digital backs outperforming MF lenses?

[kai_wiechmann]

Hi,

I am an advertising photographer and rented P30 and P45 backs, used them on Hasselblad H and

Mamiya RZ. I am in between purchasing either a P45 or P30. The 45 appears too sluggish on some

shoots on location with a 15" powerbook but a similar setup will be unavoidable at a certain percentage

of my future work. I never made a 'back to back' comparison with the same subject, the nature of my

shoots never really allows for that. But I keep hearing claims about MF lenses not being up to the task

(in which case my nod could go towards the P30). Is there any truth to that? Real life truth?

[warren_kato]

The real life truth is that there is not a yes/no answer. Remember back in the film days when you had a choice between Kodachrome and Ektachrome or Fujichrome, you picked the film trading resolution for grain and noise. If you really wanted detail, you picked Kodachrome 25, Pan X or Pan F, all fine grained films. It's the same with sensors, given a fixed sharpness for a particular lens, the more pixels, the better. The only exception is that if the lens is terrible, it doesn't matter how many pixels you have. But with H and RZ lens, it does.

[Ed_Ingold]

You can find several in-depth articles and demonstrations on this subject at http://www.luminous-landscape.com. Hasselblad has recognized this fact and introduced the CFE40/4 IF (q.v.), with special attention to center resolution and contrast.

 

If you're not using a tripod, it probably doesn't matter which lens you use.

[dave_redmann]

Do the math: a 39 MP back with a 36x48mm sensor means that there are about 150 pixels per mm. You could take a simplistic approach and say that 150 pixels per mm = 75 lp/mm, and then ask whether your lenses can deliver that. (General answer: some tests of 50-year-old Rollei TLR's report well over 100 lp/mm, at least in the center and at certain apertures; and some modern tests of current MF lenses report under 40 lp/mm, at least at the edges at certain apertures.)<P>

 

Of course, in reality it is a bit more complicated than this. Lens resolution is more relative than this, with reported resolutions often based on 50% MTF response. This is a good standard, but usually a lens can <I>to some extent</I> resolve detail finer than the 50% MTF limit. Also, the need to de-moscaic the Bayer sensors on typical backs means that their color resolution is well below what their pixel counts sugguest. And don't forget the Nyquist theorem: if you want to resolve two lines, you need a little over two pixels (maybe 2.1 or 2.2). Notice that the typical 35mm-form DSLR has linear resolution of only about 75-80% of what the pixel dimensions suggest. So there is no single answer that is easy, precise, and generally-applicable.<P>

 

Somebody here (or maybe somewhere like Reichmann's site?) has to have compared, e.g., 22 MP and 39 MP digital backs. In typical use, I would expect the differences to be subtle but often real and measurable. But I can't give you a first-hand answer.

[kai_wiechmann]

Thanks, everybody.

 

Warren, as for what kind of film I choose, it's almost always been 6x7 Kodak NC 160,

exposed at 100 ISO, lately a drum scan and along with that a photographic match print for

the client. This is pretty much what I compare my experience with digital backs to. The

safety of 'the bigger the better' and the fact that I generally like a larger format for it's

'look', makes me lean towards the P45.

 

Edward, it's funny you mention the CFE40. It's a 'V' lens, right? I posted in a separate post

asking about future 'H' lenses and Hasselblads future compatibility and I am missing a

similar lens for the 'H'.

I also posted on Luminous Landscape.

 

Dave, thanks for your mathematical input. It makes complete sense. But as you say, reality

is more complicated hence I am trying to trust my eyes. Just that I rarely have two backs

on the same shoot... But this is as close as I can get: I am right now looking at two files

from people shoots, unfortunately from two different ones, with different cameras/lenses

(P45/RZ, 50mm+P30/H2, zoom somewhere 65-70mm), but very similar lighting. Both are

processed the same way (no sharpening, 16bit HighRes): Both are similarly crisp (if that

term applies to an unsharpened file), can't really judge sharpness differences since I don't

know how exact I focused. But I shot both from a tripod indoor with Profoto flash. What's

apparent (at the same magnification) is that the P45 looks a tad less 'aggressive' with

softer color gradients, the skin a tad more real with that 'transparence' I know from film.

No huge difference, but enough for someone walking into my studio to see it immediately

at 200% (just happened) on the monitor. Thought I share this, even if insufficient as a

technically representative reference. Thanks again!

[doc_w]

"But I keep hearing claims about MF lenses not being up to the task (in which case my nod could go towards the P30). Is there any truth to that? Real life truth?"

 

Sorry if this is a stupid question, but I don't quite understand. If the lens is not "up to the job," how is the back going to make any difference? What am I missing here?

[Ed_Ingold]

Don said, "What am I missing here?"

