« Rare 1888 Keller / Sullivan Photograph Comes to Light | Main | Why I Hate Infrared »

Friday, 07 March 2008


First and foremost your experiment illustrates quite nicely that a photograph is not a copy of reality captured on paper or screen. With respect to your third point, unless you are required to document a scene exactly as shot there is always an element of interpretation in the preparation of an image for output.

Given the fact that everyone has different visual perception I don't think it is possible produce an image that everyone will judge as correct. It certainly is not high on my list of goals for for producing a successful image.

I would contend that there can be no objective standard for fidelity in a photograph beyond ensuring that all the elements that were originally captured are still present in the finished product. The camera is not a magic box for capturing reality.

"A print from the version that looked best (most realistic) on the monitor looked horribly bland as a print, and the print version looked horribly overcooked on the screen."

That's interesting. Did you experiment with soft proofing in Photoshop? That can make an image look subdued. Using different color transformations (Relative, Perceptual, etc.) in the printing and/or soft proofing might also have an effect on how different the screen and print versions look.


It's nice to see this effort; it confirms my own thoughts on "accuracy".

Some thoughts on this, and the "saturation" article.

Visual memory is trainable. Before digital, I worked at the process of remembering color, or a scheme in paintings. (See website) Now, I just "chimp" the camera. I've had good results with a series of Canon digital Elphs, neutral sttings, auto WB, taken fast under a real mixed set of different lighting conditions. WB is most important for any kind of accuracy, but I've found auto to be better than me. Since these are just "notes", not a lot of time is invested. For low light I use a monopod, smallest Gitzo Traveller w/RRS B25 ballhead. Highly reccomended!

As to lesson 3., yes, go for it, though it's nice to know there is a "base" of reasonable accuracy, within the constraints of the medium. Julie Heyward, http://www.unrealnature.com/, has some nice points about her limits, reality constraints because of what she is doing. If she's reading , maybe she can point to the comments.

Some great thoughts in these two articles, though what about that doggy?


Well Richard, all this is way over my head, but, aren't all photos on monitors viewed by transmitted light? Whereas prints are viewed by reflected light? It would seem that in the film world this would be akin to looking at a MF transparency on a light box and then looking at a print of same. There hardly no way they could look the same.

Very interesting article. If I could go off topic for a sec, I couldn't help but notice all those external drives. I recognize the brand. I was wondering if you have had any trouble with them? I ended up having to replace over $20,000 worth of those drives. I would boot up and the drives would not appear even though they were working fine the day before. I would then get a message saying "drive not formatted." If I formatted the drive (losing all my data) it would work fine for a while until it would happen again. This happened on 12 out of twenty drives. The company was no help at all.

I always feel obligated to warn about this, as I do not want anyone else to lose their data.

The doggy (Lulu) has to stay tied up outside at my brother's house because of the two cats and various critters (rodents'n'reptiles) which live inside. She considers it her duty to bark furiously at every distant bark, passing pedestrian, and (especially) every squirrel in sight--which are plentiful, since the neighborhood is blanketed with mature oak trees. It keeps her very busy. She sleeps soundly in the car on the way home.

She definitely does not like being tied up and kept away from the action. She is an in-the-thick-of-things type of dog. One of her funniest habits is that if there is ever a fight or a dominance match at the dog park, Lulu will go over and insert herself physically right into the situation. She doesn't join the fight--she is possibly the best-natured creature I've ever known, and I've known many--and she never asserts dominance herself (she is submissive)--but she loves to be *right* there where the action is.

Another funny thing she does at the dog park is that while she doesn't fetch herself, when there's a retriever fetching Lulu will run right alongside it just to be companionable. She never fights for the ball or the frisbee, she just likes to run with the runner. I call it "vicarious fetch."

Mike J.

"is there a "saturation dynamic range" akin to the luminosity dynamic range?"

