Understanding DPI: The Solution to Resolution

Did I miss something here in this article about DPI/PPI? While I completely agree that there is a big misunderstanding about PPI and the false interchangeability with DPI as though they are one and the same, I am confused by some of the statements made in the article. For example, what is DPPI as made in the statement "What this means is that you don't have to set the DPPI to something specific, the printer will take care of it."? I've never heard the term "DPPI" before. I am also confused by the remark that "My camera sets DPI to (something) in the images". A camera does not ever set a DPI (as in Dots Per Inch for the laying down of ink on a substrate) as a camera is not a printer. A camera is a device for capturing an image on a sensor, converting it from analog information to digital. After some manipulation in image processing software, either in the camera or on your computer, the image is then handed off to the printer, still as a PPI image. The printer then converts that PPI information into dots of ink. Generally the dots of ink are so small in comparison to the size of the pixel the camera produces that each pixel of information has many dots of ink used to compose it. Depending on the spread of the ink (dot gain) on the paper, printing at DPI's above 1440 are a waste of printing time and ink.
I also take issue with the statement that "My scanner requires setting DPI". A scanner requires setting the Samples Per Inch (SPI), not the Dots Per Inch (DPI). Using the terms interchangeably is wrong as they are entirely different things. A typical flatbed scanner is like the sensor in your camera. It has divisions in it as sample points. I'm not sure what the actual number of sample points is, but I believe 600 per inch (the equivalent of 600 PPI in a camera sensor) is about the max. Many scanners boast higher output resolutions, but that is usually achieved by interpolation rather than there actually being that many sample points in the scanner array.
The resolution of an image is based on the physical size of the sensor and the number of capture points it is divided into, ie Pixels Per Inch, ignoring the potential for interpolation by the camera to a higher resolution. Some cameras may allow you to hook up directly to your printer and print something straight from the camera, thus the possible reference to DPI, but I can't imagine anybody doing that, especially those who read this forum. There's not an image in the world that doesn't require some sort of correction, be it color, tonality, sharpness, etc. Somehow I don't think that people who join photography forums would ever do a direct print from a camera straight to the printer.
I certainly do not know everything there is to know about digital photography, perhaps far too little, but I find this article to be more confusing than it is instructive. No offense to the person who wrote it, just that it doesn't make much sense to me.
pjcaver

The problem is that cameras, scanners, and computer software use the term "DPI" when they mean something else. Because of this, "DPI" is in common use as a field in a variety of locations. The goal with this article is to help people understand that whatever is in the "per inch" setting is not relevant unless scanning or printing.

What Peter Jones indicates about SPI (as also stated by David Blatner "Real World Scanning & Halftones" book) is pretty much how I understood input resolution, but with scanners only.
Digital camera's sensor size in inches is what's important to know in relation to the size of the detail (hair, tree foliage, etc.) sampled on the sensor. That is known over scanners due to the fact you can't measure the sensor array unless it's stated in the scanner's specs.
A 1 inch wide DSLR sensor with 3000 pixels or any given amount of pixels across on the long end will tell you how many pixels will make up those tiny bits of detail that allow you to add a certain amount of contrast and sharpening in order to overcome demosiacing/anti-aliasing for it to not look like a cartoon or vector graphic when noise reduction is factored in.
A very informative article by Jeff nonetheless.

I'm somewhat confused too, in particular as I have been printing (through Photoshop cs5) all sized pictures on an Epson 3880. Needless to say the images come from raw files and are plenty large and loaded with details. Only recently did I resize some of them to the size of the prints I was making. And frankly, I noticed no difference/just the same, excellent quality.
It would appear Photoshop, as the article mentioned, is taking care of the DPI. Am I correct???

The confusion can be alleviated with your own eyes viewing a test print of a small crop of the image to check strairstepped edges on fine detail like hair and hard edges.
Frankly I wouldn't worry about it if you're getting good print results. DPI? PPI? SPI? At this point in the advancement of current technology you can rely on the preview of the image on your display viewed at 100% to tell you if it's going to work.