 

A lens will perform as well with a digital back as with film, only you will see the imperfections in the lens better, particularly chromatic aberation. A 39MP digital back rivals 4x5 inch film in sharpness and detail, with the right glass.

[Ed_Ingold]

Kai,

 

Hasselblad does not offer an "H" lens comparable to the CFE40 IF, but you can get an adapter to use CF lenses on an H camera. Reichmann (Luminous-Landscape) uses a digital back on a modified view camera with lenses optimized for the smaller, digital image circle. These lenses are not retrofocus, with the inherent compromises of that design.

[tim_bradshaw1]

<blockquote>It's the same with sensors, given a fixed sharpness for a particular lens, the more pixels, the better.

</blockquote>

<p>

Well, perhaps not. I went to an interesting talk the other day (I'm working at a maker of sensors, though not ones used in MF digital backs), and it also depends a lot on how good your pixels are, and that in turn depends (among many other things, but this one is interesting) on how big the `full well' capacity of the pixel is, which is (something like) the number of electrons that can be displaced in it. Smaller full well capacities mean greater shot noise, and there is just nothing you can do about that. And, of course, small full wells correspond to small pixels, or to more pixels in a given area.</p>

<p>So, in other words, you can't just keep ramping up the number of pixels, you have to increase the sensor area too, or you're just looking at noise. I haven't done the maths (I'm not sure I am up to it any more), but I suspect that MF sensors are not near this limit, but P&S ones and perhaps recent phone ones are, so from now on in it's just hype.</p>

[q.g._de_bakker]

You see the imperfections of the lens better, because they clash with the imperfections of the digital sensor.<br>The colour fringes caused by chromatic aberration, for instance, form a spatially separated colour image projected onto a sensor in which colours are spatially separated. With luck, you will not see anything of it, but how often will luck come to the rescue?<br><br>High pixel resolution sensors (i.e. sensors with densely packed small pixels) help avoid aliassing. The higher the spatial pixel resolution (sensor element per mm), the better use a sensor can make of highly resolving lenses. Else, resolution limiting devices (anti-alias filters - in effect soft focus filters) will be used to limit (!) the resolution in the image the lens projects.<br>As was said, high density sensors cause noise, and instead large wells are preferred. That means regular MF lenses are too good (!) for digital capture by quite a bit.<br>However, not many (if any) digital backs make use of an anti-alias filter, so they (lens and sensor) are a good match, more or less. MF lenses certainly are "up to the task".

[john_gleason1]

In addition to other points already raised, presumably you're talking about MF lenses that were designed for film. There's an argument that, depending on the particular lens, performance might then be worse with a digital sensor than with film because of lens coating. Mike Johnston was discussing that just today, albeit in terms of 35mm lenses. But there may be some relevance to MF lenses as well.

 

 

http://theonlinephotographer.blogspot.com/

[q.g._de_bakker]

So yet another bit of ill-informed internet mythology surrounding digital photography has surfaced...<br><br>The idea that lens coatings care from what side light comes is ridiculous.<br>Flare is caused by light bouncing between lens lens elements (and inside lens elements). At every glass-glass or glass-air boundary, part of the light that is on its way to sensor or film is bounced back (!). That light is reflected back towards the film again, and when meeting another boundary is reflected [... Etc, etc, etc.].<br><br>The fact that digital sensors, and the piecces of glass they put in front of these reflect more than film is annoying. But lens coatings do not need to be different in any way.

[lightminer]

QG - I agree with your specifics 100% but perhaps am not sure about the conclusion. Lets take a lense with 5 elements. Lets assume it is coated. Now, what I thought was the difference between the more modern 'digital' lenses and previous ones is that the 5th element, the one facing the sensor/film typically was not coated as for film this makes litte/no difference to film, while with digital coating that last inner side of glass does help. The film was less sensitive to the strays from diffraction at that level as the intensity is so low, but with digital it effects the sensor a little bit. Or - maybe does the sensor itself bounce more light back at that interface while film wouldn't? That could be.

 

Anyone know if I'm on the right track here? I've actually personally done this scientifically in building solar cells (we have multiple layers in solar cells and anti-reflection coatings help - and some people are even working on purposeful reflections as well to keep light in the system and not bouncing off of the first junction and back up and out!), but don't know the details on that last layer in the film/digital issues.

[lightminer]

I think the answer to the main point of the thread is found here:

 

http://www.schneiderkreuznach.com/knowhow/digfoto_e.htm

 

Go down to "The highest line pair number Rn, which a sensor with a pixel dimension p can transmit is equal to:" and there is a formula there.

 

It continues: "This is 33 Lp/mm (for pixel size 0.015 mm) and 40 Lp/mm (for pixel size 0.012".

