« The Henri Cartier-Bresson Show | Main | Nikon Replaces the Fast 85mm »

Thursday, 19 August 2010

Comments

Is there anyone that does such focusing tests on digital cameras and and reports the results?

cfw

This kind of stuff is really irritating because even if I determined with some confidence that I had such a problem, getting it fixed might be a nightmare. If the thing is within manufacturing tolerances, what repair shop is going to believe me that there really is a problem. How do I even get past the first line defense at the counter, who may be 2-3 degrees of separation away from the person doing the work.

You could hit 150 line pairs per milliliter? Of what beverage? :)

I seem to recall that back in the 1970's and 1980's, Norm Goldberg's camera "strip down reports" in Pop Photo would have many interesting things to say about the design, choice of materials, and contstruction qualities of the cameras he reviewed, but you'd sometimes have to read carefully and almost 'between the lines' to find the negative bits...he was the only one who would report straightforwardly that the 'shutter failed during test', or that the lens was decentered, or that the focus or framing was inaccurate and hence required servicing to be made right. NONE of the web reviewers I've seen, especially the 'big one' in London, come close to Goldberg's technical knowledge, or the willingness to act on that knowledge.

This is one area in which mirrorless system cameras will be a significant improvement over mirrored designs: Since they use the imaging sensor for autofocus, there will never be any disparity between "film" and focus sensor (be it phase sensors or the focusing screen).

I wonder if focus accuracy is any better using contrast-detect autofocus, given that the imager and the focus detector are the same object. It transfers the problem from accurate registration of the physically distinct sensor plane and focal detector to accurate detection of maximum contrast by the sensor/software combination, which I might naively expect to be an easier problem. Has anyone run tests on the accuracy of contrast-detect focus systems?

Ctein,
As you will be in Montreal for the next 10 days, try not to miss the "We Want Miles" multimedia exhibition at the MBA (Montreal Museum of Fine Arts). Dedicated to Miles Davis life, it's something not to miss.
I'm gonna see it for the second time this weekend as it will end Aug.29

Clearly insiders were aware that some things might be rotten in Denmark, but the news rarely made it to us consumers.
You should have said "Rotten in Sweden." Someone (David Vestal?) once wrote a true-life booklet titled "32 ways to jam a Hasselblad." It was said to be a popular read among photo professionals.

Bill,
You sure about that? As far as I can tell, the only source for that bit of information on the entire internet is, well, you....

Mike

'150 line pairs per milliliter'?

dude, that's some excellent Scotch!

Using Live View and the associated contrast-detection focus is something I do for lens testing; it removes a variable. (On the other hand that means my test isn't telling me how focus will be when shooting "normally".) It takes out focus-screen position and mirror position as variables in the focusing process. Of course the lens can still not mount perpendicular to the sensor for various reasons, and no doubt there are other ways to be misaligned that I'm not thinking of right now.

I've never heard of a digital sensor being physically non-flat enough to matter for focus, so maybe digital has finally killed film flatness issues, at least.

Live view manual focus or contrast-detection autofocus is the only way to go for digital. Everything else is just awful.

I never used autofocusing cameras before I switched to digital, so I don't know if they worked better for film, but I can get better and faster focusing with an old manual focus lens on my canon 5d2 by guessing and focusing using the distance scale than I can using a canon autofocus lens. I can get amazing results with live view. Using the camera the way it was designed to be used yields mush at best, even using the calibration function.

Friedrich: "This is one area in which mirrorless system cameras will be a significant improvement over mirrored designs: Since they use the imaging sensor for autofocus, there will never be any disparity between "film" and focus sensor (be it phase sensors or the focusing screen)."

True, but unfortunately there are all sorts of other problems that crop up. Test a few less recent 4/3 lenses on micro 4/3 cameras, and you can find an amazing amount of variability. Some of the algorithms still aren't there yet.

Dear Robert,

Hahaha! Voice transcription is not without its hazards. That's exactly the kind of "voice-o" that sneaks past poorfreading(sic).

-----------

Dear Nick and Friedrich,

You know, that is something I have wondered about. Seems logical they should be more accurate and reliable... But what does logic have to do with it!?

I'd love to see some real data on this.

pax /Ctein

I find myself chuckling at the anguish you and Mike have caused. First Mike got everyone worked up over bokeh....Ohmygod, my lenses aren't good enough!...Now you've got us doubting our camera bodies!

If only I could go back to those innocent days of yore when I could order gear over the phone and then take my new kit out to make photos without a worry. At least for awhile, ignorance was bliss.

