Digital Photo Blog

Have “experienced” George DeWolfe’s “Digital Photography Fine Print Workshop” book.

Didn’t say read, because Mr. DeWolfe is a talented photographer but annoyingly and extremely full of himself.

It’s this latter part that made the book so hard to read. He spends a huge amount of time telling the reader how great he is, and how fortunate they are that he is sharing his wisdom.

Very tiresome.

And, of course, there are the parts where he’s wrong.

Such as in his basic workflow, where he places noise reduction as the last thing done to the photo, prior to sharpening and printing, because “Noise…is the exact opposite of sharpening.” Huh? “Blur” is the exact opposite of sharpening.

This is a computer, folks, not a sentient being. Computers run programs that are looking, in this case, for a very specific condition: CCD noise and color noise, so it can effectively remove them.

It is NOT looking for contrast enhanced, color corrected, noise. It will never find that… because that’s not what it’s looking for.

You want your noise removal done as _early_ in the workflow as possible, not as late as possible.

As to the book as a whole, I’d give these comments:

It’s a very basic beginner’s book. If you want Fine Art Print information, get “Fine Art Printing for Photographers” by Steinmueller and Gulbins.

DeWolfe’s artistic advice is much better than his technical advice. If you want a good book on photos in Photoshop, get Martin Evening’s “Adobe Photoshop CS3 for PHotographers.”

… at least for me. As soon as I went from PS CS2 to CS3, my prints came out 1.5 or 2 stops too dark. Drove me nutz.

Search on the web, and I see I’m hardly alone. Call Epson, and they admit it… but have been working on a fix for nearly a year now. Sigh.

Then tonight, I’m determined to find it, and I start mucking about. One of the things I did was turn on print preview, with my chosen paper profile. And on a whim, I clicked on “Preserve RGB numbers” and saw exactly what the printouts look like: way too dark. A clue!

Off to Adobe help, where I find that the definition of “Preserve RGB numbers” is ” Simulates how the colors will appear without being converted to the color space of the output device. ”

Hmmm… is it possible that the output is not be converted to the printer’s color space for some reason?

If so why, and what can be done about it to fix it?

Off to look at my profiles… to discover that I have two full sets of Epson 3800 profiles… one at /Library/ColorSync/Profiles and another at /Library/Printers/EPSON/InkjetPrinter/ICCProfiles/Pro38.profiles.

Hmmm… do I have a conflict here? Should not be, but on a whim, I delete the batch of profiles at /Library/ColorSync/Profiles, and head back over to PS to see how thing work now.

And the result is… “things work now!”

No more excessively dark prints.

Is this the reason that it was so hard to find? Double profiles? Some had them and some didn’t? (The ones at /Library/Printers/EPSON/InkjetPrinter/ICCProfiles/Pro38.profiles are, in fact, inside that package, so not very obvious.)

Did I find it, or were the gods just smiling on me? Taking pity?

See if that is your circumstance; see if that fixes it for you.

Good luck!

This is for the more serious photographers and videographers on the list, and there’s no doubt that some of you (certainly the pros) already know this… but some may not.

White Balance.

The human eye is very adaptive: take a sheet of typing paper outdoors, and it looks white. Take it indoors and it looks white. But that’s just our brains at work. Photograph it out doors and it looks slightly blue; shoot it indoors and it looks significantly orange. That’s because it’s not the paper - it’s the light.

In the film days, this was compensated for by using indoor or outdoor film, but in the digital age, it’s done in your camera. Every digital camera offers “automatic white balance” and often a range of cute little icons for outdoors (sunny) outdoors (cloudy) indoors (incandescent bulbs) indoors (florescents) and so on.

And better cameras will also have “custom white balance” as well.

Now: if you go to the trouble to set your white balance for a given lighting situation, there _will_ be a visible improvement in the color accuracy of your print.

The process involves choosing “custom white balance” in your camera’s menu, and then shooting something “white” filling the frame. That setting is then saved, and used as long as you don’t change the lighting in the current environment. (If you do change it; go outdoors; whatever, you have to go thru the process again.)

Except that my description above is incorrect. The bit about “shooting something white” is where it fails… because (unless you’ve paid for it specifically) “white” isn’t white. White typing paper, for example, has optical brighteners in it which reflect more blue, making it appear whiter to the human eye.