Yes there is, in the form of the color gamut. And print is famously limited in that respect, not really able to reproduce even all colors and intensities a normal screen can show, never mind the eye.

But mostly I think you're being bitten by the fact that color perception is not independent of our expectations. We do not process our view only as if it were a flat image; we interpret things differently depending on object identification and on depth cues. And the flat print will not give us the same depth cues that reality, and so the colors, sharpness, contrast and so on will appear different to us.

What we think we see directly really is a constructed reproduction of the image we get, heavily influenced by our memories, expectations and so on. Just knowing one scene is a print is enough to alter your perception of it (and it really does alter your perception; your changed expectations change how colors and contrast is being processed early on).

Doesn't "accurate" white balance depend on how long you've been sitting in the room? There was an installation in Tate Modern (someone remind me whose work it was?) a while back where visitors got to sit in room bathed in blue light and then got to sit in a second room bathed in red and giggle at how their eyes went all funny. Presumably, if you've just walked in from outside or, indeed, a pink bathroom, your work area will look different from how it looks if you've been sitting in there for a while.


Glad to hear no dogs were harmed in these delvings into color.

Only recently have I returned to having a dog in the house, ostensibly my daughters, but we have become best buds. Renewed empathy for dogs; I was worried.


Nice article, Richard.

It comes right on the heels of an epiphany I had the other day as to why my printer seems to have lost its accuracy. I thought I needed to get custom profiles made for it, but couldn't understand why the original Epson profiles, with which I'd been happy for years, should "wear out". I bought the Epson 2100 a few years back when I was using an old Samsung LCD and sometimes a Sony CRT. It was easy to get a good match between monitor and print then.

The trouble today is that the modern LCD monitors are so over the top in terms of saturation and contrast, compared to reality. A modern LCD looks great to the eye - giving a real "wow" factor, but is way beyond what the printer can deliver. Our eyes adjust to the colours and contrast of the monitor, and we accept what we see on the monitor as "right", but with a penalty in the form of a changed perception of the prints, that now seem so dull in comparison to what we see on a modern monitor.

The older LCDs and CRTs were a much closer match to reality, and therefore to the print too.

Re the $20K worth of bad drives: is it possible you ran into a bad lot? I've seen a number of situations where my soon-to-be-former employer has had to replace a bunch of the same type of hard drives or motherboards in fairly short order. The hardware vendor would naturally never admit to anything, but once the serial numbers started changing a bit we never had (more than normal) trouble with that hardware.

Dear John,

Displays use transmitted or emitted light, 'tis true, but it's not of import. The difference in appearance between a chrome and print is 90% due to the greater density range of the chrome and(in the case of wet darkroom prints) the higher color-purity of the chrome dye set.

You match dye sets and density ranges and there's little perceptual difference between chromes and prints (one can approach this state with dye transfer prints).

The BIG difference between monitors and prints is that monitors are RGB display devices and prints are CMYK devices. RGB and CMYK color spaces overlap but are not coincident. That's what the 'out of gamut' preview setting in Photoshop is all about.

pax / Ctein

Dear Puplet,

Yup, color accommodation takes a LONG time. People used to think it happened in a matter of tens of minutes, max. Turns out it can take much longer. Move from a daylight situation to an incandescent one, for example, and it will take several HOURS for your color vision to equilibrate.

(this from a paper in NATURE mebbe 15 years ago, but no way I could come up with the reference if someone wants it, I'm afraid)

pax / Ctein

Not sure if anyone will see this, buried after so many other comments. Anyway, a couple of things :

1) Great article that shows how difficult getting accurate color is, especially from memory. I think it also, in my mind, makes it a pointless excersise to try to be too accurate. "Realistic color" is probably a better term. Of course, being slightly color blind, I've always wondered what all this talk about accurate color really means anyway.