The worst experience I've had with camera issues concerned my $1850 famous-name (American made) 4x5 field camera. There wasn't a detent setting on that camera that was anywhere near centered or squared. To make it usable I had to file and shave and shim to the point I doubt anyone else would want it.

I am really glad you brought this topic up. Makes one think twice about shooting wide open, especially using auto focus.

The thing is, after you run some tests and don’t like what you are seeing how do you fix it? If the camera is in factory specs and you send it in they just send it back saying it in within specs. Not your specs, but theirs.

latent_image,
Bokeh was always there. We just pointed it out.

Same with manufacturing tolerances in camera bodies. It's going to be the same whether you know about it or not.

And nobody's telling you you have to worry about it. That's up to you.

Mike

Verily, Ctein, you speak the truth. My own experience was along the lines of photographers telling me they needed to go to medium format because their pictures were to "grainy." So I'd look at their negatives and, almost without exception, tell them, "Well, they're out of focus. Get them in focus and they won't look so grainy."

I am truly beginning to wonder what is going on with my Pentax equipment, after reading your last two columns on lens and camera defects. It's an eye opener. At first I thought it was the vibration in the autofocus lens, then the tripod, then focus itself and finally, my eyes! I trusted that everything was alright and the fault was with my poor eyes. One day I forced a Nikon lens on my Pentax camera (No, it does not interface, just barely sits on the mount and will not turn and sit properly. Don't try it, you will ruin your camera!)and every shot I took with that macro lens came out as I saw it in the viewfinder- the opposite of the autofocus lens that I attempted to focus manually! So now I wonder, how will I determine anything!

JMR

My wife and I were in Montreal 2 weeks ago- would like to see your pictures. Please post them in one of your future columns.

I second Jacques Leonard's suggestion of th "We Want Miles" show at the Musée de Beaux Arts on Sherbrooke St. in Montreal. Loads of great jazz era photographs, plus video, music, artifacts, etc.

The main gripe I have about camera and lens reviews, whether in print or on the web, is that no-one ever tests more than one sample. In some cases the test camera may even have been specially selected by the manufacturer to show their product in the best possible light. Anyone with an elementary knowledge of statistics knows that this is not going to give very reliable results. Closer manufacturing tolerances and better quality control may partially alleviate the problem, but because a sample of only one is tested we get no information on whether these have improved over the years, other than what the manufacturers want to tell us.

In one of my earlier incarnations I worked at a second-hand camera dealership which used a, fairly basic, shutter-speed testing machine to check out new stock. We never saw a Pentax that could manage more than about 1/200 second and even Nikon Fs and Leicas weren't all that much better. We didn't see many Canons other than Ftbs, which were also pretty poor. Indeed, the only SLRs that consistently retained something resembling shutter speed accuracy in the upper part of their speed range were the Fujica ST801s &c.

Ctein- I was really trying to make myself forget the first time you posted about this- the one thing I could have remained blissfully ignorant of for the remainder of my life. Knowing that essentially every camera you buy does not focus accurately (manual or auto, reflex or RF, film or digital) is pretty depressing indeed. I guess the one silver lining is knowing that even the most expensive bodies are not exempt- and thank god for using wide angles stopped down.

Leica M cameras are of course manual focus and sans mirror, and it seems that modern digital bodies require more focus adjustments (which Leica service commonly provides) due in part, I'm told, to higher tolerances required with flat and thin sensor surfaces compared to thicker film bases.

My understanding is that modern Leica lenses are now built to closer tolerances to accommodate this issue, and incorporate other design features such as floating elements for close focusing, particularly some new fast lenses.

Others more expert than I can likely explain this in more technical-speak. I can say that I've relied on Leica to calibrate my camera/lenses for focus accuracy more frequently in the digital era. As a practical matter, however, most people would never notice in my final prints, which are not huge by modern standards.

Bill,
You sure about that? As far as I can tell, the only source for that bit of information on the entire internet is, well, you....

Mike

Afraid that I can't document it. Is Marty Forscher still alive in Florida? Or how about Ken Hansen? One of them should remember it. Or maybe the guys at Lens and Repro? Anyhow, someone from photographys "dark age." How about David Vestal himself? Didn't he used to do a column in one of your magazines?

I guess I should just stop taking pictures altogether until I can get the most lp/m.

In the past one way to tell if a film camera repair person was someone willing and able to do a "proper" job was if he insisted that one leave not merely the camera but also the prime lens that was the one most used with the camera so that best focus with the camera could be obtained where that was much more than simply focus within factory tolerances.

However both slrs and rangefinders given their mechanisms can and did easily shift focus over time, no matter how well once adjusted. The cameras that best retained proper focus over time were Rollei TLRs with Planar or Xenotar. Tessars had too much built in focus shift to get a best focus that would cover most stops. Also the TLRs had very straight film paths while medium format magazines did not.

