http://www.flickr.com/groups/33051635@N00/discuss/72157627375843536/
Category: Theory
All there is, right now (expanding with time):
http://www.dpreview.com/learn/?/Guides/
Color profiles, display and print – a “must read“. Add Scanner profiling to the lot!
http://www.dpreview.com/news/1107/11072810newprintingguidepart2.asp
Choosing a printer
http://www.dpreview.com/news/1107/11072520newprintingguide.asp
http://www.printernational.org
Paper sizes, color theory, Photoshop, InDesign, File formats.
Early writings on the emotional effect of photographs on the viewer
Who: http://en.wikipedia.org/wiki/Roland_Barthes
What: http://en.wikipedia.org/wiki/Camera_Lucida_(book)
Who: http://en.wikipedia.org/wiki/Susan_Sontag
What: http://en.wikipedia.org/wiki/On_Photography
Follower, who refers to these: http://www.multijournalist.se/
http://www.rideau-info.com/photos/mythdpi.html
Pictures, Prints, Size, Resolution
Some abbreviations or TLA’s that are omnipresent.
DPI = Dots Per Inch
PPI = Points Per Inch
Measures and their meaning – in the metric system
One inch = 1″ = 25.4 mm,
Four inches = 4″ = 101.6 mm, about 10 cm,
Fourty inches = 40″ = 1016 mm, about 1 meter,
12″ = 1 foot, about 30 cm, add one inch and you have about 1/3 of a meter,
One A4 sheet is approximately 8.2 x 11.7″, or more exactly 210 x 297mm.
(A0 is 1 square meter with the long side 1.414 times longer than the short one. Cut an A0 sheet in half, by cutting the long side and you get an A1 sheet. Cut that the same way and you get an A2, cut again for an A3 and again for A4)
A usual print size in sweden is 15 by 10 cm – approximately 6 by 4 inches – which is about the same size as an A6 sheet (148.5 by 105 mm).
Most digital cameras uses “EXIF” which allows to record metadata (information!) inside the images that are shot. Among that information is Date and Time as shot, most camera settings and ”image resolution” which can be perceived as a “DPI”-value. Older cameras might just provide “72″ here, others might allow you to enter a default value yourself… My D300 just provides a preset of 300 (not changeable in camera).
For the camera, this value has absolutely no meaning. And it still has absolutely no meaning – until you’re about to produce printed images, display it on a computer screen or a TV set.
As you resize the image to make it appear appropriately on the screen or paper – you’re “setting a scale”. For printers and screens you can see the required scaling as a grade of how fine images they can display; a “dot pitch” can be calculated from this, which in fact is the inverse of a DPI-value when taken in relation to a measure – most commonly ONE INCH.
An example: My old Laptop displays 1280 by 800 pixels. Each of these can have any of the colors provided by a total of 32 bits – e.g. a set of 2^32 distinct colors (=4 294 967 296).
The display area is 331 by 207 mm (390.4 mm diagonal, i.e. a 15.4″ DFP). 331mm is 13 inches, 1280 dots on those makes 98 dots per inch. 800 dots over 207mm makes 800/(207/25.4) = 98 DPI – so the pixels are square and displays at a resolution of 98 DPI in 32 bit color.
Printers have a similar resolution, with an exception; most of them print just ONE color in that dot (out of the available choices of color-cartridges). If there is three cartridges any of three colors can appear in that dot (in addition to the unprinted dot = paper color). To achieve an impression of finely blended colors – impressive color patterns are printed at resolutions of 1440 DPI or more. To achieve “240 DPI”, e.g. 1440/240 = 6 dots in height and width (36 dots!) may be used with a pattern of colors selected out of the cartridges.
Now note that “DPI” has nothing to with color; just how fine dots there can be produced. A certain amount of those dots can appear within one inch.
But now; how does this apply to images then? The answer is; quite loosely!
Lets have an example: a modest 6MP camera might produce images that have 3000 by 2000 pixels (e.g Nikon D80).
If I print that on the usual 6 by 4 inches (15 by 10 cm) this means I have 3000 dots over those 6 inches; 3000/6 = 500 DPI (same as; 2000/4). If the printing device DOES print at 500 DPI in full color! (go back and read above if you wonder!).
In most cases though, the printing device has a lower actual printing DPI. Then a piece of software has RESIZED your image to a DPI SUITABLE for the device; it might only handle a set of e.g. 150, 300, 600 or 1200.
If you printed the image from Photoshop (or anything); it was first handled by Photoshop, then by Windows (GDI?), the Windows printer driver, handed over to the printer itself (a hardware driver, which might be doing scaling and rasterization)
What is a nice level of print resolution then? Here is some text on that.
NOW NOTE:
The resolution given in the image clearly isn’t necessary!
Can you think of a reason to have it there?
With regard to Nyquist: Have a lok on the Wikipedia article here.
If you don’t understand what is written, just fast-forward to Fig.5 and Fig.6 below “Application to multivariable signals and images“, where Fig.5 shows what is explained at “For example, a digital photograph of a striped shirt …” in the text (fully understandable for anybody I hope, to the right of Fig.6).
Then try to understand what the sinusoidal curves on the image under “Critical frequency” means:
The black dots are “samples” of the red curve.
As you can see the blue AND green curves ALSO match the samples – even though they’re NOT the same as the red curve. This is because the “sampling” has too few dots (i.e. has too low “frequency”). If there would be more evenly spaced samples taken (more black dots!) the other curves wouldn’t fit on the dots (samples).
This is what “Nyqvist” is all about, you need to sample at a high enough frequency (at least the double) to catch a correct “image” of the information (image, music, whatever) you’re trying to record.
This is the reason why standard CDs contain samples at a rate of 44.1kHz (each sample: 16 bits each for both channels in stereo) – to cover up to 22.05kHz of sound frequency (which is about what a person with GOOD hearing can hear).
An old analog telephone line covers frequencies from some low number (I’d rather not say zero here) up to 3 to 4kHz – which is one of the reasons old modems had hard to climb past 14.4kbps – until new encoding and technology was invented (which eventually took them up to 57.6kbps).
