Talk:Dots per inch
From Wikipedia, the free encyclopedia
Contents |
[edit] British vs. American spelling
Since inches are now only an American unit* the spellings on this page should be in American English. Similarly, a page on the British monarch or Australian government would use British spellings, and a neutral article (i.e., one on molecular biology) could use either or both spellings. SteveSims 04:39, 5 January 2007 (UTC)
*Excluding a few non-English speaking countries.
[edit] Merges
I hadn't noticed before that a separate article existed for DPI until Bobblewik helpfully pointed it out. I have arrogantly decided to simply redirect DPI to this article, wiping out the previous contents of that article. Here is why:
- In a general sense, there was no information in that article that is not already in this one.
- Several factoids from that article are either vague or misleading, and are more adequately explained in this article or in related articles Pixels per inch and samples per inch. Specifically:
- DPI always refers to a physical representation of pixels per length unit. - Not true. DPI most correctly refers to printing resolution - dabs of ink on paper. The term is often used to specify what would more accurately be called pixels per inch; this distinction is explained in the current article.
- Only when outputting this image to a physical medium with a certain size (say printing on to paper 20cm by 15cm) does the DPI get defined. - This is confusing, subtly wrong, and partially contradictory to the previously-mentioned sentence that DPI always refers to pixels.
- The resulting DPI depend both on the resolution of the image... -- Not true. Even if DPI is used broadly, encompassing printer and monitor output, it still refers to a physical characteristic of an output device. The resolution of a particular image displayed on that output device has no bearing on that device's DPI capability. And DPI doesn't really make any sense in describing an image with such-and-such number of pixels; the "DPI" of an image only makes sense when given some number of inches. You can print a 5x5-pixel image on a 1200dpi printer; the output is 1200dpi. Printing a 1000x1000-pixel image does not change that.
Of course, the history is preserved, if anyone wants to extract any useful nuggets of information from it. -- Wapcaplet 20:57, 12 Aug 2004 (UTC)
[edit] Removal
From an editor: Draw a 1-inch black line on a sheet of paper and scan it. If the resulting image shows a black line with a width of, say, 300 pixels, then does the scanner not capture at 300 PPI? 128.83.144.239
I moved the above anon comment from the article to here. It was posted by way of explaining the setence "A digital image captured by a scanner or digital camera has no inherent "DPI" resolution until it comes time to print the image..." I'm not sure how it helps to explain this point, however; indeed, it seems to stem from the very misunderstanding of DPI that is being explained (that DPI and PPI are not the same thing, and SPI is yet another thing entirely). -- Wapcaplet 23:42, 23 Sep 2004 (UTC)
[edit] DPI versus PPI
DPI is mostly used to to tell how big a resolution an image should be printed at. Of course, this should be really be "pixels per inch"! But since it has become a standard, I think it should be explained here.--Kasper Hviid 18:33, 9 Nov 2004 (UTC)
- I am not sure what you mean; the details of what it means to print an image at a certain DPI are, I think, fairly well-explained in the article at present. DPI is all about printing; there is a separate article on the related but different term pixels per inch. In printing an image, three things influence the output quality: DPI (the physical capability of the printer), the number of pixels in the image, and the space in which it is to be printed. As far as I know, there is no term to describe "the number of pixels printed in a one-inch space on the paper," though pixels per inch is probably the most appropriate. The resolution of an image sent to the printer (that is, the number of pixels) is unrelated to DPI of the printer. Maybe this should be explained better in the article? -- Wapcaplet 00:12, 10 Nov 2004 (UTC)
- I has always throught of DPI as "How many pixels should be printed per inch"? This is a wrong, but common use of the word. As you said, there is no term to refer to "the number of pixels printed in a one-inch space on the paper", but this is probably why the word dpi has been used instead. For instance, at www.lexmark.com they tell that "resolution is measured in dpi (dots per inch) which is the number of pixels a device can fit into an inch of space." And at www.olympus-europa.com, they tell that a 640 x 480 pixels image at 150 dpi will end up as 10.84 x 8.13 cm in the print. Since this has become a commonly used standard, it deserves to be accepted in the article as a official use of the word dpi, along with a note that this really is wrong use of the word. The "pixels per inch" article dont tell anything about pixels per inch in print, but only about the screens resolution, something I have never understood the point of. Kasper Hviid 09:54, 10 Nov 2004 (UTC)
I wasn't able to find the references you gave on lexmark.com or olympus-europa.com, but it doesn't surprise me that DPI would be used in this broad way in documentation intended for the general consumer. I've seen scanner software that uses DPI to mean samples per inch. "Dots" is a fairly general term for most people; pixels, color samples, and ink spots could all be "dots" to most people. I don't think there's any call for distinguishing an "official" use of the word. It's official if people use it that way, and (I suppose) if it's defined that way in the dictionary, which it is. It's really only the more technical among us who care to differentiate DPI from PPI and SPI.
