Heliar lens.

I'm really impressed with what can be done with this equipment.

 

[thompsonkirk]

In considering lenses for the work you do, I suggest you weigh contrast just as heavily as resolution. I looked at the lovely tones you're achieving on your website and can imagine how beautiful they look in your Piezo prints. But in shifting from film to a digital sensor you won't have the same tonal curve with toe and shoulder that you're accustomed to on film – it'll be just a straight line. This means it will be harder to extend the highlights in the way I can see on your website, and contrasty files will complicate the issue. (I'm not so worried about shadow detail, because you can tease a lot out of Sony files.)

The Sony FE lenses, whether prime or zoom, are contrasty, to a degree that someone accustomed to your tonal range might regard as downright nasty. Batis and Loxia will give you almost as much contrast. For this reason many who use A7rII for the kind of work you do shy away from them and prefer the older Mandler-designed Leica R lenses or the equally good but much cheaper Zeiss lenses in Contax/Yashica mounts. These are retrofocus lenses, so they don't create problems in the corners as wider-angle Leica M lenses usually do. No, they won't produce all the corner-to-corner sharpness or maybe even the center sharpness you're asking about; but 16x20, these days, is not a large print; so I think you'll get as much resolution from them as you need.

Above all, I would not go out and buy 'a bag of primes' to get started. I'd try one Sony FE or Loxia lens and one Leica R lens. See if, with the extended tonal range you want to print, an FE or Loxia is as satisfactory as an older and less contrasty design, and whether the sharpness of the latter is satisfactory.

FWIW, I'll mention my own preferences. I mostly use the Leica Tri-Elmar MATE and WATE, which are retrofocus lenses, with a 25mm C/Y as a bridge between them. I also have Zeiss C/Y primes and the wonderful C/Y f3.4 35-70.

I've only recently joined you in the 'Piezo cult' with the Pro inset and am delighted by the expansive midtones. I don't have as much trouble controlling the highlights as I did with ABW printing, but still have to work to keep the highlights well differentiated.

Kirk

 

[KenLee]

Hi Kirk - Thank you for your articulate and helpful post.

 

[thompsonkirk]

Thanks, Ken, in turn. I wrote my post after looking at your galleries on the web, but before reading your thoughtful/helpful articles. I had the embarrassed feeling that I'd been preaching to someone who probably knew ahead of time almost all that I'd written. Excuse me if I seemed to be talking down to someone so experienced and well-informed.

Kirk

 

[KenLee]

To the contrary, I had not thought about contrast at all.

At first I thought that a lens can't deliver more contrast than already possessed by the scene: it can only introduce veiling flare to some degree. But now I'm not so sure.

Over the years of working with view cameras and film, I "evolved" to avoid scenes that exceed the capacity of the equipment and process, so hopefully I won't end up with too many situations where the image needs heroic rescue. Professionals have to deliver the goods, no matter how challenging the conditions: it's a relief to be an amateur.

To carefully explore the huge variety of lenses out there takes time and money ... but I'm patient. So far I really like a few old lenses (as you recommended). The 55mm Micro Nikkor is my all-around first choice. If it didn't require an adapter, it would be very compact on a Sony camera.

If Microsoft can make software for Apple, why can't Nikon make lenses for Sony ?

 

[Vivek]

Ken, You can mount the MN 55/2.8 in a Sony - L39 adapter. Remove the 39mm t mount and use the 3 screws in theadapter to secure your micro nikkor. After this, it will only allow focus from infinity o 0.3m and not closer. I use an Olympus Macro 50/3.5 this way.

 

[thompsonkirk]

HI, Ken –

"At first I thought that a lens can't deliver more contrast than already possessed by the scene: it can only introduce veiling flare to some degree. But now I'm not so sure."

Well, sometimes veiling flare is welcome! A number of other visible differences depend on lens design: bokeh, micro-contrast, chromatic aberration, vignetting. For example, the newer f0.95 Noctilux has better center sharpness and much less vignetting than the f1, but surprisingly more CA.

