Besides Leica, what's the sharpest digital lens?

[tom_kat]

<p>Anyone have an opinion of what the sharpest digital lens is on the market today?</p>

[Matt Laur]

<p>At what focal length, speed, and on which format body?</p>

[mukul_dube]

<p>Probably more realistic to ask which lenses give results that satisfy their users. I am a film Leica fellow, but I would not place Leica lenses on a pedestal. The products of Zeiss and Schneider and Angenieux, to name only a few European makers, have been at least as good.</p>

[starvy]

<p>Hi Tom,<br>

I notice that you are fond of old Minolta glass. I use one of these as my main digital lens doctored to fit an A mount body. I had used newer minolta lenses as well as Sony in the same body. The new plastic Minolta prime was sharper but the older Rokkor lens produces such fine character that I prefer that. I get the impression that sharpness means a lot to you? May I ask what type of photography best suits your taste? <br>

Zeiss would get you that sharpness and more but it is the latter that would make someone buy it over Japanese brands. When people choose Leitz or Zeiss glass they are buying something extra other than the sharpness. You could say that for the avid Nikonian it is the same with that brand of glass.<br>

If you are shooting digitally or even digitally manipulating film it might be worth spending a little on the sharpening filters. I quite like the free unsharp mask in Photoshop but have a filter called focalblade that allows me a little more creativity when I have more time.</p>

[tom_kat]

<p>Thanks but I don't want to use filters.</p>

[Matt Laur]

<p>OK. But can you throw us a bone, here, and qualify your question just a bit? It's unanswerable as-is.</p>

[bill_tuthill]

Sigma 70mm macro?

 

You could do your own homework and total up MTF scores on photozone.de, although it is hard to account for zoom and varying max apertures.

[colin carron]

<p>I agree with Matt. There is no way you are going to get a sensible answer to that question without a bit of narrowing down the field. The best advice I can come up with is to read the reviews on such places as Photozone and FredMiranda etc. </p>

<p>However : lenses vary from sample to sample so a good copy of the runner-up lens could easily be better than a poor copy of the winner.</p>

<p>Also : a lot depnds on the user. A good lens used in a careful manner to maximise sharpness will often get sharper results than a superb lens used less carefully.</p>

<p>And.: the traditional way to get greater sharpness is to increase the senor / film area and move up a format or two. If your really serious about sharpness and you need big enlargements how about large format? If you are not enlarging that much the final degree of sharpness you are looking for probably doesn't matter anyway.</p>

<p> </p>

[Greg M]

<p>The Olympus 50mm f2 Macro Digital Zuiko would be one. Of course, it does you no good unless you own a body that can use it, but there you go. </p>

[derek_hofmann]

<p>Which lens is sharpest, in terms of lp/mm, in the middle of the frame, at its sharpest aperture?</p>

[bj_larsson]

<p>What format? Crop? Full Frame? Medium format? </p>

<p>How are you going to compare between them? </p>

<p>The question is to general to be useful... or answered :)</p>

[Ed_Ingold]

<p>A digital sensor has a fixed pitch, which represents an absolute limit on the resolution it can achieve. The practical limit is somewhat smaller due to various losses, approximately 85% of the theoretical limit.</p>

<p>The combined resolution of a lens and digital sensor is always less than either one measured individually*. This value is approximately the root-sum-square of the uncertainty of each component. If the lens and sensor were to have the same resolution, the composite would be about 72% of that value (1/sqrt(2)). This rises to about 94% of the sensor resolution if the lens has four times that resolution. Further increases are negligible.</p>

<p>For a sensor with a pitch of 6 microns (e.g., 12MP in an APS-C sensor), the optimum lens would have a resolution of about 400 lp/mm. This is not unreasonable for an high quality lens for 35mm cameras, such as the f/2.8 Nikon zoom lenses. You would achieve the same net quality per pixel with a 9 micron pitch (12MP, fx sensor) and a 200 lp/mm lens.</p>