“White” as far as the digital camera (or digital video camera) is concerned is (here’s the key) -equal- amounts of 100% red, 100% green and 100% blue. On a scale of 0-255, for example, that would be R255, G255, B255. White light is an equal mixture of all the primary colors.

Your white typing paper is probably something more like R240, G248, B255. And those are not fixed numbers for typing paper… what I’m saying is that your white typing paper is really R?, G?, B?.

What you’re tying to achieve in white balancing your camera is, not surprisingly, “balance.” You want to photograph something under the “custom white balance” setting in your camera that is EQUAL amounts of R, G, and B.

And fortunately, that’s easy and inexpensive: take a trip to your local photo store, and buy an “18% gray card”… because gray IS, by definition, equal amounts of red, green and blue.

Take your white balance setting off of gray card, and you’ll see your color accuracy jump _way_ up.

“Balance” achieved.

Finally, as many of you know, if you also include the gray card in one of the photos in a given lighting situation, you can use that later on in Photoshop to help adjust the photo colors was well.

hth

Tracy

HDR Sucks.

OK: most HDR is poorly done, and that’s what sucks; even with CS3, there is no free lunch.

“HDR! Better than sliced bread! Faster than a speeding bullet!…”

As I write this, it’s the winter of 2008, and you cannot pick up a photo magazine, computer magazine, or visit a website without someone screaming at you about HDR photos.

“High Dynamic Range” or, as I call it Too Much of a Good Thing.

Remember this? Back when the Macintosh first came out, and WYSIWYG and Postscript came along, it was at long last possible to create straight lines, and use various types without having to be a graphic artist.

Proof that PostScript did not a graphic designer make, was amply supplied by thousands of barely readable newsletters sporting 46 different fonts on each page.

It was horrible. Look - just because you bought a hammer, doesn’t mean you’re a carpenter, but it does make every problem look as though it can be solved with a nail.

Well, here we go again.

Photographers have historically been plagued with low dynamic range (the amount of detail from the blacks to the whites, usually expressed in EVs or f-stops.)

Early tintypes and early film had a limited range 5-7 stops. Film runs from 8 to 19 or so stops. A digital camera runs from about 9 to 12 stops.

The human eye is about 7 stops. That is, when you’re looking at something in particular, your iris settles on a size, exactly like the aperture of a camera, and at that given size, you have the ability to see a light intensity of about 150: 1 (or just slightly greater than 2^7 - [7 stops].)

Since early photography was also about 7 stops, it was well received… but ultimately a little frustrating. Looking at the photo as a whole, the image was perfectly satisfactory, but because a photograph is a fixed moment, we always take the implicit invitation, and look into it, trying to pick out interesting details. And with a limited EV range, the detail was lost, either in the shadows or the highlights.

Thus photography became at least in part, a quest for additional range. But that quest can be carried only so far, before it starts going “over the edge,” and then no longer looks “right.”

It’s that threshold we seem determined, like a child with a new toy (or a wanna-be graphic designer with Postscript fonts) to exceed.

Look up from reading this right now. Glance around. Look at bright areas and dark areas. Really look - how much can you _actually_ see in the darkest shadows? What happens when you look at something bright? Can you see it, while at the same time, see into the shadows?

No, of course not. The iris opens and closes. In fact, you cannot -clearly- see anything except the tinniest field just around your direct line of sight, and focus. (Your mind may fill in assumed details, but as I said: really look, paying attention to what you can, and cannot, see.)

My point is that what you see is 7 EV… and if you stare into the shadows, you’ll open them up a bit… but not more than 7 EV. Ditto for bright areas. And at no time will you ever see more than 7 EV. It’s a sliding, but fixed, range. We don’t see much detail in dark areas (our eyes “have to adjust”) and humans are very bad at seeing in the dark. We do much better with more light… but it can get to be too bright for us as well, although it’s not as several an impediment as the low light level issue is.

So I’d suggest that about 12 will give us enough leeway to look into the shadows, and recoup some of the blown out whites. (The famous Zone system had 11 tonal areas.)

