« Test Your Photo Knowledge | Main | Leica Replaces Chairman Lee »

Friday, 22 February 2008


My favorite bit of silliness is when someone says something like "the difference between Tri-X and HP5 is in the midtones" or even refer to "the Tri-X look".

Is that Tri-X shot at ISO 200 or 1600? Machine developed or small tank processed? Constant agitation, intermittent agitation, or stand development? D-76, D-76 1:1, or Diafine? Scanned from the film or looking at a print on FB? Printed at grade 2.5 or grade 4.5? Tri-X 35mm or 4x5? And so on...

Bjorn Rorslett (a well-respected user and reviewer of Nikon gear) has come to very similar conclusions with regard to the supposed differences in depth of field between APS-C sensors and "full-frame" 35mm sensors. He did an extensive series of tests of the Nikon D2x and D300 compared to the (full-frame) Nikon D3. He has repeatedly stated that the available depth of field calculators are based on algorithms that produce results that don't hold up in real life. Moreover, there are so many variables that come into play that even if you (think you) can measure the differences in DOF, they will almost never be relevant in real life. He has concluded that the difference in DOF between the two formats is less than one stop. In any case, it is certainly worth browsing through his posts and empirical tests on this subject over at nikongear.com. The results obviously are not Nikon-specific, so they should be of interest to anyone.

Best regards,

In the past I have flooded the Leica Forum with B&W images taken with FM3a and 50mm 1.4 Nikkor just to satisfy myself that this Leica thing is no more than a rich man's hoax. I've always wanted to own a Leica but couldn't afford it. Having Leica users drool over my Nikon pics and reassuring each other that only Leica glass could be responsible for the quality of these images made me feel better.

As for the rest of the discussion I agree perfectly with Mike, photography has become too much of a gear hype. Although I sometimes envy people taking multiple pictures of ordinary ducks with a Canon EF 70-200 2.8 IS L fitted on a MkIII. But then again each to his own.

Mike, you're definitely going to end on somebody's black list again. :-)

Besides, I couldn't care less about the sensor differences if the photo is good. Not about the sensor size nor about the name on the camera nor about the lens on that camera.

Your friend Michael Reichmann said somewhere on his site that you have to care about the lens and the camera because if, for instance, you didn't have a telephoto you couldn't take the photo that requires a telephoto. I'm simplifying wildly, but it boils down to that. Well, I don't think that's exactly true.

If you have a superzoom compact, it's basically the same as if you have a big black camera with a cannon of a lens on it. Yeah, DoF, blah, noise, nyah this and urk that. A good photo is a good photo.


I'm not sure what would cause you to make such a claim. I assure you that when I shoot my dSLR in monochrome mode at ISO 3200, the resulting JPEGs have a beautiful noise pattern that CLEARLY matches the grain structure of Tri-X.

By contrast, I find the B&W results from my compact digicam at ISO 400 fairly distasteful. Far too reminiscent of the grain in pushed HP5...


Well said - most measurbators have very inflated ideas of their own capabilities to sniff out offensive sensors/lenses/brands. The key word is "offensive" not inferior, because it the measurbator is a defensive stance by people who feel threatened about their equipment not being the best. Love to see you post some 4/3 v.APS (may be some v.1/1.8 for fun?) shots to show up their baseless chauvinism.

Well Mike once again you've highlighted the critical issues (should measurbator have another vowel - measurabator??).
Andrzej Wrotniak has made a similar observation (http://www.wrotniak.net/photo/oly-e/sensor-size.html). It does appear on this issue (sensor size) as with others, certain reviewers can detect different values and characteristics and as with all things some do this more thoroughly than others. After identifying any such issues the question remains how important is this, under what circumstances is it relevant and does it matter to the average user? How does the impact of this issue compare to other factors such as lens quality, exposure etc. Will I be able to see any difference on a decent size enlargement? Too few reviews go this far and provide appropriate real-world interpretation of results. Over here in the UK we have the British Journal of Photography which for significant models publishes two sets of reviews, a technical and scientific (verging on the impenetrable) report by Ander Uschold followed the next week by a 'real world' review concentrating on a more practical assessment. This works as well as any other approach but still leaves questions unanswered. It’s in the area where you and the Sean Reids of the world provide us with some insight and one that is generally ignored by the average review.

So, which lens was it?

