[Note: The example print sale ends Sunday night. —MJ]
I came upon the scene right after visiting Pier 24 in San Francisco. Happily for me, I was carrying my trusty Olympus Pen E-P1 with my favorite lens, the 45mm ƒ/1.8 (reviewed here). ISO 100, open up two thirds of a stop to compensate for the bright sky, ƒ/5.6, and Ctein's your uncle.
Once I got a good look at this photograph on my large monitor I knew it was something special. Lovely exquisite detail everywhere, delicate tracery in the clouds and tones on the bridge tower, and an amazing degree of sharpness. My, do I love that lens.
In order to eke the very last bit of quality out of that data, I super-sampled the image in Adobe Camera Raw, pulling it into Photoshop as a 16-bit, 28-MP (6144x4608 pixel) file in ProPhoto RGB color space. Why the upsampling? Because I have a (possibly superstitious) belief that ACR, having access to all the Raw camera data, will do the better job of extracting every last bit of detail from the file, and I know from experience that more pixels get me better results when I run deconvolution routines to maximize sharpness.
Once I had the image in Photoshop, I ran the ContrastMaster plug-in from Photo Wiz to kick up the gradation and tonal separation in the subtle, fine detail just a bit. It did a great job of restoring the delicate tracery in the clouds and the subtle variations of light on the bridge tower that I had seen in the original scene. It also kicked up the noise (just more subtle, fine detail so far as the algorithm is concerned).
Next I ran Topaz Labs' InFocus with a one pixel de-blur radius using the blur function calculated for this particular photograph. Deconvolution algorithms like this produce a genuine increase in resolution and sharpness. I use them routinely on all my digital photographs...just a touch. They'll also sharpen up and enhance any noise or grain in the photograph. Between the two plug-ins, I had more noise in the sky than I really wanted to be seeing. Normally I have no complaints about the noise levels in this camera at low to moderate ISOs, but here I was really pushing hard.
I needed some careful noise reduction. I didn't want to lose any of the fine detail and subtle gradation in the bridge, and noise reduction algorithms always clobber a bit of that. Time to make a mask so that I could apply noise reduction, using Topaz Lab's DeNoise, to the sky alone. Making that mask required lots of fine tuning and a certain amount of pixel painting where the bridge cables were almost the same tone as the sky. There are limits to what even the best masking software can do. I had a couple of hours of fiddly work getting the mask just so. Such is life.
From that point on, it was pretty straightforward work. A few masked curves adjustment layers for dodging and burning to get local brightnesses just as I wanted them, another curves adjustment to get the overall tonal placement just so. Surprisingly, no color correction; the camera had nailed it perfectly. That almost never happens.
I printed the finished photograph out on my Epson 3880 printer as a 17x22-inch (15x20" image area) print. The printer settings were 16-bit output (I've never seen this make any difference in a print, but I figure it can't hurt), 2880 dpi with high speed off, and finest detail on.
Twenty minutes later I was looking at a very decent first print. A bit of curves tweaking to get the print tones just so, a little more dodging and burning, a bit of pixel painting to catch a bit of cruftiness that it slipped past my masking, and there it was. A superb print of a photograph that still makes me very happy every time I look at it.
That's it for this time, folks. I'll be taking next week off, while I deal with all the orders. See you again in two weeks. Meanwhile, keep those cards and letters coming in.
Note: Links in this post may be to our affiliates; sales through affiliate links may benefit this site. More...
Original contents copyright 2012 by Michael C. Johnston and/or the bylined author. All Rights Reserved.
Featured Comment by Sven W: "Hmmmm...is this exercise a measure of a Micro 4/3 camera or Ctein's excellent post-processing skills?
"Quote #1: '...I think exemplifies the very best that a Micro 4/3, 12-megapixel camera can do. It's not a hero experiment; I can go out any day and make photographs of this technical quality. I just can't see any possible way to make a much better one.'
"Quote #2: 'I had a couple of hours of fiddly work getting the mask just so. Such is life.'
"The two quotes imply Ctein spends 2+ hours post-processing a typical image. I suspect that's way more than what most people do for their 'any day' images. Truth be told, I don't have enough post processing skills to spend two hours on one image. ;-)
"I'm not making a pot-shot here...just getting my head around Ctein's level of dedication! I spend around 20 minutes fixing up my images in post and have been thinking about upgrading my camera to reduce time spent in post. Maybe I need to re-think this!
"As an aside, what level of camera would produce the (equivalent) bridge photo as a JPEG? A D7000 or a D800, for example?