<p>Most modern lenses are sharp enough that resolution alone is not the most important factor. Chromatic aberation has a significant effect on sharpness in digital capture, but is not readily detected in MTF measurements. I find CA in my Hasselblad lenses using a digital back (CFV-16) which is not visible in film images. Software to remove CA noticeably improves sharpness on an A/B comparision (e.g., using Phocus or Lightroom software).</p>

<p>The highest resolution/pixel and freedom from CA for digital photography is found in Schneider "digital" lenses used on modular cameras with medium format digital backs. They are not even especially expensive, once you have the back, of course. See various articles in Luminous-Landscape.com for details.</p>

<p>* Resolution of lenses is normally made using high resolution, high contrast media, and is limited by the composite value. This renders resolution tests found on Photozone.de and DXOMark, mounted on a DSLR, nearly meaningless. Lenses can be measured independently by the aerial method, whereby MTF scans are made using a microscope and photometer on the aerial image.</p>

[mark_pierlot]

<p>First of all, there's no such thing as a "digital" lens. Lenses are optical instruments, and thus are inherently "analogue."</p>

<p>Secondly, as far as I understand, macro prime lenses are the sharpest.</p>

<p>Thirdly, as others have said, which lens is sharpest depends entirely on your applications. For example, the sharpest fast prime I have ever used wide open is the EF 85/1.2 L II, but that lens is useless for some uses.</p>

[Ed_Ingold]

<p><em>First of all, there's no such thing as a "digital" lens. Lenses are optical instruments, and thus are inherently "analogue."</em></p>

<p>Sorry to disagree, but you misunderstand the statement. A "digital" lens is a lens optimized for digital capture. Several reliable companies advertize "digital" lenses, including Hasselblad (CFE40/4 IF) and Schneider. These particular lenses are especially sharp in the center of the field, at the expense of the corners, and have exceptionally low CA.</p>

<p>I'm somewhat suspicious of lesser marques who claim that their "digital" lenses constrain the exit rays to be parallel. That describes collimation, and the view through a collimator looks like a pinpoint of light no matter where you look.</p>

[berg_na]

<p>The response of the lens/sensor is calculated by multiplying the response of each component, i.e., system MTF= lens MTF x sensor MTF, not added in quadrature as you implied.<br>

The Nyquist frequency for a 6um pixel pitch is 83 line-pairs/mm and for a 9um pixel pitch it is 55 lp/mm.</p>

 

<blockquote>

<p>This value is approximately the root-sum-square of the uncertainty of each component. If the lens and sensor were to have the same resolution, the composite would be about 72% of that value (1/sqrt(2)). This rises to about 94% of the sensor resolution if the lens has four times that resolution. Further increases are negligible.<br>

For a sensor with a pitch of 6 microns (e.g., 12MP in an APS-C sensor), the optimum lens would have a resolution of about 400 lp/mm. This is not unreasonable for an high quality lens for 35mm cameras, such as the f/2.8 Nikon zoom lenses. You would achieve the same net quality per pixel with a 9 micron pitch (12MP, fx sensor) and a 200 lp/mm lens.</p>

</blockquote>

[jorgen_dalen]

<p>Having browed a few photozone mtf diagrams (but not all) it seems that the sharpest for full frame (in the center) is the canon 35mm 1.4L with LW/PH of 3869. The sharpest overall lens for full frame seems to be to zeiss 100mm macro.<br>

Canon 35mm 1.4L test:<br>

<a href="http://www.photozone.de/canon_eos_ff/516-canon35f14ff?start=1">http://www.photozone.de/canon_eos_ff/516-canon35f14ff?start=1</a><br>

Zeiss 100mm macro test:<br>

<a href="http://www.photozone.de/canon_eos_ff/508-zeiss100f2eosff?start=1">http://www.photozone.de/canon_eos_ff/508-zeiss100f2eosff?start=1</a></p>

