There are many many erroneous statments out there that try to "explain" ppi/dpi/screen rez/print size/etc.
Just study up at Wayne’s. I know of no better all around education on the subject matter:
http://scantips.com And even Wayne will mix ppi/dpi once in a while.
The prob is that even mainstream vendors now refer to dpi quite often, even when talking about images. Most scanner apps do. Which just helps confuse the issue, especially for folks new to the concepts.
Even image editors use the term "wrongly". Adobe is one of the few that still use "ppi" for images.
And no, monitors do NOT have an inherent "dpi" (or ppi). They may be said to have a "working ppi" depending on the physical size of the monitor itself and the pixel dimensions it is set for, but this is not a native specification of the monitor.
This concept is further blurred by the fact that most LCD monitors have a "native" pixel dimension resolution at which they look best, while CRT monitors can manage a wide variety of screen resolutions at which they look fine.
Mac
In addition to what Mac said, the main reason for suggesting 72ppi for screen use is that you usually need to choose some kind of resolution when getting your files into your computer. Usually a scanner, say, will only offer ppi/dpi as a choice for setting your file size, or an even-more meaningless dimension like inches.
While it doesn’t really matter (only pixel dimensions,as Mac said), if you choose 72 you are getting the smallest possible file size for storage purposes, and usually that will give you enough info for screen display.
Of course, the downside is that by scanning at 72 you are creating a file that won’t be much good for most other possible uses, like printing.
I did read all of Wayne Fulton’s very long "Say No to 72 dpi: Its a False Notion". But I am a nobody. One of the coauthors (Christopher Dahl) of the book I quoted worked for seven years for Adobe preparing instructional stuff. The other person (Craig Hoeschen) has been doing this kind of stuff for 25 years.
But let me see if I understand the situation. Is the following published claim false? "72 ppi is the most common image resolution for monitor displays."
Actually, they aren’t making a claim about monitor resolutions. They are making a claim about _image resolution_ "for monitor displays". But do we care what the most common resolution is for images on monitors? Don’t we care about what the most fitting resolution is? Is 72 ppi the best choice. But when do we choose this? I can see that if you are scanning you have to decide a resolution. But normally, one would want to scan at the highest resolution possible, wouldn’t one? Photoshop Elements indicates that the resolution of the pictures I get from a standard Kodak CD is 256. Nothing I do in preparing the photo for posting on the Web changes that number. And if I don’t choose 72 ppi when I am preparing images scanned by others, why should I choose 72 ppi if I get a scanner. What is wrong with posting things that have the number 256 in the resolutions box?
As far a printing goes, dots per inch is fixed for my printer isn’t it. Is someone saying that it makes a difference what the number is in the ‘resolution’ box under resize image?
I get the impression that despite everything that Wayne Fulton or others say, there is still supposed to be some residual role for resolutions using the number 72. But why is it 72 and not 73? Why not 80? (I am talking about good reasons, not historical ones).
Other than possibly when scanning, does a person _ever_ have to use the ‘resolution’ box under Document Size in the Image Size thing from the Image menu?
Every Photoshop Elements book I have looked at (including Richard Lynch’s, I think), and some of the recommended Web sites, invokes the number 72. Are they all misguided?
Thanks for any help.
–Paul (Bullen)
Here is something Wayne Fulton says:
"Yes, I know that Adobe Elements or Photoshop menu _Save for Web_ is absolutely determined to scale the image to 72 dpi in _every_ case. I can’t think of any rational reason for it — the 72 dpi value will always be ignored. My guess is that is that it is only because there as still so many people who would think it is wrong otherwise." (p. 24 of 26)
This led me to do a Save for Web of a file for which the number in the resolution box is 256. Then I opened the file and checked the resolution box, and lo and behold it was now 72. (But contrary to Fulton, it is ppi, not dpi.)
But there is no indication of this on Save for Web. This gives rise to a few questions. (1) What does Fulton mean by saying that "the 72 dpi value will always be ignored"? (2) Are we unable to choose any resolution other than 72 ppi, if we use Save for Web? If so, is there a good way to save for the Web other than using Save for Web? If this is all ignored, as Fulton says, then none of this matters. But if it is not ignored, is 72 ppi unquestionably and always the optimum dpi for image to be viewed on computer monitors?