Mike, you're right that the difference in sensor size between APS-C and 4/3rds isn't very significant. However, the difference between FF and APS-C is much more significant, especially when making fine prints. If you look at this chart (http://en.wikipedia.org/wiki/Image:SensorSizes.png), we're talking about a 50% sensor size increase from 4/3rds to APS-C but more than 160% sensor size increase from APS-C to FF. But back to the discussion at hand, is there really a large contingent of photographers that put down the 4/3rds system simply on the basis of sensor size? I always assumed the complaints were about the general lack of lenses, design of the bodies, and the inability to do super wide-angle. That and the fact that 4/3rds bodies haven't provided the significantly smaller size/weight bodies and lenses (relative to APS-C) as intended.

Hi Mike,

Writing sensible down to earth things on the internet will not work. Stop it, will you.

(Those Oly 510 articles on TOP a few days ago were really sensible too.)

There are multi-megabytes of pixel-peeping tests conducted at 100% screen magnification, but almost nowhere can you find anyone who will tell you if that "difference" they just spotted can be seen by anyone on a print. Too much data, too little useful information.

If anyone has any doubts about the impact of sensor size on the resulting image, just take a look a the image gracing Dewitt Jones column, Basic Jones, in Outdoor Photographer’s March issue which can also be seen at http://www.outdoorphotographer.com/content/2008/mar/basic-jones.shtml. It was taken with an iPhone. Anybody wanna buy a D3?

"That and the fact that 4/3rds bodies haven't provided the significantly smaller size/weight bodies and lenses (relative to APS-C) as intended."

I'm not specifically trying to defend 4/3rds here, but I recently learned that this was *not* originally intended. (Although I was one of many who assumed it was.) It turns out that all *Olympus* ever claimed was that long telephoto lenses would be lighter, smaller, and faster in the 4/3rds system...as, indeed, they are. Olympus apparently never officially claimed that the normal zooms and bodies would necessarily be smaller and lighter. People can correct me if that's wrong (with citations, I mean).


One of the fundamental tenets of psychophysics is Weber's Law, which states that the detectable difference between two magnitudes is proportional to the magnitudes themselves; e.g., humans can tell the difference between 1 oz. and 2 oz. but not between 100 lbs. and 100 lbs., 1 oz. Mike, you're a psychophysician!


The problem is not the 4/3 chip size, but Olympus has apparently been let down by outsourcing to the wrong chip maker, who's R&D is a couple of generations behind Canon, Pentax, etc.

Good point Mike. In my case, I switched from Canon (10 and 20D) to Olympus 4/3 (E-330, E-510, E-3) when I could not discern any resolution differences between the 20D and E-330. Other factors motivated the switch, color rendition, overall lens quality, live view, original and best automatic sensor cleaning, etc. In other words, it is the camera and system as a whole that makes more of a difference to me than the relatively minute differences in APS vs 4/3rds sensors - especially given the normal crop of the 2/3 aspect of the APS sensors.

Full frame sensors, on the other hand, do seem to have an advantage in low light situations, at least when they keep the number of pixels lower and allow the photoreceptors to be larger, as with the Nikon D3 and Canon 5D.

I'm not specifically trying to attack 4/3rds here (I'd have bought into it 4 years ago instead of Canon if they'd had a decent relatively affordable long telephoto/zoom) but:
"The advantages of 4/3 at a glance
* smaller and lighter lenses and camera bodies"
(I'm excluding the other 6 bullets) - Olympus E System, Frank Spath, 2003, p 18.

To make a test like this truly scientific, you would have to be comparing two sensors produced by the exact same semiconductor process, and use those sensors with exactly equivalent lenses. The greatest fallacy behind all those claims made on the respective merits of 4/3 vs. APS-C is the implicit assumption that all other parameters are equal. They are not!

To even claim that two sensors manufactured a couple years apart are the same simply because they bear the same part number and have the same resolution and size is completely ludicrous. Semiconductor processes evolve constantly.

IMHO, it is most likely that any person capable of accurately sorting prints from various sensors, if there is such a person, will actually be basing his/her judgment on many factors not even remotely tied to sensor size...

To be able to make an opinion for yourself and then claim a preference for the output of one camera vs. another is what critical judgment is all about. But to attempt to substantiate your claim based on specifications is only showing lack of self-confidence, and has no scientific value whatsoever.

So, let's go back to the Decisive Moment Digital camera. Can we now specify a smaller sensor size than was previously thought necessary? Would 2/3", like on the Nikon Coolpix 8400, be big enough? Or maybe 1 inch? Just thinking out loud here.

Yeah, but...

The reason people used large-format cameras was so that they could print as large as possible while still holding onto a quality level that knowledgeable art enthusiasts found acceptable, whatever those levels might have been. In other words, judgment was made at the extreme -- not at the average, with average photos looked at by the average man.

I remember years ago looking at a Shiela Meztger landscape in a NY gallery, and thinking, that whatever camera and film combination she'd used, she'd taken the print at least one or two sizes too large -- the grain had become unpleasant and obtrusive. If she'd had a larger neg, then that size might have been okay.