 

Various lenses can get up to 120 lines/mm - I've never been sure if lp/mm is exactly the same as lines/mm (might one be 2x the other? If you count 'line pairs' instead of lines there would be half as many, but probably I'm being too specific). Assuming they are the same, various medium format lenses have different ratings and some go as high as 120 (Mamiya 7II), and most of the Mamiya stuff I have is in the > 60 and > 70 range in its best f-stop.

 

If the sensor can get to 0.006 mm then you would need 80 lp/mm, and that would start to outdo all but the best lenses.

 

So, until digital sensors get to 0.006 mm we are fine!

[q.g._de_bakker]

Light,<br><br><i>"the 5th element, the one facing the sensor/film typically was not coated as for film this makes litte/no difference to film"</i><br><br>You must try to understand that this is, quite simply, not true, nonsensical, evidence that you do not understand how reflections 'work', and why lenses are coated.<br><br>Again: it does not matter which of the elements would not be coated, the effect will be the same.<br>Read what was already said, and try - not just to "agree" - but understand what it says.<br><br>So yes: somebody does <i>"know if [you are] on the right track here"</i>.<br>You are not. Far from it.

[q.g._de_bakker]

Light,<br><br>The "p" in "lpmm" does indeed mean a pair: two lines.<br>Lenses do go up to 120 l<b>Pmm. Line pairs, not lines. And even beyond.

[q.g._de_bakker]

</b><br>My apologies for the open tag! Only the "P" was to be bold.

[lightminer]

Remember that if you don't have an intense light source in the picture (i.e., the sun) then this probably doesn't matter toooo much.

 

 

http://www.outdoorphotographer.com/content/2005/mar/gb_lenses.shtml

 

"Another difference between film and digital is the relatively mirror-like surface of image sensors. In some cases, the sensors? smooth face can reflect light back through the rear lens elements, where the beams bounce around until re-exiting the rear glass and striking the imager as flare or a ghost image. The Canon series of ?/2.8-speed, L-series zooms, such as its EF 70-200mm ?/2.8L IS USM, have optimized lens element shapes and anti-reflection coatings to minimize or eliminate these ghosts. Unlike the special small-format digital lenses, this lens series is a part of the regular 35mm lineup and provides outstanding image quality on both film and digital cameras.

 

The Tamron Di and Di-II-series lenses, such as its SP AF17-35mm ?/2.8-4 Di LD Aspherical (IF), feature improved multi-coating and interior light baffling to combat off-the-sensor reflections. The lenses offer improved resolution, contrast and freedom from flare with film SLRs as well. This also is true of Sigma?s new Super Multi Layer Coatings now used on allSigma lenses.

 

Canon and Nikon also have taken care to subtly curve the protective front filters on their super-telephoto lenses. The filters? new meniscus shape disperses the reflections from the image sensor that would otherwise bounce off the rear of the filter and straight back onto the sensor?s face. The curved filters now are standard on all Canon fast IS super-telephotos, like its EF 500mm ?/4L IS USM; Nikon has introduced a curved filter on its 300mm ?/2.8G ED-IF AF-S VR Nikkor"

[lightminer]

QG Bakker - I don't know what is in your past that makes you talk to people the way you do, but you should spend some time meditating on it. But back to the topic at hand:

 

I worked in a lab at Stanford for 2 years on anti-reflection coatings in solar cells and I know the calculus of multi-layer transmission systems, so..... Not sure what you mean.

 

Also, logically, saying that the last layer is not coated has no direct logical correlation at all to understanding such systems, (the first is a historical fact, and the second is a set of concepts in optical physics) they are quite independent, so your statement is implicitely false without any defense required.

 

Tell you what - I'm interested in 'discussions' not arguments. I leave my links as they are and you can go argue with Schneider and all the other lense companies who have changed their designs to minimze final reflections from the sensor.

 

Enjoy your inflammatory arguing - its not for me!

[lightminer]

Bakker - you did the same thing on another thread I started

 

http://www.photo.net/bboard/q-and-a-fetch-msg?msg_id=00LUZ8

 

where you got into an inflammatory argument with someone else, and you were wrong there as well! Dude, seriously, put the photo stuff aside and think about life for a bit. Relax. Meditate. Socrates said an unexamined life is not worth living - seriously, I'm not trying to put you down or be angry at you or whatever, but you seriously have to think about how you interact with other people. Do you take landscapes? Go enjoy the wilderness and find peace. Okay, enough said. Too much probably, but there it is.

[q.g._de_bakker]

Light,<br><br>It appears it is you who has an attitude problem. Just because you do not like the answer.<br>Now do what Socrates asked everyone he met, and go dig deep inside your inner self to look for the answer to why you are behave like an idiot.<br><br>You agree, though obviously do not understand (amazing, for someone claiming to be scientifically involved with this subject matter)?<br>Than either explain why you think it is wrong, or take your own advice and disappear into the wilderness.