Thanks. –Paul (Bullen)
Paul: This ppi thing has definitely got you by the throat! 🙂
With respect to web and e-mail images, think about the screen which will display them. In most cases, monitors are set to either 1024 pixels by 768 pixels or 800 pixels by 600 pixels – regardless of whether it’s a 15-inch monitor or 21-inch monitor. At a given monitor setting (let’s take 1024×768) the ppi of the 15 inch monitor is less than the ppi of the 21 inch monitor – but an e-mailed or web-based picture that’s 1024×768 will fill both. What I do when preparing an image that I know will be displayed on the screen is start with my camera image – which is 2272 pixels by 1704 pixels – and downsize the pixel dimensions to 800×600; I know that an image of those pixel dimensions will fit comfortably on most monitors. In doing that downsizing, I NEVER look at ppi – it has no meaning, zero, nada. Two ways to get 800×600: Save for Web, which lets you put the desired pixel dimensions in an area of the dropdown in the lower right; and Image<Resize<Image Size with Resample box checked, where you simply put the new dimensions in the Pixel Dimensions in the top section of the dropdown. In that case, there also will appear a document size in inches and a resolution in ppi; what do they mean? NOT MUCH! They just tell you that if you were to print an image at those inch dimensions, the resolution of the image sent to the printerwould be so many ppi – or if you sent an image to the printer at that ppi, it would have those inch dimensions. But they have no meaning with respect to web or e-mail.
The printer resolution vs. image resolution defies simple explanation; some good treatises from Richard Lynch and others were included in a long (and contentious) string from about a month ago. If you’re still curious about that and have a bunch of time and coffee, you might want to check out that thread.
:o)
Chuck
"72 ppi is the most common image resolution for monitor displays." (Hoeschen and Dahl, p. 36).
But do we care what the most common resolution is for images on monitors? So what if everyone is doing it. The notion of a ‘standard’ isn’t relevant in this situation, is it? Don’t we care about what the naturally most fitting resolution is? Is 72 ppi the best choice?
The authors go further a few pages later: "For on-screen viewing of Web images, 72 ppi is a standard and safe resolution." (p. 45)
So, they go beyond saying "everyone does it" to saying it is "safe". Is 100 ppi less safe? How about 75 ppi?
–Paul (Bullen
So is the "resolution" box under Document Size in Image Size thing from Resize on the Image menu the place where we decide how much ‘information’ to send our printer when printing a photo?
And is it the case that aforementioned resolution box has no other function, whatsoever? It has no practical relevance to preparing images for the Web?
With respect to printing, however, I take it that the number in the resolution box does matter. Depending on the printer, the paper, the picture, and the purpose, different numbers will be appropriate. And they will range from 50 to 350? In this case, there really are some ‘safe’ numbers, in the sense that given the difficulty of determining the best number under each circumstance, there are some numbers for some general classes of circumstances that will get us dependably between 80 and 100% of optimality?
–Paul (Bullen)
Paul, yes, that’s right – the resolution in the Document Size does indeed matter when printing. For virtually any printing that I can think of, you want a minimum of 150 ppi – much less than that and you’ll get to see your pixels as little squares or circles. I shoot for 200 to 300 ppi as my 80-100, although if I’m making a small print and the resolution is higher than 300, I don’t downsample to get rid of the extra pixels; the printer will either use them well or ignore them.
Chuck
I shoot for 200 to 300 ppi as my 80-100, although if I’m making a small print and the resolution is higher than 300, I don’t downsample to get rid of the extra pixels; the printer will either use them well or ignore them. — Chuck
Chuck,
(1) What does "my 80-100" refer to?
(2) Why does the print’s being small cause you not to downsample? Is downsampling done to speed up printing?