So, while I agree in general with your argument, I would suggest that **at the extreme**, knowledgeable people would see and react to differences between a 4/3 shot and one from a Canon 1DsIII.

Furthermore, that difference is baked in the cake, and the photographer, no matter what his skills, will not be able to escape it.

All reasons, IMHO, that the ass will eventually fall off the 2/3 system, and bring us back to the cry of all those old Olympus Pen-F owners: "Mine is still better." 8-)


Though I agree that it's pretty challenging to discern formats and similar perspectives in pictures, I still have yet to see a cell phone photo that compares even to a compact camera shot, strictly in terms of image integrity. Ironically, on Pbase I've seen many shots from D3s and 1Ds that might as well have been taken with a cell phone for all their artistic merit.

One of my favorite pictures I've ever taken was on a 2 megapixel compact with a 1/3.2" sensor! I can print it at 8x10 and it looks fine. Okay, some sharpening artifacts from the jpeg engine, but I'm not ashamed to lay it beside my DSLR prints.

Once again we are falling into the gear queer cry of having the biggest/fastest (fill in the blank). Perhaps at the extreme end of the spectrum there are those who can tell the difference between 4/3 or aps whatever, but I suspect most of the hoi polloi can't tell or cares about advantages of either system. Just like the high end audio stuff I used to lust after, I found that I really couldn't tell the difference between solid core or silver stranded cables. In photography it still comes down to properly framing, exposing and being there to get that great picture. Instead a lot of us end up staring at the screen trying to figure out if we have right sensor/camera. I'd get out and shot now, but I didn't bring the right camera, film, lens combination for the flat lighting outside :>)


"Full frame sensors, on the other hand, do seem to have an advantage in low light situations, at least when they keep the number of pixels lower and allow the photoreceptors to be larger, as with the Nikon D3 and Canon 5D"

Will (and Patrick),
True, but then, full-frame vs. APS-C does hold up to my ~2X-area standard--full frame is 2.3X the area of (Nikon) APS-C (a little more when compared to Canon's slightly smaller sensor size), and 3.8X the area of 4/3rds. So if 2X is the approximate threshold for differences to be visible, then FF should indeed be visibly better than either of the smaller formats--even when just looking at pictures.

Mike J.

The EXIF data on the pic above shows it was taken with a cell phone camera while on hold with your auto insurance agency.

What did I win?

Not trying to sound defensive, just trying to inform...

"I always assumed the complaints were about the general lack of lenses, design of the bodies, and the inability to do super wide-angle."

While I won't question that the design of the bodies don't appeal to everyone, I would like to point out that there are 38 lenses for the 4/3rds system, including the 7-14mm (equivalent to 14-28mm) which I would consider a a super wide-angle, wouldn't you?


It seems to me that the big issue at the moment with four thrids is that olympus seem to be having trouble sourcing high quality sensors from the likes of kodak, I don't know if its that the manufacturers aren't willing to invest in the format or what but they don't seem to be putting much effort into getting high-iso sensitivity out of the four-thirds sensors.

Image quality aside, I enjoy my 1.6x Canon viewfinders much more than the Olympus Evolt series viewfinders. Image quality be damned - if I can't effectively see, or have to strain to see, what I'm shooting (not just composition, but focus, too), the camera's usability plummets and photos don't get taken.

mmm, I've taken to stalking the photo section of Borders bookshop at lunchtimes and looking through the pictures. Most were shot with the celluloid style sensor and, in the PJ books I favour, you can see all sorts of grain artifacts etc. Love the pictures though.

One thing about the whole process is that it encourages me to print more of my stuff out and then look at it for a few days. Some things seem much better than I expected and others....

I'm sure that you're right Mike, I couldn't tell the difference between sensor sizes by and large, though I can see the difference in an 18/16 by 12 print from my old compact and the 5D.


Gee if the image were a bit larger and I knew what the f/stop was ...

I'd guess 80mm at f/16 or 50mm at f/8

Lets see. I can't tell what the car in the foreground is but isn't that a late seventies Honda Accord next to a Chevy Suburban? There is a pretty big size discrepancy between those two , so there isn't as much foreshortening as it looks like at first glance, in fact the far side of the car in the foreground is about 4/5 the size of the near side, so the more I look at it the more it looks like 50mm or maybe shorter even though it feels like 80mm at first glance. Then you have that facing north in the late morning light that flattens everything out, so I'm thinking shorter and shorter, but with a 28, your shadow would be in the picture and the ground doesn't look like a 28, so even though I voted for 85mm , now I'm thinking that it looks a lot like a 35mm at 5.6 . But wait, that car in the front is tilted up , so the camera doesn't really capture as much vertical angle of view as it looks , so now I'm going back towards 50mm

man that's a complicated picture

Are there really no more digicams with 2/3" sensors? That used to be a very popular size, and considerably larger than 1/1.7".