Fortunately, I own a camera that will provide me with about 12 EV, but what many do, and what I’m discussing here, is the 3-shot, 2 stops over; 2 stops under; one right on; blend-them-together technique so avidly discussed these days.

But, and here’s the problem: it’s not “evenly spaced.” That is, when you overexpose by 2 stops, you’re making the shadows 2 EV brighter. And in the 2 stop underexposure, you’re cutting down the highlights by 2 stops.

In short, you’ve giving the viewer a 4x extension of the light in the darkest areas where we do not see well, which is completely abnormal, since humans don’t see into shadows well, and then, with the underexposure, you’ve _cut_ (not extended) the range on the high end.

In short, you have a picture with a 9 EV range, pushed toward the shadows.

I’ve seen on the cover and in feature articles of well respected journals, read by many a professional, photos with such “high dynamic range” and they look terrible!

The problem is what happens to the middle range: the main perceptual area of the photo as a whole: it becomes “thin.”

In fact, I actually mistook the cover of a recent “Photo Techniques” for a pencil and watercolor drawing on first glance… as did my wife, who is a painter. It was a photo taken inside an old, trashed and abandoned room, with windows opening to a sunny day outside.

Yes: I could see into all the shadow areas, and make out details out in the bright sunlight. However (and here’s the point) I have never had such an experience in my real life, using my real eyes… nor has my wife, which is why, on first look, we took the photo to instead be an illustration: it simply had no relationship with reality.

In fact, it reminded me of what happens in poorly done audio these days: it’s compressed (I’m talking frequencies, not file size here) beyond all belief so that it can be played loud. Musical nuance is gone; harmonics are gone. It’s a large, flat monotone without any human expression…

… and it fails as music.

Most HDR is similarly overdone: “brightness” is just tacked on to the image in a haphazard way, “opening up the shadows.”

It fails as a photo, as it’s no longer within the range of human experience.

Fortunately, just as we are no longer stuck with 46 fonts per page, and the wanna-be graphic designers eventually returned to their day jobs, leaving the trained designers to produce nice works with no more than 2 fonts per page, there _is_ good HDR.

It’s not simple to do correctly, but at least there is a correct way to do it.

First, make sure that you’re not shooting automatic anything! Particularly, be sure automatic white balance is off. Turn off auto-focus. Use a tripod. Shoot -2, 0 and +2 EV.

The, for your print, you open up the shadows, but not necessarily all two full stops. You put the extended range in the _middle_ of the picture, not at either end. Leave a bit of blow out if it’s called for. Leave some pure black.

You want darkest pure blacks; blacks with only a hint of shape; blacks with some tone; blacks with some texture. If you open it up so much that none of your blacks fall in the first three categories of that list, you’ve overdone it.

Extend the detail in areas of the photo, not all over it. THINK about what you are doing. If you are sharpening to draw the eye, then provide the extra range there, where the eye is drawn. That’s what masks are for!

Open the shadows a bit, but only in the deepest areas; do not let it spill over into the middle tones. Opening the shadows should be done very modestly if there is a 150:1 tone range (ie daylight) but can be done more strongly when the tone range is 20:1 (a darker subject) because that is the way our eyes work.

Do not increase the shadow detail in the middle range (or at least not much) - keep it at 8 or 9 EV… for exactly the same reason. (Ansel Adams considered the bulk of the image to be within the center 9 zones.)

In fact, go look at some good black and white photos, and notice the tone range. After all, EV is a measure of luminosity, not color. Study the prints of Weston, Adams and the other great B&W photographers. There you will find a rich, velvet and lucious range of tones, all of which invite investigation.

Personally, the first thing I do when working raw is turn off the color and look only at the luminosity, in LAB, and get that set right. Then I pop the color back in.

Having HDR can be a good thing, yeilding spectacular and moving prints… or it can be like a newsletter with 46 fonts per page.

If photography were a entirely mechanical craft, then Photoshop could do all the work for us, but it’s not: it’s an art, which takes human experience, skill and judgement to do correctly.

You want a great photo? Plan on spending some quality time with it.

Here’s an introduction to how to do color matching; what it is; how to set it up; and why it works. If you’re new to digital photo printing, or just confused as to why your prints don’t look like what you saw on screen, this simple “no geek speak” intro is for you. I wrote it years ago, but it’s just as valid today as it was 4 years go.