Thanks,
–Paul
From what I can tell, changing the number in the resolution box does not affect pixels per inch. It just makes the picture smaller. If you really were changing just the resolution, shouldn’t the picture stay the same size, but just have fewer pixels? This would be achieved by making the pixels bigger (it could theoretically be acheived putting a black space around pixels–like molecules in outer space). So even though images say indicate their Resolution to be 72 ppi after they have gone through Save for Web, there has actually been no resolution change whatsoever. The number in the Resolution box only affects printing. And since pretty much no one would want to print something at 72 ppi, the only reason that number ends up being 72 after Save for Web is because there are a lot of people out there who will (mistakenly) think there is something wrong if they don’t see a 72. If it weren’t for the power of the myth of 72, that number would be left alone. Why? Because then if someone decided he wanted to print the image, it would print well. But if I am right (and it would odd if I am), I can manually change the number back to whatever it was before–or lower. I could make it higher, but there would likely be a noticeable loss of quality. Is this right?
Paul (Bullen)
P.S.
I just changed an image to a Resolution of 1 ppi. It is not one pixel per inch on my screen (there might be about 50 in an inch)–which would be consistent with what I said above, I guess.
Paul,
Here’s how I understand things:
Monitors don’t display an image differently based on the ppi setting. All that is important for displaying an image on a monitor is the number of pixels.
If your image is 600×400 px, it will require 600×400 px of your screen "real estate" to display it. Depending on how you have your monitor configured, your pixels may be larger (i.e. 640×480) or smaller (i.e. 1024×768). In either case, the number of pixels required to display the image remains the same. The image will appear larger or smaller on screen, correlating with display resolution.
When you send the image to a printer, NOW ppi/dpi becomes important. It tells the printer how many of your image’s pixels to cram into each linear unit. Taking the hypothetical 600×400 image, if ppi/dpi is set at 200, then the image will print out at 3×2 inches. Set the ppi/dpi at 100, and you get a 6×4 printout.
Pixel dimensions (height-x-width) determine on-screen image size, taking into account the monitor’s resolution setting.
Ppi/dpi determine printed image size, taking into account the image’s pixel dimensions (H x W).
Changing the ppi of an image only changes the instructions sent to the printer and, thus, the printed image size.
Changing the number of pixels changes the monitor display, and the printed image.
Hope I haven’t muddied the waters any…
Byron
Paul, sorry about the shorthand; the ’80-100′ was referring to your note where you talked about your 80% point. I’ll look for 200-300 ppi virtually all the time.
The print being small means that there are plenty of pixels in my original image to allow me to exceed 300 ppi on the print. Here’s an example: Original image = 2272 by 1704 pixels straight out of the camera Desired print size = 6 inches by 4.5 inches
Resultant resolution = 378.67 ppi (2272/6=378.67, 1704/4.5=378.67
On the other hand, if I print the same image at 10 inches by 7.5 inches, the resultant resolution is 227 ppi (2272/10 or 1704/7.5), close to the low end of the range I use for printing.
Let’s do one more example. Let’s say instead of my camera, I had used my scanner. Now ppi (or dpi, as most scanners erroneously call it) comes into play on both ends. Suppose I have a 4 inch by 5 inch original and I scan it at 300 ppi/dpi. I open it in Elements, and now it’s a 1200 pixels by 1500 pixels image (4 inches x300 ppi=1200 pixels, 5×300=1200). Suppose now I want to print it as an 8×10. I go into Image<Resize<Image Size and with Resample unchecked, I change the document size to 8 inches by 10 inches. What happens to the resolution now? The Document Size dropdown does the math for you but it’s going to be 150 ppi. (1200 pixels/8 inches=150 ppi, 1500/10=150). The resultant 150 ppi is very low for printing; now I see I should have scanned at 600 ppi.
Chuck
Paul, expanding on this topic further, changing the ppi resolution in the Image<Resize<Image Size box does different things depending on whether resampling is checked or unchecked. Let’s take the simple side first – resampling unchecked.
With resampling unchecked, you’ll see that the pixel dimensions boxes are ‘grayed out’; the pixel dimensions of the image are locked. Under those circumstances, when you change ppi, the document size in inches changes – increase ppi and the document size decreases; decrease ppi and the document size increases. With resample checked, you are not affecting what will be seen on the screen at all – that’s the pixel dimensions, which are fixed. You are changing the document size and print quality; you can either print a small picture of very high quality (ppi) or a very large picture of very poor quality.