Contrast-detect auto-focus or focusing on digital bodies in general is not without its problems, even if the sensor is flat. The problem is that sensor assembly is much thicker than film. It has a micro-lens layer, an AA filter, and a RGB filter over the actual sensor. So which one of these is the actual plane of focus?

So I guess the real message of the article is to stop worrying and learn to love the imperfections. :)

Yes, the "tolerances" are a problem. When there were problems with E-3's viewfinder (some copies had a slightly tilted view), I read about people getting their cameras back untouched as being "within tolerances".

BTW, the softness is what apparently makes a part of the "film look".

A system-block-diagram level explanation of how this stuff works, backed up with what typically goes wrong and/or what the technology limits are, would be a good article.

Well, I'd like it, but then I enjoyed a tour of the 787 electrical power system test and development lab a few weeks ago (professional perk) so I s'pose I'm exceptional... And the "artists" need to know their tools.

The wonder of having a trained, up-to-date expert to hand is that they know what current practice is and have perspective on what's important. Wikipedia only goes so far with that.

Y

Canonet GIII QL 17!

That was the first camera I was allowed to use in my early teens. But it never crossed my mind to start measuring its characteristics. I enjoyed it very much.

When I was allowed to get my first own camera, I just had to have an f/1.7 lens. Thus, Minolta XG-1 with the 50 mm f/1.7. I still have it.

"Bokeh was always there. We just pointed it out..." Exactly. I'm not saying bokeh and body faults aren't real. I'm saying ignorance was bliss...sort of. The more I know about technical issues, the more time, energy, and money I spend addressing them. It takes a toll on the creative side. Some of my best work was done when I was clueless, but there is no going back, nor would I want to.

Autofocus has been driving me crazy since I have started testing it. My D90 seems to have a tendency to focus too near. On the other hand, live view AF seems to be perfect and show what the lenses (which I at first thought I was testing) are capable of.

My problem is that I do not know how much focus accuracy I should expect. Is the AF on my camera off and could be adjusted?

Another focus problem: On my D40 the focus indicator seems to be accurate, the focussing screen not. And that is a pity, since I bought a Katz Eye screen for it. (The problem is the same with the original screen, just harder to pinpoint.) Again, is that somehow to be expected for this low end camera? Could it be fixed?

For those in pursuit of technical perfection (or as close to it as they can get) it is indeed distressing to discover further evidence of the shortcomings of their equipment. They become fixated on all the things their equipment "should" do but can't, at least not up to the standard of perfection.

Photographers who are primarily interested in the pursuit of memorable photographs tend to take a more relaxed approach. In the absence of egregious focusing or exposure errors, they worry more about subject matter, composition, lighting, and framing.
Neither approach is intrinsically more valid than the other, but it should be obvious that technical perfection alone won't result in excellent photographs.

Given how much work testing is, and how expensive top-grade cameras and lenses are when obtained through normal retail channels, and how non-remunerative magazine publishing is, I don't imagine we'll ever get many reviews with a decent sample size.

But having most review samples provided by the manufacturer certainly does eliminate any chance of learning about manufacturing tolerances and QA!

So I read most test reports as giving me an indication of the best I can possibly hope for from a piece of equipment. Plus useful commentary about UI and so forth.

I remember the "strip down" reports in Popular Photography with great fondness. Even back then the weakness of tests based on a single manufacturer-provided sample were obvious to me. Expert opinion on construction and such could, in my opinion, fill in some of the holes that left (of course it's subjective to some extent, we only had one expert's opinion).

Not that I find them a useful guide for photographic gear (I'm not an ordinary consumer of photo gear!), but one of the many reasons I admire Consumer Reports is that they take the trouble to get and test multiple samples, and obtain them through normal retail channels (at least for most reports, back when I read them regularly). (On the other hand, I really don't think the Miranda Sensorex was the best camera around in late 1969, though it wasn't at all a BAD camera.)

Dear Terry,

Yeah, I've run into that problem. My first GA645 was spot-on in both near (active) and far (passive) autofocus. I lost that one and bought a replacement-- it back-focussed by about 6 inches in near distance. The passive system was fine. Fuji tuned it up for me three times; each time it came back to me with the near-focus system back-focusing. I finally gave up.

~~~~~~~~~~~~

Dear Fred,

Wrote two column after my Montreal trip last year-- Google my previous columns for URLs.

~~~~~~~~~~~~

Dear Iain,

There have always been rumors among readers that manufacturers cherry-pick samples for review. Could happen... in theory. In practice, the number of defects and misadjustments I've seen in the review equipment I get convinces me that this happens rarely, if ever. There's just too high a percentage of crap, from every company.