As for the purpose of describing screen resolution, I suppose it's probably useful in calibrating a computer display to a printing device. If a print shop needs to have things displayed on their computer monitor at the same size they will be printed, monitor PPI is useful. I thought about adding to the pixels per inch article to include the idea of pixel density on paper, but while it makes sense to me, I know of no other instances of PPI being used in that way. If you find such a usage, let me know! -- Wapcaplet 22:59, 10 Nov 2004 (UTC)
- Since pixels are actually dots (PDF), it comes out that PPI and DPI are actually the same thing. --88.153.32.35 12:50, 24 June 2006 (UTC)
---
Yes, since dpi and ppi are obviously used interchangeably, this interchangeability should be explained here... so that novices looking here for definitions can understand what they find to read in the real world. It seems extremely arrogant to say "Wrong" and "misuse" when obviously pixels per inch has always been called dpi. And still is.
Do some few say ppi? Yes. Do the vast majority say dpi. Absolutely yes.
1. All scanner ratings are specified as dpi, obviously meaning pixels per inch. They dont say "samples per inch", they all say dpi, which we all know means pixels per inch. Scanners create pixels, not ink dots. Who are you to call every scanner manufacturer wrong?
2. All continuous tone printers (dye subs, Fuji Frontier class, etc) print pixels, and call their ratings dpi too (colored dots also called pixels). Who are you to call all these manufacturers wrong?
3. The most current JPG image file format specification claims to store image resolution in "dots per inch". The most current TIF file format specification claims to store image resolution in "dots per inch" They are referring to pixels... there are no ink dots in image files. Who are you to call these authors of the most common file format specifications wrong?
http://www.w3.org/Graphics/JPEG/jfif3.pdf (page 5)
http://partners.adobe.com/public/developer/en/tiff/TIFF6.pdf (page 38)
4. Google searches on 7/14/06 for
"72 dpi" 17,200,000 links
"72 ppi" 124,000 links
(138 times greater use of dpi... a couple of magnitudes more useage)
You may be aware that 72 dpi topics are never about printer ink dots.
When calling everyone else wrong, a wise man would reevaluate his own position. The Wikipedia author who claims misuse of dpi is obviously dead wrong. It is probably only his wishful thinking that the world OUGHT to be as he wishes it to be, but it is just his imagination, and this Wiki definition is definitely WRONG.
The two terms are obviously interchangeable. Wake up, look around, where have you been? Pixels per inch has ALWAYS been dpi. Yes, dpi does also have another use. So what? Almost every English word has multiple meanings and uses. However which term is best is not important here - this is certainly not the place to decree it (as attempted). Both terms are obviously used with the same meaning (pixels per inch) and that matter is long settled. Say it yourself whichever way you prefer to say it, but we obviously must understand it both ways. Because we see it everywhere both ways. So this both-ways phenomena needs to be explained in the definitions here. Without bias. About how the real world really is, not about how some author might dream it ought to be.
WHAT IS IMPORTANT is that beginners need to know the two terms are used interchangeably everywhere, with both terms meaning pixels per inch, simply so they can understand most of what they will find to read about the subject of imaging. There is no reason to confuse them even more by telling them everything they read is wrong. Wiki is wrong. The Wiki definition can only totally confuse them.