It’s so long however, that I’ve put it on a separate web page, which you can find here. (www.tracyvalleau.com/colorprofiles.html, just in case the link gets broken…)

hth

Epson 2200 banners longer than 44 inches can be made by using GIMP print.
(Originally posted 2/5/05.)

In fact, longer than the roll paper you can buy (which is 32 feet.)

Here’s a couple of things to be aware of:
1) there are no profiles for GIMP. (I’ve made my own using PrintFIX, but the license won’t let me share them. Sorry.)
2) The default GIMP settings over-saturate the image.
3) The rotation direction is 180 degrees opposite from that used by the Epson drivers: instead of left-edge first, it’s the right edge.

And one last tip: if you run out of ink (OK: before you run out of ink, and when you see the blinking lites indicating low, and you decide you don’t want to risk it) do NOT use the software to pause the printer for reloading!! (You’ll have to start over again.)

Simply hit the ink button on the front of the printer, and replace the cartridge. It will continue printing seamlessly.

How to avoid tearing your hair out when trying to print roll-paper banners on the Epson 2200 using the Epson drivers. (originally posted 10/28/04)

I lost a huge chunk of hair, and about 24 hours trying to figure these hints out:

If you’re not using the GIMP print driver, (see other blog entry) then here’s how to print out a banner on the Epson 2200 using the Epso-supplied printer drivers.

Seems that all this has to happen:
1) paper cannot be 13×44″, I set a custom paper size to 12.5×43.75 (web reports seem to be true: if 44 or larger, the driver doesn’t orientate the photo properly)
2) turned off centering (after centering the image)
3) had to set it for 2880, as 1440 caused the job to refuse to print (stopped it without running it).

Also note that the image will load with the wrong paper size in print setup dialog. Before moving to print w/ preview, go to page setup and set proper paper size, and then save the file. (Yes, the save seems to be necessary)

In print/preview, verify that the proper page size (see above) is, in fact the one showing.

use a 2880 profile (Mine was SP2200 prem.luster 2880.icc)

In print dialog, select “roll” printer (not borderless; not ‘borderless banner’)

Select paper as paper-saving cut; cut sheet (but I did it without the cutter installed…)

The “Shutterbug” editor, and apparently a number of readers, don’t understand the nature of digital. (originally posted 2/13/05.)

I’m in a unique position vis-a-vis digital photography, as I’ve been a photographer (with darkroom experience) for 46+ years, and have some nearly 30 years of detailed computer experience (I was a programmer) as well. I jumped on the digital photography bandwagon with the very first digital camera, and have long since abandoned film.

It is from this perspective that I’m making the following comments, and taking to task George Schaub, editiorial director of “Shutterbug” magazine.

George has been on a rant for the past two issues, complaining about the unknown longevity of digital media. “What ever will happen to our precious photos with this media?” he wonders. He seeks “… a viable medium the will not be lost to us later in our own lives, and to generations ahead.”

“…one of the main detriments folks see to digital is just that - that we have yet to be convinced that it’s a reliable keeper of memories. Digital memories, stored as bits and bytes are not, like a well-stored print or properly processed piece of film, hard -wired; they are virtual and sit on a medium that more often than not seems transitory.”

To his credit, George goes on to point out that, in fact, film and prints are -known- to fade over time, and to shift colors. “It took years for the photographic industry to first own up to, and then do something about creating more stable color images.”

“Perhaps it will take just as much time for the digital industry to own up to and do something about the reliability of the bits and bytes on which many of the memories we create today reside.”

This strikes me as not only silly, but a bit misinformed.

Current film and print technology (non-digital) is not permanent; in fact, it begins to change color and chemistry the instant it is removed from the bath… although it may take 20 years for the effect to be noticed by the human eye.

Yet he takes to task the digital industry because he “feels” insecure and things “seem” unreliable. He calls “bits and bytes” (they are the same thing: a byte is just 8 bits) “virtual” while lauding a print as “hard-wired.”

Well, I’m not sure what that means, if anything at all. Prints and film are molecules and grains of silver - molecues and grains are not photos. They change constantly and deteriorate. What is any more “virtual” about a bit? Once burned to a CD, that area of difference is just as “real” as a burned grain on a photo.