Well, I'm not rich enough to have a comparison between particular models of cameras. And I'm not quite sure, that in terms of no-noise producers have reached the level at least close to the best-phisically-possible one. But being the physicist it's very easy to tell how big the differences will be WHEN the cameras reach that level.
The best-phisically-possible situation is the situation, when the amount of the noise is limited only by the fluctuations of the photon flux seen by different pixels.
In such a case, when we take sensors with the identical pixel count, but the different sizes then, under the SAME lighting conditions, the sensor 2 times smaller lineary (i.e. as the 4/3 sensor in comparison to the full frame one) will give the same noise for the speed exactly TWO stops lower, because it's pixel have 4 times smaller area.
That means, that future Oly-E will get just the same results on the ISO 400 setting, as the future FF CaNicon set on ISO 1600.
Going that way, if we get the Oly-E WITH anti-shake and FF-CaNicon without one, the former can COMPENSATE the 2 stops lower "usable" ISO by the ability to shoot with 2 stops longer times.

The question is, if end when we can get the electronics of the camera so noise-free that the noise-on-the-photo level will be limited only by the number of the photons collected?

"Image quality be damned - if I can't effectively see, or have to strain to see, what I'm shooting (not just composition, but focus, too), the camera's usability plummets and photos don't get taken."

Amen, brother. Even though I'm not as sensitive to this particular issue with the Olympus E-510 as you are, I know exactly what you mean. The E-510's viewfinder is definitely "an acquired taste." I, on the other hand, am more sensitive to issues of size and weight. If a camera is big and heavy, I don't want to take it out for casual shooting. Its usability plummets and photos don't get taken. (A paid gig is another matter.)

In exchange for the size of the E-510 I'm willing to put up with the small viewfinder. Others may not, or they may have other issues, all based on personal, practical needs and preferences. But--and here's where I bring this back around to the central point of this thread--these needs and preferences should be PERSONAL and PRACTICAL, not based on theory, conjecture, second-hand information, and blind prejudice. If you don't like the E-510 or any other camera because it truly cramps your style, I have no argument with you. If you tell me you don't like it because "there's no way to get good pictures with a sensor that small," I'll have a hard time taking you seriously.

I'm guessing 50mm in the sample because I've read some of your writing about the versatility of the 50.

(tongue in cheek): If you're extremely demanding and only the highest image quality will do AND if you believe that 24x30 inches is too small of a print size then you couldn't possibly shoot anything less than the absolute best and biggest format going.

Mike, thanks for the voice of reason. Earlier today I was pondering the related issue which has come up above. ie 12 MP ff vs 12 MP aps at low iso, how would you know. I realize this has can of worms written all over it. So I apologize in advance for opening it.

"Are there really no more digicams with 2/3" sensors? That used to be a very popular size, and considerably larger than 1/1.7" "

Not as much larger as you might think...2/3" sensors were only 8.8x6.6mm, for an overall area of a whit over 58 square mm. Compare to 1/1.7", which is 43 square mm.

Mike J.

Mike: While I mainly agree with your thoughts on sensor size I can only really evaluate my personal experience. My digital experience started with a CoolPix 990 then on to the Canon D60, 1DMKII, and now the 5D. Each step I found gave me noticeably better results. However, at the same time, I was also moving up to better printers and vastly improving my Photoshop skills. I'll have to go back and revisit some of those old files someday...
Also, I am absolutely sure that one should only comment on the quality of a print (NOT the screen image) at the desired size and viewing distance. Anything else is a waste of time.

Am I glad you got that off my chest!!!

Years ago my roommate bought into the Monster cable fad, he honestly believed he could hear a difference that was made by these over priced speaker wires (due to the lower resistance don’t you know). I switched out his monster cables for regular 22 gauge speaker wire. For three weeks this went on until he noticed the change while dusting.

No; he admitted, he could not tell the difference.

With all the variables that are left to a photographer’s whim, I doubt that any could tell the difference between images created with any brand camera.


There's an old story that made the rounds a few decades ago, a young man entered a camera store with an old box camera and a shoebox full of pictures. After showing the photos to the sales clerk, he asked if they had a camera that would take better pictures. The clerk sighed and said "No, but we've got some that will make them easier."