[kelly_flanigan1]

The curvature of the rear elements last surface; the one that faces the sensor *sometimes*is tweaked when a version is made for digital. This is to help reduce IR hotspots. Also the rear elements coating is sometimes to blame for added reflections too; and thus considered along with a lens design tweak. Resolution for somebody in the optics field has always been in line pairs; whether mentioned or not. Only slimball marketing chaps goose data; thus one might say one sold 48 shoes today as a shoe salesman; and just be 24 pairs if one has no real worth or honesty. I real engineering folks do goose numbers to look good; its unprofessional; abit slimball and sleazy. Some lenses were even coated before ww2; the old Kodak Bantum Special; the Kodak Ektra lens lines. If one places many different LTM or M lenses on the Epson Rd-1 digital or Leica M8 ; some old ones might have a tad of reflections with their rear elements; some might not. The comments about the last surface not beaing coated and others are seems bizzare; all the coated lenses I own from the 1940s on have coated rear elements.

[q.g._de_bakker]

Indeed. (Except that line pairs per mm are an excepted, clear, and commonly used unit. There is a reason for that too. All above board).<br>And it makes sense. Light is passing through the last element too. That means it can be reflected off that element too. That reflected light will be passing through all the other elements, partly reflecting (and again, and again) at every interface it meets.<br>And that does not just hold for light passing the second but last (the front surface of the last element), but the last interface (the reaqr surface of the last element) too.<br><br>If you are going to coat a lens, you are going to coat all (!) elements on all (sides).<br>And indeed they did: the assertion that they did not is historically incorrect (!), and the added assertion that there was a reason why they didn't need to is a show of misunderstanding and/or ignorance. (Neither are, Light, something that is made any better by having worked at a Stanford lab. But i do indeed like your <i>"I worked in a lab at Stanford [...], <b>so.....</b> Not sure what you mean."</i> )<br><br>The higher reflective properties of silicon sensors with glass filters in front of them may indeed be an added problem.<br>But any suggestion that there is help to be found in lens coatings is a red herring. Nonsense. Nothing else.

[lightminer]

QB - ideas already explained above. Links are everywhere where lense manu's are explaining how they are changing things, here is another from Canon.

<p>

Kelly - could be that the 'change' is they are making it thicker as the intensity of reflection has increased.

<p>

http://www.canon.com/camera-museum/tech/report/200308/report.html#t9

<P>

 

Improving Compatibility of EF lenses to Digital Photography

<P>

Measures against flare and ghost particular to digital photography

The image sensors in digital cameras have higher specular surface reflection than film. Due to this characteristic, when there is intense point light source within the field, the light reflects off the surface of image sensor, causing flare or ghost. To solve this issue, Canon has adopted a new approach to the optical design of EF lenses.

<P>

Adoption of meniscus lenses

<P>

Most super-telephoto lenses have a protective glass at the front. When this glass has a flat surface, the light that reflects off the image sensor reflects back from the protective glass, occasionally resulting in spot-shaped ghost.

To eliminate this phenomenon, all of Canon's large-aperture IS super-telephoto lenses adopt a meniscus lens shaped protective glass.

<P>

The meniscus lens is spherical and has curvature with same direction on both sides. By employing the meniscus lens as protective glass, the reflected light from the image sensor forms an image in front of the image sensor, and then diffuses. The majority of the diffused light deviates from the image sensor, thus preventing ghost (see Fig. 4).

 

 

<P>

Fig. 4 Flare/Ghost Suppression by Meniscus Lens

<P>

? Optimum lens shape and coating

<P><b>Even when using lenses without protective glass, flare and ghost may occur resulting from complex reflection of the light which reflects off the image sensor into the lens. To prevent this effect particular to digital photography, Canon has optimized the shape and coating of the lenses</b>, including the EF17-40mm f/4L USM and the EF24-70mm f/2.8L USM.

<P>

Specifically, each lens element adopts a different design, in order to reduce the amount of repeated reflection inside the lens. <b>Furthermore, high-transmission multi-layer optical coatings are applied to lens surfaces which are susceptible to reflection. This allows any light reflected off the image sensor to escape out of the front of the lens, thus reducing flare and ghost</b> (see Fig. 5).

<P>

In addition, appropriate balance between multi-layer and single-layer coatings contributes to the excellent color balance.

[q.g._de_bakker]

Light,<br><br>I'm sory, but any suggestion that there are things to be solved by the lens' coating is, and remains, nonsensical.<br>Lenses already are coated. These coatings already do what they are supposed to do.<br><br><i>"Making it thicker as the intensity of reflection has increased"</i>???<br>You are sure you are a scientist?