« Underwater! | Main | Canon 40D Price Drop »

Tuesday, 10 June 2008


Nicely done, Mike. I think you have convinced me to upgrade to PSE6. Thanks for the push.

Fun seeing others trying the possibilities that Autopano, Panotools, Photomerge and their likes provide. I've started trying this more often in the last month or so as a way to increase the flexibility of my primes. My standard 28mm (short normal on my 7D) can take in a lot if I flip into portrait orientation and shoot 3-4 frames. Even more so my 14mm which has become a lot of fun for social shots in cramped cafés when the subjects know what's going on..

I've admired both Koudelka and a lot of lesser known XPan-shooters in the past, now it's my turn :-)

In a sensible world, the widespread availability of convenient and efficient stitching software should produce a pause in the 'megapixel race'. Canon's Eos-1Ds mk III is already limited more by lens resolution than by its 21 megapixels. The modest resolution improvement over its 16 megapixel predecessor is barely visible in huge enlargements of photos taken under ideal conditions. It's approaching physical resolution limits imposed by optics and diffraction. The obvious improvement in noise reduction at ISO 800 with the newer camera is a lot more useful in the real world. Here's hoping the big camera makers declare victory at this megapixel level and devote more effort to high ISO performance, ergonomics, higher resolution articulated LCD's and (hopefully) a black & white version lacking the Bayer filter array.

After all of the countless times Mike J. has praised good old-fashioned, "unmanipulated," one-shot, timeless photographs, it is nothing short of amazing to see him not only singing the praises of composites but actually *doing them himself* and bragging about them! I can only imagine what's next....

you should try autostitch too... depending on the type of shot, sometimes the results are great.
one way to get around a blog's limitations is to upload on an account on something like imageshack or flickr (it could be on private there) and link to the picture url on the blog with html.
that's what i do when i want to adjust the size as i like on the blog and still have it click-able to full size.

Dear Geoff,

I entirely agree with your overall points, but one sentence gave me pause:

"It's approaching physical resolution limits imposed by optics and diffraction."

Not really. Addressing this common misconception in detail would consume a whole column, and I'm rather backed up on topics, so I'll just be brief(er) [ loquacious grin ].

A 21 megapixel full-frame sensor has a pixel pitch of about 150 pixels per millimeter. That corresponds to a lens operating diffraction-limited at f/10. That is not at all a difficult standard for a high quality 35mm lens. The really excellent ones show near-diffraction-fraction-limited resolution at f/8 and some even at f/5.6. We know that for a fact from optical bench and in-camera tests back in the film days. Optical design principles and quality didn't take a sudden nosedive when digital came in. Those good lenses are still out there.

Somewhere around 80 megapixels you really will run into a wall where no lens you can get will be able to make use of more megapixels. But we're a long way from that point.

Sure, many lenses can't exploit ultra-high resolution cameras, and none can fully do so at small apertures. So how is that changed from the film days? Not one bit. If you wanted to take full advantage of super-sharp, super-fine-grained films, you used only the best lenses and you tried to use them near optimum aperture. Same holds for digital.

Meet the new boss, same as the old boss.

~ pax \ Ctein
[ please excuse any word salad. MacSpeech in training! ]
-- Ctein's online Gallery http://ctein.com
-- Digital restorations http://photo-repair.com

I had already had Elements 6 when Ctein's post showed up. I had not tried the panorama feature but I sure did after reading what he had to say. It really worked great for me. I have a lot to learn about making the images, composing and all that. If I knew how to post my first attempt at it I would, but I don't, so I can't. I don't know how to put my blog site on here and I don't know if Mike would allow that anyway. I shot it with my Pentax K100 D on a tripod in the vertical position everything on manual. Worked good, give it a go. E

I like the pano shot because not only does it demonstrate Mike's point--32MB camera on the cheap, lurking in your camera bag right now!--but the composed picture is actually quite nice! Not to mention the horizon is straight--or is that PSE6, too?

Forget about the bacteria in the water. The risk is that if the air intake for the car's engine is low enough, water might be sucked into, and hydrolock, the engine. If that should happen, the car will stop, right in the middle of the water, as the engine will be ruined.

I think the only way to link to a site in these comments is to just provide the URL, which will automatically become a link.

I would have no problem with you linking to your site. I get a lot of "spam" comments that are essentially just advertisements, and I do like to keep at least a little control of what sites are linked to (sometimes there's a reason why I haven't linked to certain sites), but I don't have any problem with "sincere" links. You're a longtime reader whose name I recognize. No problem with that.

Mike J.