I have been giving this quite a bit of thought recently and the point I keep coming back to is not the size but the shape, surely the best shape for a sensor is a circle, that is best reflecting the shape of the lens. Film was generally a strip and so squares and rectangles made sense, but the image deteriorated out to the corners and if the sensor was a circle then this would be constant, you have available more of the prime area for the lens. Hey, you can always crop in photo shop. (Is that a slogan?) Imagine never having to rotate your camera? Thats something that makes me think...

My favorite of these discussions is the 16-bit versus 8-bit debate. Although there are a few specific situations where 16-bit is clearly better (gradients in Photoshop, bad raw converters), I have yet to be convinced that converting to 8-bits actually does anything bad to an image. Well, beyond "damaging" the histogram.

Dear Bert,

Absolutely, positively right! Sensor characteristics differ so much over time and model that sometimes it's not a case of even comparing apples and oranges; you're not even talking about the same phyla!

It would be kind of like me saying that 6 by 7 format was superior to 35 mm and not taking into account that I've got my camera loaded with Konica 3200 while you've got Fuji NPS 160. Technically, your photos are going to blow mine away.

Even in the rare cases where things are equal, people overestimate the significance of some of the differences. Humans can perceive incredible amounts of fine detail. There's an 8 fold difference between the amount of fine detail it takes for people to think a stand-alone print looks acceptably sharp and the finest detail they can perceive. In terms of pixel counts, that's 60-fold! Is it any wonder that folks argue about how many pixels are truly enough?

But people are really lousy at perceiving incremental improvements in sharpness. Under controlled test conditions, you can just barely see a 12% improvement in resolution. In terms of pixel counts, that's 25%. Yet people obsess about whether they should trade in their 8 megapixel camera for a 10 megapixel camera.

And that's assuming all qualities are equal, which we both agree they aren't. People need to learn to look at the real images and real tests, not their notion of what the theoretical specifications are. Otherwise they're just buying shiny baubles dangled in front of them.

pax / Ctein
[[ Please excuse any word-salad. ViaVoice in training! ]]
-- Ctein's Online Gallery http://www.ctein.com
-- Digital Restorations http://photo-repair.com

Dear Bernard,

You nailed it. The Sensor Size Queens are laboring under the delusion that today's sensors work near the physical limits of the universe. That would be kind of like declaring, back in 1970, that you could never get faster, finer-grained films, because the silver halide crystals had to be so big to capture enough photons.

What's the physical limit? Well, very roughly ...

Film typically runs a couple of percent efficient. I don't think there's ever been a photographic film made that even got up to 5%. Agfa had some new sensitizing technology that would have pushed up to around 20%, but it never got used.

Scientific solid-state sensors can count almost every photon coming in and even tell you its precise energy (color). It gives "full color" photography a whole new meaning. In any case, building scientific sensors that have 70-80% collection efficiencies is no problem. But the technology is a long way from what you're getting in your cameras.

So, when digital cameras start producing results comparable to film at, oh say 30 times film's ISO, then you'll probably be closing in on the classical limits. In other words, get a digital camera that with no noise reduction has just as fine "grain" as Portra 800 at speeds of 25,000 or so, and you'll be approaching said limits.

After that there are some quantum-mechanical tricks for reducing the "irreducible" statistical noise severalfold, but I don't even want to think about when those will be commercialized. More than 10 years, less than 50.

Regardless, the future's so bright you're going to have to wear shades. [grin]

pax / Ctein
[[ Please excuse any word-salad. ViaVoice in training! ]]
-- Ctein's Online Gallery http://ctein.com
-- Digital Restorations http://photo-repair.com

Measurbators have been around for a long time. In the mid 70's I worked at a large camera store in Chicago. One joker came in wanting to buy a 80~200 f4.5 Nikkor for his pristine F2. He insisted that we open several new boxed lenses so he could look through the lens(off the camera). He was looking for a lens that was free of even the tiniest bit of dust between elements when examined closely in bright sunlight. After rejecting the first two or three samples the store owner, Mr. Altman, told the guy no more boxes were going to be opened for his inspection and if he wasn't going to buy one of the three lenses he had already checked, to please shop elsewhere. I doubt the guy ever loaded film in his camera, it was just jewelry.

It's also possible the guy was suffering from OCD (obsessive-compulsive disorder). OCD can be functional (when it helps people focus on real tasks) or dysfunctional. Eons ago I sold high-end stereo equipment for a short time, and we had a few customers who were obviously in search of a state of perfection that, shall be say, greatly exceeded what was rational. Such disorders aren't entirely voluntary, so spare a bit of compassion for sufferers--or, at least, pity.

Mike J.