<p> </p>

[jorgen_dalen]

<p>Having browed a few photozone mtf diagrams (but not all) it seems that the sharpest for full frame (in the center) is the canon 35mm 1.4L with LW/PH of 3869. The sharpest overall lens for full frame seems to be to zeiss 100mm macro.<br>

Canon 35mm 1.4L test:<br>

<a href="http://www.photozone.de/canon_eos_ff/516-canon35f14ff?start=1">http://www.photozone.de/canon_eos_ff/516-canon35f14ff?start=1</a><br>

Zeiss 100mm macro test:<br>

<a href="http://www.photozone.de/canon_eos_ff/508-zeiss100f2eosff?start=1">http://www.photozone.de/canon_eos_ff/508-zeiss100f2eosff?start=1</a></p>

<p> </p>

[john_robison3]

<p>I've always thought the sharpest lens is a tripod but perhaps that was back in the film days.</p>

[buffy_h_f.]

<p>The FA Limited series from Pentax (31mm f/1.8, 43mm f/1.9, 77mm f1.9) are quite sharp although they are not made for digital as the DA series is.</p>

<p>If you shoot RAW, all your digital photos will require at least a modest amount of sharpening no matter how sharp the lens.</p>

[anand_dhupkar]

<p>Also - the needed size of the final print will influence the sharpness -</p>

[marcsharp]

<p>A good tripod...?</p>

[Ed_Ingold]

<p><em>The response of the lens/sensor is calculated by multiplying the response of each component, i.e., system MTF= lens MTF x sensor MTF, not added in quadrature as you implied.</em><br /><em>The Nyquist frequency for a 6um pixel pitch is 83 line-pairs/mm and for a 9um pixel pitch it is 55 lp/mm.</em></p>

<p>MTF curves are a graphical representations of Laplace transforms. The transforms themselves can be multiplied (and graphed from the results), not particular values themselves. The root-sum-square is an approximation. For a more rigorous analysis, in Laplace form, see <a href="http://www.normenkoren.com">www.normenkoren.com</a>.</p>

<p>Assuming an 85% "efficiency", I agree with your values for 6 micron and 9 micron pitched sensors. I would interpret the results to mean that a lens with about 320 lp/mm resolution would extract essential all the available resolution from a 6 micron sensor, whereas you could do the same for a 9 micron sensor with 200 lp/mm. In other words, the larger the pixels, the less is demanded from the lens.</p>

<p>The Photozone.de values are determined by fitting the lens to one of several digital cameras. The value approaches the theoretical maximum (4288 for a 12.3 MP sensor), assymptotically. This means that results for various lenses will have only minor differences, regardless of their quality. If you use a camera with a different resolution (or pixel size), the results for the same lens will be different.</p>

<p>On the other hand, Photozone results for distortion and CA are very useful, and not found elsewhere in such a clear representation.</p>

<p>Let us assume, for the purpose of this discussion, that a good camera support is used, and not trivialize the argument by saying a tripod is needed. With minor variations, a human being shakes a camera about 20 MOA/sec, which limits the resolution for any lens unless the shutter speed is at least 3x the "focal length rule" value.</p>

[bill_tuthill]

Jorgen, thanks for doing Tom Kat's homework! Compared using Canon APS-C, the Canon 35/1.4 attains 2077.5 LW/PH at f/2.8, while the Sigma 70/2.8 macro attains 2071 LW/PH at f/4. (Zeiss not tested with APS-C) The Sigma 70/2.8 macro and Zeiss 100/2 macro have much better edges, though.

[images_in_light_north_west]

<p>Ziess 21mm</p>

[bill_tuthill]

Nope, compared on full-frame Canon, Zeiss 21/2 achieves only 3561 LW/PH at f/4, versus 3776 for the Zeiss 100/2 macro and 3869 for the Canon 35/1.4.