2) The Nature Photographers Net post above might be oversaturated to many, but those kinds of shots are definitely not meant to be completely accurate. In any case, if you want to see some terrific shots, most of which fall into the "realistic" category, check this gallery out :


Richard's comments mirror my own experience.

I was never a great printer in the traditional darkroom. I did a lot of printing, don't get me wrong, but no one would accuse me of being a fine art printer.

But I do have some skill at processing photos for CMYK offset reproduction, since that's a fair part of my current position. CMYK printing is more art than science, with the added stress that you won't see the results until the very expensive job is starting to run on the very expensive press, and by then it's basically too late to change anything.

A good CMYK conversion looks just terrible on a computer monitor. Over sharpened, over saturated, too much local contrast -- yuck. But an image that looks great on a monitor looks flat and pale and soft and ugly on press.

I finally went out and joined the digital darkroom revolution last fall, putting an Epson 3800 on my desktop. It took months of work to get prints that made me happy -- until I realized that I don't want to match my monitor -- what I want to do is make great prints. Now I think of them as comparable to my CMYK offset print jobs (though not as extreme), and I'm getting much better prints. They key in my case is to substantially boost local contrast (using the Unsharp Mask tool), and add just a touch more saturation than I think it needs. And, in the case of the Epson, bump up the shadows in Levels just before printing. Works wonders.

i didn´t get the point on this experiments about reality. You know... everyone has it´s own perception of colors, saturation, contrast, etc... I think you´ll get to a point that a photographer to "make" a picture will be unecessary... a robot can do it perfectly calibrate with reality.

I was wondering...if you continue with this search you´ll need to ignore BW pictures, because the saturation is wrong. Velvia pictures...too.

BTW.. i am a photographer from Brazil. I love to read your blog and i think you always have a good point to discuss here. Keep up the good work, and forget about reality... there is no reality just interpretations of something we don´t know at all.

A perfect display to print match (by their nature) isn't possible, but you can get a fair way there by setting realistic display luminance (for my work area about 90 cd/m2), tuning the colour temperature of the display to that of the paper and having controlled viewing conditions.

That said, if you want a good starting point for colour you may want to read up on this:


ACR and Lightroom by themselves are poor tools for accurate colour but zeroing settings (other than white point and exposure) and applying the calibration script created by the author of the above paper will get you surprisingly accurate colour. Then you'll need the means in Photoshop to adjust tonality to the desired output medium without the hue and saturation corrupting side effects of most Photoshop tools (you can forget the RGB composite curve). Read the other papers on the above site, or check out Lobster 2 available from here:


After you've got your workflow sorted it comes down to aesthetic judgements about what representation (departure) you want in the print. There's a lot more to satisfying prints than just having good profiles (whatever "good" means). All this comes under the definition of "craft". There's an unfortunate preoccupation with equipment (cameras, monitors & printers) as having an overriding bearing on the results which couldn't be more wrong.

Stephen Best
Macquarie Editions

I have to admit that my first prints from the same printer (Epson 2200/2100) were something of a disappointment.
They looked nothing like what I could see on my monitor(CRT) even though I was printing to the correct paper profile etc.
Much duller and lifeless in line with the transmitted light versus reflected light variation. However, after a day or so (and several ink cartridges) softproofing in Photoshop I eventually arrived at print output on enhanced matte which is very satisfactory.
I used a standard print calibrating image and saved the softproof tweaked curves and H/S as an action. I applied curves to match the grey step wedge in the calibrating image and H/S on any shifted colours.

The result is I am very pleased with the output from the Epson 2100 and my only beef with it is the pissy small and expensive cartridges.


Your statement, "Matching the luminosities mattered a lot" caught my eye. I seem to remember from a biology class (many, many years ago) that this is a result of the evolutionary process: luminosity differences are what allow us to separate our food (prey) from the natural background. That's why rods (the luminance detectors) are something like 20 times more plentiful in our eyes.


The comments to this entry are closed.



Blog powered by Typepad
Member since 06/2007