John Robison,

ALTMAN'S....great store! Old school camera shop. The only one left is Central, just down Wabash from where Altman's was. Of course there is Helix and it's still pretty much the same but not quite. Helix bought all of Altman's fixtures..cabinets, what not...some GUY told me that.

My Dad bought me my first camera at Altman's when I was in 7th grade..a used Minolta SR-3. I also bought my first enlarger there a couple of years later...Omega B-22.

I used to go downtown on Saturdays with my photo buddy, Rabbit (his school nickname...black kid with some huge front teeth). Rabbit and I took night photo classes at our local art center. On Saturday, we used to catch a movie at one of the theaters... Oriental/State&Lake/Chicago/United Artist/Marina City..Get a hamburger at a coffee shop called the "The 17" on Randoph, just West of State. Both of us always had cameras around our necks. Sometimes we got in trouble for taking pictures of people. Rabbit got hit on the head with an umbrella by an old lady on the bus, she put a nice dent in his fro.

Then we'd go into Altmans and look at everything before jumping on the #1 Drexel bus for a ride home.

Downtown in the loop was ALIVE and city-natural back then. What a kick...I loved Altman's

I voted for the 28..again! I always vote for the damn 28 when given the chance.

Mike, good point about the OCD but of course in the 70's such things were not broadly known. (I'm an ADD-Packrat camera wise). On the sales floor I got a reputation of handling the oddball requests. One fellow wanted to put a M39 Leica telephoto lens head on his Spotmatic. After another salesman told him "no way" I buttonholed the customer before he went out the door and in short order had him on his way with the lens focusing just half a millimeter past infinity on his Spottie. After that the other salesmen sent all the strange equipment questions to me.

Thank god someone is putting the record straight on 4/3rds. I'm getting rather tired of being sneered at by some folks for going with Olympus.

Me, I own an e410 (the baby version of the e510 with a different battery and no image stabilization, and smaller compact body. Otherwise identical in their internals), and, apart from being a bit difficult to handle because of it's small size, has been absolutely amazing. I've been eyeing off an E-3, but thats around $3200 with the 12-60mm lense in Australian dollars. They're dollars I don't have at the moment ;).

I used to go downtown on Saturdays with my photo buddy, Rabbit (his school nickname...black kid with some huge front teeth). Rabbit and I took night photo classes at our local art center. On Saturday, we used to catch a movie at one of the theaters... Oriental/State&Lake/Chicago/United Artist/Marina City..Get a hamburger at a coffee shop called the "The 17" on Randoph, just West of State.

What are you Raymond Carver ;-)

You made my day

My solution to forum-based pixel-peeping? 12" monitors for all. Viewing at 200% is so much less fun when you can only see a clutchful of pixels at a time. Staring at a 24" screen is like standing on the brink of chaos.

Puplet (from his ever-trusty 12" Powerbook)

To me one of the areas where this psychological principle is most easily demonstrated is the world of wine. I enjoy wine and have spent some energy trying to find tasty "value" wines that I can afford to buy and drink frequently. I try to find wines that are on the "sweet spot" of the curve graphing price versus quality. Turns out there are quite a few really enjoyable wines within the immediate vicinity of 10 bucks. I also have been attending some friendly wine tastings. There is absolutely no doubt in my mind that for almost anyone except a professionally trained taster, if I take my favorite $10 wine and serve it from a Chateau Lafite bottle (particularly if the person drinking it is sitting in a high end restaurant where he/she expects to taste great things) they will rave about how great it is. If I serve it from a 2 buck Chuck bottle they will turn up their nose. Expectations have a huge effect on perception. Hence, we have blind wine tastings. I noticed the same thing when I sold high end audio years ago and you can fill in the blank for your area of interest (fly rods, golf clubs, fine art, etc.).

Secondly, it seems to me that most of the measurbators out there are mainly interested in buying/owning/collecting the "right" camera/speaker/wine/flyrod so that they can be sure they posssess the "best" one, and they don't actually produce much photography (in the case of cameras). They then spend an awful lot of bandwidth defending their purchases which is another way of defending their taste or protecting their investments. It has little to do with aesthetics or art. Reminds me of the eleven year old boys that were in my bunk when I was a summer camp counselor years ago. They argued incessantly about who owned the best shoes (Nike or Adidas).

Many of my favorite photographs were taken with 1950's era film and lenses that wouldn't pass muster on today's image quality tests. Moreover, many of my favorite photographs of the digital era are being made by folks using small sensor digicams because the "noise" produced by these sensors happens to look great in their black and white photos. The best camera is the one that helps you to express what you want to express in your photographs and that you enjoy using enough so that you want to go out and make more photographs.