"Anyway, thanks for the print offer and the write-up. You've given me some ideas to explore."
Ctein replies: As I said in response to a similar question on Wednesday, "...while I did my usual superb custom printing job (he said immodestly) this is no trickery. The processing I did to turn this into a fine print is pretty straightforward; it is about what is inherent in the medium. It's just ordinary Ctein custom printing magic, not how clever I can be at manipulating pixels. I'm not trying to win some argument by hook or by crook, here. So far as I'm concerned, I won that argument years ago. I'm only giving people an opportunity to see for themselves instead of having to take my word for it."
In other words, this is about what the camera is capable of, not what I am capable of.
As I mentioned in a the Comments, I do typically spend hours custom-printing one of my photographs, but the time for any particular photograph runs from less than an hour up to several days. That puts me a cut above most other people, but if you're trying to show what a camera is capable of, you really need a superb print. Otherwise it's like trying to show off your audio system by hooking it up to $10 speakers.
An interesting question you and others have raised: would a camera with more pixels or a larger sensor reduce the amount of work I do printing? The answer turns out to be no. I've printed everything from six-megapixel quarter-scale sensor to 40 megapixel medium-format sensor photographs. The amount of time it takes to print a photograph to my satisfaction depends entirely on the photograph, not the format.
For example, the fastest-printed photo I alluded to previously is this one. Printed in half an hour from start to finish. From the Olympus. It just happened to be one of those 'perfect negatives' that comes along once in a blue moon and takes very little work, and you just nail it on the second print. (11x14-inch prints, in this case. My printer can't print fast enough to turn out two 17x22" prints in half an hour.)
Contrariwise, this crop from a medium format sensor took quite a few hours of work.
Can't really explain to you the whys and hows in more detail without writing several book-chapters worth of verbiage.
Featured Comment by Alan Fairley: "It sounds like you essentially applied two rounds of sharpening once you had the photo in Photoshop. Did you also apply capture sharpening in ACR? If so, how strongly? As a 4/3 and Micro 4/3 user, I am very intrigued by your workflow. Thanks!
Ctein replies: I routinely apply a modest amount of sharpening and noise reduction within Photoshop to my digital photographs. The results are so dramatically better that it is the norm for me, for any Bayer array photograph I deal with, everything from six megapixel quarter-scale sensor to 40 megapixel medium-format sensor photographs, with or without an anti-aliasing filter.
Usually I leave the Raw conversion sharpening in ACR at the default setting unless I'm dealing with a photograph with pinpoint specular highlights. Then it puts ugly rings around them. Absent that, it's on.
Once I have the photographs in Photoshop, I typically use Smart Sharpen at a radius around a quarter pixel and a percentage in the 50% range, for photographs ACR-converted at normal resolution. Tune to taste. Smart Sharpen is a deconvolution routine; it actually does improve fine detail, not merely enhance edges. I'm quite astonished at how much more detail there is in those Bayer array photos.
I also usually apply a low level of noise reduction using a third-party plug-in (there is no one best plug-in; some work better with some photographs, others with others. No rhyme nor reason to it).
Between the Smart Sharpening and noise reduction, I can produce a photograph with less noise than the original but the same sharpness and detail, the same noise as the original but better sharpness and fine detail, or something in between. Depends on how much of each tweak I use. Usually there is one combination of the two which works very well over an entire photograph.
Sometimes it works a bit better to apply sharpening before noise reduction, sometimes the other way around.
I also happen to own Topaz Labs InFocus which is a deconvolution plug-in. It's more sophisticated than Smart Sharpen but not as good at small radii, because it increments the blur circle in one-pixel steps.
In this particular photograph, since I was supersampling it, I needed to de-blur with a radius more like one-two pixels for maximum quality. I did a bunch of pixel peeping and decided I liked the plug-in better than Smart Sharpen, just by a bit.
By itself, this amount of sharpening doesn't produce killer noise. Yes, there is noise gain, but it's tolerable. What made it a killer was having applied ContrastMaster. It's a plug-in that improves local tonal separation and detail, but it's really hard to tame and it really latches onto noise.
As for the very good question of why I didn't do the masking before I did the sharpening? Well, because I didn't think of it. Normally, I don't need a mask, as I implied above. In this case I was far into the process before I figured out that there just wasn't going to be a set of compromise settings that would work for the clouds, the sky, and the bridge, all at the same time. That's when I built the mask. Had I known when I started I was going to have to go that route, it would've made a lot more sense to do it at the beginning. Given that it took me a couple of hours to create that mask, you can understand why my head didn't want to go that way. (Lazy smile.)