I suspect that it’s just ignorance at work here. It’s fear of the new. Photographers are not generally computer gurus and what they don’t understand, they fear. They may think it’s “virtual” because they can’t see it, but I assure you it’s as real as any negative.

And in terms of longevity, the irony of his recent editorial is underscored by an advertisement in the back of the same magazine for a CD with an expected life of 300 years! That’s 15 generations; twice as long as the current history of photography as a medium!

Let’s look at longevity of digital media. Unlike analog media (film) digital is simple: it’s either on or off. No gray areas: the bit is either there, or it isn’t. The CD is either burned or it’s not. This simplicity leads to reliability, as opposed to the infinite shades of gray in the analog world.

It further leads to 100% -perfect- duplication. Once you shoot on film, you cannot duplicate that negative or slide with 100% fidelity… but a digital photo you can, a thousand times over… and each one IS the original! Worried about storage? Why, with the ability to make an infinite number of perfect originals? Further, after storing that original for 20 years, if you’re worried about the medium itself beginning to deteriorate, just make a 100% perfect copy on a new CD, and you’re good to go for another 20 (or 300) years.

Try that with film.

The recording media industry doesn’t need to “own up” to anything. The details of what is known and what is not known are discussed ad-infinitum within the computer world. Maybe not the photo world, but the discussion is there, to be read by anyone who is interested.

The National Bureau of Standards has run longevity tests on CDs and is running them on DVDs now. That data is public knowledge: your tax-dollars at work.

Manufacturers are working diligently to make ever more durable and long-lasting media, just as they did with the quality of film and paper, evolving over the years.

There is no “dirty little secret” in the closet; nothing “the industry needs to acknowledge” nor anything it is hiding.

Media as it is now, properly stored, is perfectly stable: bits don’t change by themselves, like magic. They are not “virtual” (whatever that means) nor subject to alteration on a mysterious whim.

Can a digital media file be destroyed? Sure. So can a negative or a print… it’s just different things that destroy them.

Look: digital is not a panacea - it is just another medium. And as another medium, it is subject to the constraints of that medium, which are different from the constraints of film and chemicals.

So, if what George is complaining about is permanence, all I can say is “What? 300 years isn’t enough? What do you want? Isn’t that at least 250 years better than what you have now?”

Outside of the fact that you cannot hold them up to the light to see an image, I cannot think of a single way in which digital storage of files isn’t superior to film.

What is at work here is not media problems, but ignorance. It’s a case of “the devil you know, versus the devil you don’t.”

Ignorance isn’t bad: it’s simply the state before becoming informed. I’m immensly ignorant of thousands of subjects.

Here it is in a nutshell: take a photo in the analog world, and you are 100% guaranteed that it will eventually deteriorate and fade from human history. Store it as a digital file, and with a bit of human help copying it every 20-300 years, there’s no reason that original image, exactly as it came out of the camera, won’t be here at the time the universe ends.

You decide which one is “virtual.”

Once again, it’s necessary to correct photographers about computers. Here’s why digital photos are NOT “codes.”

Oh good grief: words have MEANING!

In a recent (July 2006) Editor’s Notes in Shutterbug, George Schaud says “… digital images are composed of codes, either/or binary ‘addresses’ that can be swapped for others without much regard for the original codes set when the image is recorded and processed.”

Leaving out the adjectives, that becomes “codes are address” and thus, “digital images are composed of addresses.”

Well… no; they are not. Digital image files are composed of numbers, from zero to something, intended to be grouped in threes, one each for red, green, and blue.

(Yes, a number is represented by “bits” - binary switches that are either on or off, but I have not twiddled with bits since the early days of assembly language programming, and trying to get a photographer to make the mental jump from binary numbers to, er… “codes” borders on intentional obfuscation, if not the self-agrandizement of “jargon speak.”)

There is nothing mysterious about this, and calling them ‘codes’ is merely confusing. They are just numbers representing the intensity of light hitting that particular sensor.

And they are not even remotely ‘addresses.’

Look: an address is an address. Think of your mailbox. That’s an address. There is nothing in it… it’s just a container. You can put a magazine in it, or you can put a letter in it.