Will you now admit that these conclusions also apply to audio equipment?

bokeh (I'm a little fuzzy on that)

"I assure you that when I shoot my dSLR in monochrome mode at ISO 3200, the resulting JPEGs have a beautiful noise pattern that CLEARLY matches the grain structure of Tri-X."

In an 8x10 print from a Tri-X 35mm negative you can see fine detail, like for example the pattern in a fabric, while the "grain" shows up in both the out of focus areas and smooth areas with little detail.

Digital grain tends to cover the entire image like an overlay.

There is a difference when it comes to high ISO performance. It's clearly visible at ISO 1600 and up if you're comparing the newest bodies in each format (FF, DX, 4/3rds). You'll also see a difference in DoF effects shooting wide-open with fast glass. The high ISO issue is very simply that you need to use smaller sensor sites to get competitive pixel counts with 4/3rds, and even though 4/3rds has gotten to the point where the high ISO performance is better than high ISO film, the larger sensors have also benefitted from the same improvements (Compare the D3 and E-3 at max ISO, I suspect you'll find the performance similar. But the D3's max ISO is 3 stops higher than the E-3's).

Both cases are specialist cases (I actually run into the first with my work). For most applications, you are correct.

Measurbator is one of the best terms I've heard lately. Great post, but honestly Mike, these doses of common sense doesn't help the photo industry. You know, bigger is better, more is better, we need excuses for consumering, we don't need satisfaction, got it? :-)

Steve R.,
Reminds me of when I was watching TV one time with a friend and, on some promo, the Voice inside the Box said, "What does a $400 bottle of beer taste like?" and my friend growled, "Like a bottle of beer." Past a certain point, image and status have a lot to do with consumption.

I have a counter-story too...I had a friend, a painter, who got into photography in a big way for a few years. He had somehow acquired an old Nikkormat. The camera had been dropped and the lens was frozen onto the camera; the outmost element of the lens was also cracked, right down the middle, and the shutter would only work on "B". My friend would go out at night and take the most fantastic pictures, handheld time exposures where he would guess at how long to leave the shutter open. He approached his photography like an artist--all process--and his best portfolio when it was all said and done was a magnificent set of photographs.

Then his car go broken into and his ratty old Nikkormat was stolen (thieves recognize as valuable anything shaped like a camera). He got a new FE with the insurance money, but he was never really able to do good work with it. He said it never felt right to him and the pictures just didn't look the same.

Mike J.

Two observations re sensor size and "pixel peeping"

Pixel peeping isn't new , I spent years trying different developer, film, enlarging lens, and enlarger head combinations before I settled on my standard recipe, and a lot of that was based on using a 25x micro-sight grain focuser. I guess that would be called grain peeping or something but it was common enough that it was considered normal behavior and didn't warrant a pejorative name. After all, who was it that was buying all that diafine , acufine, rodonal, tech-pan etc. if there wasn't a lot of interest in micro detail?

All things being equal a larger sensor should have higher quality, but my sony R-1 usually blows away my canon 1Ds in terms of image quality.

Regarding nostalgia for the 2/3" digital sensor (which Mike has pegged as bigger, but not THAT much bigger than 1/1.7" sensors), the largest digicam sensor is reborn in Fuji's latest bridge camera, the FinePix S100FS. The camera is due out any time, and has some interesting features, albeit in a large package. No pocket cam here.

"Measurbator is one of the best terms I've heard lately."

Not my term--I can't take credit for it.

Mike J.

"I guess that would be called grain peeping or something"

I think the term you're looking for was "grain-sniffing." You're right, it was very common.

Mike J.

Let's see... if your thoughts are worth two cents, why do people offer only a penny for your thoughts? Those cheapskates. And that's not even counting inflation.

In reference to the original intent of the four-thirds system as it relates to overall camera size, there is an enlightening essay to be found at the offical Four-Thirds site (http://www.four-thirds.org/en/index.html).

Katsuhiro Takada, the Olympus researcher in charge of the development of the system, "felt that this was the perfect opportunity to review the optimum size of SLR cameras. Indeed, Mr, Takada had long had an ideal camera size in mind— the size of the OM-1. The OM-1 has long been regarded as the original compact, lightweight SLR camera, the camera that first challenged the conventional wisdom on what an SLR should be.

'It was not too big to cause problems in daily portable use and it was easy to hold in your hand and easy to control whenever you wanted to take a picture. There was no doubt in my mind that the OM-1’s size was the ideal size.'"

I'm sure that most folks, as you point out, Mike, will not be able to tell differences in prints of fairly comparable content taken with a 4/3 or APS-C sensor.