I would not dare question such an authority on the issue. My comment was based in part on a recent on-line article I found here (http://luminous-landscape.com/ under "what's new") reviewing the subject of optical resolution limits at great length. This article was a bit too technical for me to follow easily, but the gist of it was that real-world lenses tend to become the resolution limiting factor at about 16 megapixels for a full frame 35 mm format D-SLR. This certainly matches my experience; comparing a 16 mp Eos-1Ds mk II to the 21 mp mk III version, the improvement in resolution is barely visible in very large prints from photos taken under ideal conditions. Like many landscape hobbyists I'm addicted to zoom lenses for their convenience, but it's evident that even the very best zooms become the limiting factor in the resolution chain. Using a high-end prime lens would presumably solve this, at a big price in real-world usability.


Dear Geoff,

First off, that article:


appears to be excellent. I don't have the time to go over in meticulous detail but I spot-checked it on a couple of points that most other experts get wrong, and they got every one of them right. Which means in all likelihood they know more about this subject than I do (because there's certainly something that I've gotten wrong). The article may not be 100% correct, but it's the best thing I've seen yet. These guys are GOOD.

That said, what they say doesn't contradict what I told you. In fact, they report testing top-end lenses that produced 160 line pair per millimeter (a 3 micron pixel, half the size of the ones you're talking about). We're within a factor of two of that holy grail and that means that we are "approaching" the limit. But we still don't hit that wall smack on until somewhere around 80 million pixels.

I think a place where you have erred is forgetting that resolution only goes as the square root of the number of pixels and that's only in theory. I have long told people to never buy a new digital camera if it has less than 50% more pixels than their old one, unless they KNOW it's substantially sharper or it has other image quality improvements or features they need.

16 versus 21 million violates that rule. All things being equal, it will only give you a 15% improvement in resolution, which is half the difference between a bar group in a fine-discriminating resolution target, and that's just barely visible in controlled tests. In the real world, it pretty much never matters.

Things are never equal. The new camera might have a better sensor and better electronics and signal processing which give it proportionately higher resolution. Or it might be worse. So in the real world, increasing the pixel count by 30% might give you anywhere from a 5% improvement in resolution to a 25% improvement in resolution. But on average, it's just not worth bothering with.

(A good example of this was the Canon 11 megapixel versus the Kodak 14 megapixel cameras. The Canon was sharper than average; the Kodak was blurrier. The result was that there was almost no difference in real resulting power between the two cameras.)

That doesn't take into account that the inevitable focus errors in the camera, combined with non-infinite lens resolution means that the actual difference you see will be even less than that.

In other words, you got seduced by pixel numbers [ smile ].

~ pax \ Ctein
[ please excuse any word salad. MacSpeech in training! ]
-- Ctein's online Gallery http://ctein.com
-- Digital restorations http://photo-repair.com


Just out of curiosity, what camera/lens were you using when you made this?



"Just out of curiosity, what camera/lens were you using when you made this?"

Pentax K20D and 35mm f/2.8 DA Macro.

Mike J.

Dear Ctein-
Thanks again for the excellent information.

I am a major fan of your work. Your book "Post Exposure" should be a primary source for anyone remotely interested in printing, chemical or digital. The chapters outlining the characteristics of human visual perception versus photographic film and printing papers are far and away the most lucid description of imaging fundamentals I have ever seen. The only reason I haven't enthusiastically reviewed it for Mike's blog is because it seemed mildly inappropriate since you write for the site.
Any chance you'll be updating this wonderful book for digital capture and printing?


Dear Geoff,

Thank you very much for the kind words about POST EXPOSURE. Yes, it does have a lot of very useful general printing information. My editor has asked me about the prospects for coming out with a third edition of the book, but so far we haven't seen a reasonable way to do it-- surprisingly little of the content of the existing edition is obsolete and one of the requirements for a new edition is that at least 20% be new material. So what 20% do I cut out? We couldn't figure out a 20% that wouldn't do more harm than good.

But recently, (with regards to DIGITAL RESTORATION) I hit upon the idea of taking the excised content from the previous edition of the book and putting it up online when we do a new edition. This takes a lot of the editing strain off of revising the book. It also makes older information still available to people who are using older techniques or software, and it provides free publicity for the book.

What I need to do, in my copious spare time (ha!) is see if applying that principle to POST EXPOSURE would make it possible to come out with a more digitally-savvy version of the book. Focal Press, to their credit, is happy to keep it in print indefinitely in the form it is, but it only sells a trickle of copies every year.

~ pax \ Ctein
[ please excuse any word salad. MacSpeech in training! ]
-- Ctein's online Gallery http://ctein.com
-- Digital restorations http://photo-repair.com

The comments to this entry are closed.



Blog powered by Typepad
Member since 06/2007