For an overview, here's the best case one can make that lens renderings differ in rendering micro-contrast – and that there's a definite 'Leica look':

The Leica Look – ARTPHOTOACADEMY

And some folks think/feel/imgine that there's a '3D' look to images with middle-aged Zeiss lenses, because of the way the in-focus plane transitions into the out-of-focus areas.

IMO modern lens designers are trying too hard to create pore-and-nose-hair sharpness by maximizing resolution and microcontrast. This results in a tendency for the viewer of an image to notice '2D' surface texture at the expense of a more rounded '3D' impression. This is a religious belief of mine – I know of no scientific proof.

Kirk

PS re: having 'evolved' into a fairly narrow dynamic range: Me too – but I'm trying to devolve so I can work in dappled sunlight and still attain a full range of highlight/shadow detail by blending two exposures. Even when LR/ACR retrofits the highlights and shadows of such exposures into a good-looking histogram, the highlights still fail – in a way that's especially obvious with the smooth tonal gradation of Piezo prints. But try bracketing an N and an N-2 exposure; post-process identically; select the highlights from the underexposed file using Select Color in PS; move/center the highlights-only layer onto the N exposure file; and use the opacity slider to select a level that blends the gentler highlights into a full-scale file. Or insterad use luminosity masks (I have Tony Kuyper's) for narrower selection of the highest zones of the N-2 file – haven't tried this yet.

Because he documented all those darkroom manipulations with water-baths and the like, I'm pretty sure A. Adams would have welcomed this kind of flexibility in digital processing.

 

[KenLee]

It can be helpful to compare identical images where the only difference is the taking lens. Otherwise the evaluation can be 'anecdotal'.

 

[Pradeep]

:


IMO modern lens designers are trying too hard to create pore-and-nose-hair sharpness by maximizing resolution and microcontrast. This results in a tendency for the viewer of an image to notice '2D' surface texture at the expense of a more rounded '3D' impression. This is a religious belief of mine – I know of no scientific proof.

Kirk

Kirk, I think it is market forces that dictate everything these days. If you look at the work of high-end photographers in the wildlife genre, whose limited editions sell for over $10K each, they are razor sharp, with each facial hair/whisker standing out. This has led to everyone, from the lens makers to the consumers of such media to lean towards more sharpness and contrast than 'character'. Of course the fashion industry may have different needs.

Personally, I have never understood the 'draw' or 'look' of a lens. Have owned Leica glass, Phase One, Canon and Sony - many of their flagship products in fact. To me the only notable differences are in the distortion, sharpness, bokeh and contrast but I cannot tell them apart otherwise. They do not relate a story to me. But I admit I am a philistine in such things. It may be that distortion of a certain kind makes a lens more appealing, like a small mole on a woman's face can make her look more beautiful or 'interesting' at least, as opposed to 'clinically perfect' but boring.

At the end of the day it is all in the eye of the beholder, IMHO.

Here is what is popular these days, at least in my area of interest: http://www.la-photographie-galerie.com/artists/photos-7/files/page17-1005-full.jpg

 

[KenLee]

...Personally, I have never understood the 'draw' or 'look' of a lens.

Because there are so many sloppy lens comparisons online, it's hard to see any differences even when they exist.

The differences are seen in the flaws of lenses. By definition, images made by flawless lenses look the same. Of course, there are no flawless lenses.

 

[Pradeep]

Because there are so many sloppy lens comparisons online, it's hard to see any differences even when they exist.

The differences are seen in the flaws of lenses. By definition, images made by flawless lenses look the same. Of course, there are no flawless lenses.

Agree, Ken. As I said earlier, it is the distortion (flaw in your terminology) in a lens that lends it a special character that may appeal to a person. The less the flaw, the less the special appeal. Manufacturers are working hard therefore to create lenses with minimal flaws.

Don't know if that is necessarily a bad thing. I suppose it depends on what one is after. Me, I am incapable of appreciating subtle variations.