The magazine in the mail box can be swapped out for a letter… but the address on your mailbox does not change, as George seems to think (…”addresses” that can be swapped…”)

When one of those numbers, representing a level of light is loaded into a computer’s memory, it has to reside somewhere… and -where- it resides is it’s address. Each time you load that particular file, the address where a given pixel resides is free to change, just as you can put the same magazine into many different mailboxes, and each mailbox can hold various things.

The reason computers are useful at all, is that you can change the numbers inside the address, say from 200 to 220, making the computer interpret that pixel as brighter than recorded.

I think George has a vague idea of that, or perhaps even a firm grasp, but he does a terrible job of explaining it, using words that have a specific meaning in the wonderful world of computers, without properly understanding them.

“Code” is a programming instruction, telling the computer to do something, like “add 20 to the number that is in address 34567.” (”Code” does NOT mean “number.”)

“Address” is a memory location that can contain a number.

“Binary” is base-2 numbering; II = 3; IOI = 5; IIIIIIII = 255. Computers use binary numbering because computer memory is just a bunch of on/off switches. 255 is 8 switches in a row all turned on.

The rest of George’s note is pretty much accurate. But I do wish he’d either do a better job of explaining, or a better job of understanding.

In either case, his failure to do so results only in confusion, as when the reader later hears the words “code” “address” and so on, used properly, s/he will be forced to decide which version was the proper meaning.

I can’t imagine him calling developing fluid “water” and there is no excuse for using the wrong words in the digital domain either.

The next entry, called “A Bit of a Nibble…” actually explains, in simple and clear terms and analogies, how computers work, and is aimed at anyone interested in what’s really going on in digital photography.

While Scott Kelby may be the world’s leading computer author, that doesn’t make him right… Correcting misinformation for digital photographers. (Originally posted 2/12/05.)
Scott Kelby is, according to Amazon.com, the world’s leading computer book author. His fame rests in his works about Photoshop, and I’d be the first to admit that he’s made great contributions in that realm. On the other hand, I started working with computers when he was about four years old; Photoshop at version 1 and digital cameras when they first came out.

So, I’ve assigned myself the task of correcting the misinformation in his books, and misinformation about digital photography and computers in general. I’ll not limit this to Kelby, but will begin with him.

In his book “The Photoshop CS Book for Digital Photographers” (New Riders, 2003) he states on page 149 that “…16-bit offers 65,536 possible levels in each channel.” This bit of misinformation is promulgated throughout third-party literature on Photoshop.

One can be slightly forgiven for assuming that 65K worth of levels are used, since 2^16 = 65,536, but then one would equally assume that 16-bit raw files have 65K levels too, and for the exact same reason, both are incorrect.

In fact, Photoshop allows 32,768 levels, because it uses only 15 of the 16 bits. This can be verified by simply putting the info palette into 16-bit mode, and clicking on “display 16-bit values.”

On the same page, he states “To get the benefits of editing in 16-bit, you need to shoot Raw (sic) 16-bit photos…” Well, sort of. Most raw files are in fact 12 bits of data, with 4 bits padded with zeros (bringing the total bits to 16.) In short, the camera is recording 4096 levels of information (0-4095), so that even loading in a raw image does not result in a 16-bit range of data, nor even a 15-bit range (neither 65,536 levels or 32,768) but in fact 4096 levels per channel.

While this is truly counting the angels on the head of a pin, as 4096 levels per channel is still a stunning amount of information, I merely hope to set the record straight. 256 levels per channel yield 16.7 million possible colors (although not all would be distinguishable to the human eye) 4096 yields a staggering 68 billion color possibilities.

And just FYI, the difference between a million and a billion is about the same as the difference between two weeks and 32 years.

Next, another oft-repeated bit of misinformation about Photoshop is recycled on page 145 of the same book: “…press option-delete to … fill with black…”

In fact, option-delete fills with the foreground color, whatever it may be. (Command-delete likewise fills with the background color - again, whatever color that happens to be.)

(To use the keyboard to fill with black, one has to first press d to reset the default colors of black in the foreground and white in the background. Only then can one be assured that option-delete will fill with black.)

Next up: one photography magazine editor and his failure to understand digital technology…

… stay tuned.