I'm sure some reader can come up with a case of established cherry-picking, but I'm confident it's not even close to common, having tested hundreds of products from a dozen major manufacturers.

As for testing multiple samples, anyone with "an elementary knowledge of statistics" also knows that for a complex system, you have to test a LARGE number of samples before randomness improves over any one sample. For simple single-variable systems, "noise" goes down as the square root of the number samples-- to hit even the 75% confidence level would require testing ten of anything. For complex, interrelated variables, it goes down much more slowly.

In a few cases, I have statistical data. Back in the 80's I tested well over 100 top-tier enlarging lenses. Also, testing many, many products of a class over the years give me a trend sense.

Beyond that, what you want is simply impractical.

~~~~~~~~~~~~~~

Dear Jeff,

As I've mentioned in previous comments and columns, the factory-fresh M6 and lens kit I tested exhibited very substantial focus errors. Leica rangefinders are *precise*; they are not necessarily *accurate*!

pax / Ctein

I shutter to think how many truly great photos were taken on equipment that was truly inadequate for the job. And how many pictures of cats were taken with perfectly adjusted cameras.

Iain,

diglloyd.com does test mulitple samples, at least when the first one shows a defect. He is, however, only testing high end gear, so it's of academic interest to me.

I'm surprised you didn't mention the vacuum system Contax installed in the RTS III to make sure the film was flat. A further indication that most cameras didn't achieve that standard.

"I've never heard of a digital sensor being physically non-flat enough to matter for focus, so maybe digital has finally killed film flatness issues, at least."

David - That would be correct. The tolerances for flatness in electronic chips are very very narrow. At each phase of layering a semiconductor, they are polished (planarized) to withing a few nanometers of being truly flat.

Of course, that still doesn't fix everything if the sensor isn't perfectly aligned with the body and lens. Or if the filters, microlenses, etc. on top of the sensor are off slightly.

"Contrast-detect auto-focus or focusing on digital bodies in general is not without its problems, even if the sensor is flat. The problem is that sensor assembly is much thicker than film. It has a micro-lens layer, an AA filter, and a RGB filter over the actual sensor. So which one of these is the actual plane of focus?"

It doesn't make any difference what the actual plane of focus is, since you are seeing the results of the whole system. It is a closed loop. The same with whether the sensor is perfectly aligned or rotated etc. Unless you are doing copy photographs of flat work with the lens wide open, it doesn't matter. What you see is truly what you get. I'm using a fun old Century Precision Optics Tele-Athenar which has a field of focus that is so curved that the center of the image can be focused on infinity and the edge is focused at 10 feet or so. It works great in live view, just get what you want in focus in the "plane" ( parabola is more like it) of focus.

Will in-body image-stablization system cause focus ploblem over time because it invovles the frequent movement of sensor?

off and on over the years I have wondered about this sort of thing - the calibration, I mean. The ability on my Nikons to "fine tune" the focus for individual lenses and this article has brought it to my attention again. My results with manual and AF are variable enough (across lenses and apertures) that I suspect some sort of adjustment would be helpful. But I have no idea at all where to go to have this done. There is no way I am going to touch a single screw on the body of my D300 to give it a DIY try! :-)

Who does this sort of work?

Disassembling and adjusting aside, how do you actually check the focus? I've spent a fair amount of money having cameras and lenses straightened out recently. I've not been 100% happy with the results, and I'm thinking it's time I learned how to do this myself. I have an ME Super I bought off eBay as a second body but it's very misaligned, too much so for me to want to use it (and also too much to be able to sell it on with a clear conscience!) so that can be my learning one. I also have a funky old rangefinder I already spent £75 getting aligned, and it's not much better for it, that can be another learner :)

Dear Mark,

I really cannot speak to that, because it is not of overwhelming importance to me.

But, if it is to you, then, yes, you should stop making photographs immediately. Future generations will thank you!

pax /Ctein

Dear Jacques et.al.,

As it happened, Bayla and I forewent the Miles exhibit and (at the exhortation of Jon Singer) went across the street to see the Studio Glass exhibit. Wow.

I repeat. Wow.

It's in the back of the Decorative Arts collection, both part of the free exhibit areas. Studio Glass is up through January.

Singer was right-- it's a must.

There are also some dynamite pieces in the DA wing. At least four or five I would sell my soul for.

pax / Ctein

Ctein,
Thanks for the update. Great to know you went to see the Miles exhibit (I'm a true Miles fan) and I also appreciate the info about the Studio Glass exhibit. I'll be back at the MBA to see it.

The comments to this entry are closed.