Total gobbledegook I'm afraid - having read it three times I'm none the wiser. Yet the article says "you know what you need to know"! Which I assume is nothing? Maybe it doesn't matter? Sorry Jeff, clearly I missed something here!

What do you need to know:

1) The setting is irrelevant if you are not printing.
2) The setting may be relevant if you are printing.

It's really simple, but if I just said that, nobody would get it. This was read by a number of people prior to publishing, so I'm not sure why the message isn't getting through.

I tried to accommodate to the article's style and make the most out of it, but I still, after reading it twice, can't get much meaninful info from it.
To the author: I kindly ask you to rewrite it. I did get the point that PPI and DPI are irrelevant while dealing with the electronic image. But the statement "it might be relevant when printing" does not seem clearly explained. I believe/trust you, but I don't understand you. And a point that stayed open: what about at the scanner?
Thanks for your efforts, though, albeit with less than great results, for me anyway.

Great article Jeff. I've been explaining theses differences to people I work with for years. And still I have to deal with it even with Art Directors in arvertising agencies.
Pixels are the only things we need to know.
Even if DPI is useful in printing I still like to think with pixels.

I like the fact that you've tried to explode the myth of dpi - it's generally not relevant to photographers until they need to print and is really a great [and continued] source of confusion.
That being said, by trying to cover every area where dpi [etc.] is used or possibly a factor [e.g. on screen] I think you've made the issue more complex, not less, as you're trying to a] explain resolution and b] why dpi is not often relevant but the text starts with dpi and then goes on to resolution
I would have focused on the fact that images have certain intrinsic properties - like height and width, and that the resolution in which they are reproduced varies depending on the properties of the device that is used to reproduce the image, be it a screen, an iPad, or a printer, rather than on the properties of the image itself.

Just had a sort of resolution revelation yesterday examining detail captured in a 113MB high quality dpreview.com sample jpeg from a Pentax 645D MF digital camera with 1.75inX1.25in./7264x5440 pixel count sensor. Go figure how many PPI that comes to.
Didn't see any more refined detail than I've seen off a regular 8-14MP APS-C DSLR capture. And this was examining a profile of a child's headshot that nearly filled the entire frame meaning the subject was quite close to the lens.
I saw a lot of very fine dithered noise that didn't add to the amount of perceived detail. It just seemed like a waist of HD space for such a huge image.

No disrespect to the author but I am confused as well. Even more than before. My understanding is as follows. The camera depending on the size of the chip and file's setting will determine the size of the final image and as well, how many pixels fit in that space. All data contained in those pixels cannot be change. However, through the proper interpolation, the image size can be increased, mostly for printing purpose and in order to achieve the best possible print, the DPI resolution should be matched to the printer needs. When scanning, the DPI setting will determine the final image size and the larger (within limitations) the DPI, the smaller detail will be achieved. This makes me to suspect that the number of dots needed to make a print - increase in number - by a higher DPI setting. However, I've done some experimentation with small captured files that need some serious and delicate editing. For the most part, a few pixel edits, such as 3 to 5 pixels. I increase the size of the image in terms of pixels and DPI together in Fractals and when I work in that file, it is possible to work much smaller areas compared to the original size. For the most part, I find Fractals tends to avoid bleeding and contrast from one color to the other one, when the resizing is not exaggerated. The edges tend to remain relatively sharp for as long as no large compression has been applied in the original file. I've edited images of 800 pixels, enlarged it to 2200 pixels and after blending some edges (in some colors only) and a softening ot in Neto noise remover, I get a flawless photo quality final file. If the softening was large, I pass the file again through Fractals but adding film grain (usually set to 21). I've made very large prints from those files that way. Original DPI was 72 and I increased it to between 180 to 300 DPI. Usually my working file is in PSD format and the final file is JPG and save it in the highest setting <12>.

I will look at some editing to make it more clear. However, the issue can be summarized very succinctly:

Whatever is in the box labeled DPI or PPI is irrelevant for a digital image until the time you create a file to print. Even then, it may be irrelevant - most outside printers ignore the setting and use the file size. In addition, it's worth understanding that image file resolution is x pixels by y pixels and has nothing to do with DPI or PPI. There are no inches in a digital file.