Beginners do need to know the two concept differences (your two definitions), but once the concepts are known, then the terms are almost arbitrary.. We could call them "thingies per inch". The context determines what it means (like all English words), and if the context is about images, dpi can only mean pixels per inch (ppi can mean that too). If the context is about printer ratings, then dpi can only mean ink dots per inch. 71.240.166.27 03:20, 14 July 2006 (UTC)
[edit] Printer advertisements
I see several printers advertised with "4800 x 1200 color dpi" and such. Is this some kind of industry conspiracy to redefine the term "dpi"? Or am I misunderstanding something? Example: [1] -- Anon
- Nope, sounds like the most appropriate and correct possible usage of DPI to me, assuming those figures are what the printer is actually capable of. Now, if a scanner is advertised with some DPI, then Samples per inch is what is actually meant. Many times a scanner is advertised with its interpolated sampling resolution, since that number is often much higher than the actual optical resolution; a good consumer scanner may only be able to capture 1600 samples per inch, but the samples are often scaled (either in the hardware or in the scanning software) to much higher resolution, such as 19,200, and of course "19,200 DPI" looks better in an advertisement than "1600 DPI." Whether printer manufacturers use a similar strategy, I don't know; I do know that the reason the two DPI figures are often different is that one of them is a horizontal resolution, determined by how finely the printing heads can be controlled, while the other is a vertical resolution, determined by how finely the paper feed roller can be controlled. Finally, if you see a digital camera advertised with some DPI, buy a different brand, since DPI has no meaning in that context unless they are referring to the quality that might be achieved in printing a digital photo at a certain size (and then, pixels per inch is probably more appropriate). -- Wapcaplet 18:11, 28 Nov 2004 (UTC)
[edit] Color
I am concerned about the following statement:
- This is due to the limited range of colors typically available on a printer: most color printers use only four colors of ink, while a video monitor can often produce several million colors. Each dot on a printer can be one of only four colors, while each pixel on a video monitor can be one of several million colors; printers must produce additional colors through a halftone or dithering process.
Computer displays work in a similar fashion to printers: they use a combination of different amounts of the primary colors (in this case, the additive primaries: red, green, and blue) to produce a wide range of visible colors. Most printers use the (subtractive) primaries and black in different combinations and patterns.—Kbolino 02:18, 10 February 2006 (UTC)
[edit] What is DPI dependent on?
"The DPI measurement of a printer is dependent upon several factors, including the method by which ink is applied, the quality of the printer components, and the quality of the ink and paper used."
This is not true, or at least is confusing.
The DPI in the printing direction is dependent on the head firing frequency and the linear print speed. The DPI in the advance direction (perpendicular to the printing direction) is dependent on the spacing of actuators (e.g. nozzles for inkjet) on a head, and the angle of the heads. Each of these can be multiplied by use of interleaving/"weaving" using multiple passes and/or multiple heads.
What the sentence above may have been trying to get at is that different print modes can use different firing frequencies, linear speeds, interleaving factors, etc., and the effect of ink and media settings in print drivers is often to change the print mode (possibly in addition to other software settings that don't affect DPI). Also, different print head technologies may improve at different rates in terms of firing frequency, actuator spacing, etc.
On the subject of advertisements, I strongly suspect that some of the dpi figures quoted in printer adverts are inflated. You can inflate a dpi figure by:
- counting different colors as more than one dot (this may be what "4800 x 1200 color dpi" means -- I expect it is really 1200 x 1200 in 4 colors)
- counting each dot printed by a variable-dot head as more than one dot
- saying "equivalent dpi" and making up a random number.
This kind of creative arithmetic is all the result of trying to munge various resolution and quality factors into a single number for marketing purposes. It's similar to how clock speed used to be used to indicate how "fast" a processor was. At its worst, it can lead to distorted technical decisions that maximize DPI with no improvement in, or even at the expense of quality (just as the Pentium 4 design was distorted to maximize clock speed).
It's unlikely that someone could see a visible improvement in resolution above about 1000 dpi with the unaided eye at a normal reading distance. The extra quality that you can get from a higher dpi than that is not due to an increase in resolution; it's due to a reduction in "graininess" (and possibly better hiding of head defects) from using smaller drop volumes, which requires you to use more dots in a given space to achieve the same ink density.
To meaningfully compare printers, you need at the very least to know the volume of ink in a drop (for inkjet heads as of 2006, this can vary from about 1 to 80 picolitres), including what "subdrop" volumes are possible in the case of variable-dot heads, as well as the real dpi figure in each direction. The overall quality will also depend on halftoning algorithms, the gamut of the inks used, color management, positioning accuracy of the printer mechanism and any encoders, head defects and how well the print mode hides them, etc. The intended application is also significant: to give an extreme example, there's no point in achieving "photographic" resolution in a printer that will be used to print billboards -- although color gamut would still be very important for the latter.
DavidHopwood 00:16, 5 June 2006 (UTC) (working for, but not speaking for, a printer manufacturer)
[edit] Metric
"There are some ongoing efforts to abandon the dpi in favor of the dot size given in micrometres (µm). This is however hindered by leading companies located in the USA, one of the few remaining countries to not use the metric system exclusively."
I wouldn't blame US companies for this, even though I'm an enthusiastic S.I. advocate. Software interfaces to RIP packages and driver APIs require dpi, and there's no compelling reason to change them. Despite this, it is possible for a printer controller implementation to be internally almost S.I.-only. DavidHopwood 01:20, 5 June 2006 (UTC)