But, it's also clear that Canon and Nikon's APS-C sensors and imaging engines markedly outperform Oly's 4/3 sensor with respect to dynamic range, RAW headroom, and noise. If those attributes are not reflected in the content of prints one may be comparing between a 4/3 sensor and an APS-C one, then the differences, as you say, may not be noticeable. But if the content DOES contain a dynamic range of tones causes the 4/3 sensor to clip highlight detail, one can tell the differences even on a low-res image on the web; witness the article by Gordon Lewis "E-510 Part 2-Dealing with Overexposure" as a perfect example. Those image problems are a direct result of Oly's 4/3 sensor lacking dynamic range and RAW editing headroom.

If attributes like dynamic range, noise, and RAW editing headroom are of lesser importance to 4/3 system owners, your points are probabaly well-taken. But if these attributes are of marked importance, as they are for me personally, a 4/3 sensor isn't going to cut it, especially when Canon and Nikon consistently keep giving us markedly better performance for these attributes, not worse.

While there is something undeniably cool about Oly and their pretty nice camera bodies and their absolutely superb glass, the fact of the matter is that Oly has increasingly painted itself into a corner with the 4/3" sensor system.

These realities will become ever more clear when Canon releases what will be a full-frame "consumer" level DSLR at some point in time in the reasonably near future....something along the lines of an $1800 5D with 10 megapixels, but with a full-frame sensor. This camera, with it's huge pixels, markedly lower noise performance, and full-size sensor WILL result in image quality that you'll clearly be able to see compared to 4/3.

Dear MWG,

Converting a file from 16 bit to 8 bit doesn't do any harm.

It's what you do after the conversion that might do harm [grin].

If you're converting just before output, it has no effect whatsoever. That's because printer drivers and image viewing programs today are, unfortunately, 8-bit. That does cause visible degradation of prints in color-managed systems, but it's not like you can do anything about it.

If you're not massaging your tones or colors very much, it doesn't matter whether you work 8 bit or 16 bit. A fairly good rule of thumb is that if you're not changing of the slope of the curve (in any of the channels) by more than 25% anywhere, you will not see the difference. In other words, if four out of five of those histogram bins are populated, the results are not likely to be visible. On the other hand, if you're changing the slope locally by 50% or more (very common if you're doing any kind of major color adjustments or if you're trying to improve shadow or highlight tonality and detail) then that will visibly degrade the results. In between? Shrug.

The one real exception to this rule is if you're doing manipulations in both RGB and Lab space. There are severe round-off errors converting between the two. These are almost entirely eliminated by working in 16-bit mode, but you only need do it temporarily. In other words, convert to 16 bit just before you do the RGB to Lab back-to RGB stuff. When you're all done and back permanently in RGB space you can convert back to 8-bit.

Unless your system's seriously choking on 16-bit-sized files, though, I'd be inclined to do everything that way. It's one of those cases of can't hurt, might help. Then you don't even have to worry about whether you've wandered past the point of visible damage in your massaging. And, somewhere down the line, there will be output devices that will be able to take advantage of the extra bit depth, and then you'll see a measurable improvement in quality.

pax / Ctein
[[ Please excuse any word-salad. ViaVoice in training! ]]
-- Ctein's Online Gallery http://ctein.com
-- Digital Restorations http://photo-repair.com

Looking for a difference beetwen half-frame sensors which has almost the same area doesn't make sense. They just are half-frame and the rest is a marketing buzz and masturbation.
There is a difference between half-frame and full frame sensors, but I think it is only valid for two groups of people.
1. Those who want really big prints.
2. Those who want to use their great lenses, and not have to buy a new ones just because they have a new camera. Unfortunatelly photo industry marketing focuses on big and slow zoom lenses and hardly noone have a good series of small wide-angle primes for their half-frame sensor. Why stop using those great Contax lenses?

Hmm - it depends what you are comparing. I've owned perhaps more Canon DSLRs than is sensible at one time or another. On the whole, I would agree that I would be hard pushed to distinguish an 8x10 print from a 5D/20D/1D3 shot at 100ISO.

At 400 ISO, I am certain I could pick the 20D.

Furthermore, for certain subjects - eg sunsets - I could tell you which was shot with a 5D or 1D3 (at any ISO) due to the better tonality of the 14bit 1D3 images. (on other subjects I may not be able to tell).

I guess in average conditions, there isn't much in it. It's when you push things a bit that you begin to see the strengths/weaknesses of a given sensor. What does change over time is the amount of pushing required to reveal the differences.



@Void: If you want primes, look at Pentax. They've got a nearly full set (14, 21, 28, 31, 2 35's, 40, 43, 50, 70, 77) with more coming (30/1.4, 55/1.4)

The comments to this entry are closed.