True resolution is measured in lines per mm. Their are too many variables in DPI....some dots are square....some are round....some are rectangle. That's why in the medical field...it's measured in lines/mm.

You should have a clear understanding of resolution before writing a solution. Film is about 400 lines/mm...that's why you can blow it up big. The best lens produces about 75-100 lines/mm....digital cameras...the sensors....CCD, CMOS, JFET....produce about 10-20...this is were it gets good.

The human eye...can only distinguish about 5-8 lines/mm. resolution comes into play when you blow it up.....and i mean way up.

Dave

This is about digital files. They have resolution measured in pixels in each direction, and that's it.

DPI is only valid when it is used to express the number of ink dots per inch that a printer sprays out. PPI should be used for everything else, but since a small printed pixel looks like a dot people call it a dot and use DPI instead of the correct PPI. I really dislike DPI used interchangeably with PPI. It makes any math used less intuitive and confuses people.

In math you can't mix apple and oranges; when doing calculations all labels must be the same. Oh yeah, sez who?

I have 48 apples that I want to put into boxes that hold 4 oranges each. How many boxes do I need? Simple:

48 divided by 4 = 12 boxes.

Since an apple is about the same size as an orange, you get the right answer but the question looks confusing.

I have 48 apples that I want to put into boxes that hold 4 apples each. How many boxes do I need? No one would get confused about that.

A. If a digital image is 3000 pixels long and I print it at 300 dots per inch, how many inches long will the print be?

B. If a digital image is 3000 pixels long and I print it at 300 pixels per inch, how many inches long will the print be?

Any newcomer not realizing that some people call printed pixels "dots" will be confuse by question A. but could readily grasp question B.

Solving B. 3000 pixels divided by 300 pixels/inch
Invert numerator and denominator of the divisor, pixel labels cancel out and you end up with 10 inches.

Solving A. 3000 pixels divided by 300 dots/ inch
Invert numerator and denominator of the divisor, no labels cancel out and you end up with 10 pixel inches per dot. Now that is confusing.

Mr. Spirer had to use the term DPI in some of his comments because that is the term that manufacturers and editing software often use albeit incorrectly. I'm sure he doesn't like it any better than I do.

What would be helpful here is some guidance as to the best/optimal print output. If I'm looking to make a 24 by 36 inch museum quality print, what sort of input PPI and output DPI would I need?

That's a very different topic that depends on a number of factors, including ink, paper, viewing distance, framing, etc.

Some comments were added here:
http://photo.net/digital-darkroom-forum/00ZoMs

its probably easier to learn russian than it is to figure out this ppi/dpi thing, just impossibly confusing !

Da tovarich

First of all I'm disappointed in the fact that the statement about a file size like "I have file that is 300 DPI high res file"" was not addressed here... this is the most common bad use of "DPI" in my dealing with people and files. I have to tell them there is no such size, it is meaningless with out a designation of Hight X Width--you just have a density of dots other wise-that file could be .2 inch by .2 inch at 400 DPI or it could be 10"X10" at 400 DPI and would be completely different size files, suitable for two completely different applications. I also do not agree that digital files do not have sizes… my monitor has a physical space and a physical resolution and so my digital file takes on those characteristic as it is displayed and how the setting in my image browsing software also affect how the file is displayed to me and how it is defined as I work with it electronicall--do to me it dose have H X W in digital existence.

Tom, when you output your digital file to anything that will allow us humans to see the data in it (a printer, a monitor, an iPhone) it leaves the realm of the digital and becomes part of our physical world, modifying reflected light (print) or modifying transmitted light (monitor). And yes, then it takes on physical dimensions. But until then, the only dimensions an image file has are its pixel dimensions (x pixels high by y pixels wide) and its storage size (measured in bytes).
Have a read again of what Dave Fitch had to say above: "the resolution in which (images are) reproduced varies depending on the properties of the device that is used to reproduce the image, be it a screen, an iPad, or a printer, rather than on the properties of the image itself."