When you check Resample, the Elements engine goes into gear to give you pixels that your camera never saw. Suppose we start with an image that has the following description:
Pixel dimensions: 1500 by 1200
Document size: 5 inches by 4 inches
Resolution: 300 ppi
Let’s say now that I want to bump up the resolution to 400 ppi. Taking care that the resample image box is checked, I put in 400 ppi. Look at what happens to the numbers:
Pixel dimensions: 2000 by 1600
Document size: 5 inches by 4 inches
Resolution: 400 ppi
Note that the pixel dimensions have increased. Where did the extra pixels come from?? The camera only captured 1500 by 1200; the program calculated the rest based on some mathematical algorithm. It spread out the original pixels and put some new ones in between to take up the space. This is referred to as ‘upsampling’, and as it’s based on math and not on real image capture, it’s likely to degrade the printed image somewhat and is usually avoided. In our example, the printed image would still be 5 inches by 4 inches, but now the printer would be delivered 400 ppi instead of 300 ppi.
And what about the screen image of the one we increased to 400 ppi with resampling checked? Well, it didn’t fit most screens to begin with at 1500×1200, but now it’s even bigger: 2000×1600 and will need to be scrolled both ways even more.
Suppose we were to start over with that 1500 by 1200 image and get it down to a size that would fit on most anyone’s screen – let’s say a picture with a maximum width of 800 pixels will do that for us. In Image Size, we see the following to start:
Pixel dimensions: 1500 by 1200
Document size: 5 inches by 4 inches
Resolution: 300 ppi
Now, I make sure that Resample Image is checked, because I want to change the pixel dimensions. In this case I want to get rid of pixels, i.e. downsample. I simply change the 1500 pixels to 800 pixels; the 1200 will change automatically to 638 pixels (forgot to mention: Constrain Proportions should always be checked!)
Now the dialog box should look like this:
Pixel dimensions: 800 x 638
Document size: 2.67 inches by 2.13 inches
Resolution: 300 ppi
In this case you have ‘downsampled’, i.e., thrown away pixels. The image will fit nicely on the screen, but it sure will be a tiny print.
That’s enough for now – going back to bed!
Hope that helps…
Chuck
In my opinion, a discussion like this is almost useless. There have been some very good technical discussions presented here, but my eyes just glaze over when I try to understand everything that is being said. I think what everyone should do is experiment until they find a routine that works for them, and then use that routine as a base to begin from. I remember back when I shot a lot of film everyone would just sit down and enjoyed looking at photographs when I decided to share them. But now, oh my! What kind of a camera and you have? How many megapixels does it take? What is your printer? What kind of paper do you use? What DPI did you print? And after all those questions are answered, the photograph is just set aside and forgotten. So what is the point of all this? All I want to do is share my pictures. Yes, I have a routine that works for me. But if I were to outline it for you all I’m sure I would have countless people tell me how wrong I am. But I don’t care. I have found something, through EXPERIMENTING, that works for me. And I use it as the base for every photo project that I start.
Joining the string rather late, so this may have been covered.
The reason 72ppi was chosen as a standard for monitor resolution goes back to the print and typesetting origins of desktop publishing. In print there are 12 points in a pica and 6 picas in an inch. Therefore a foot has 72 picas and an inch is 72 points. By using 72ppi for monitor resolution you maintain this relationship and a 72 point font is approximately 1 inch tall on screen for a capitol letter like an X or an M. This is actually an approximation and will vary with different font families, and even the 72 points to an inch is not exact due to changes in standard measurements from the original French system, but it is a fair guideline.
Rich
Well, perhaps that has some basis in fact (the measurements are certainly right) but of course CRT technology wasn’t based on a font size. It wasn’t as if early on they had the capability of making monitors capable of showing the same number of pixels as can be acheived today. Even once video adaptors could achieve higher pixel dimensions, some monitors were limited.
Early on when computer desktop publishing was taking off, a professional standard system (on Macintosh first) was a 15" monitor running 800×600. At this monitor size and rez, a picture will print at same size as seen onscreen at 72ppi (within a millimeter or so anyway).
That is however, an accident of the math, and further perpetuated the myth that "monitors have 72dpi". Anymore, of course, actually "working ppi" on a monitor may vary from 40 something to 90 something, depending on physical size of monitor and the pixel dimensions set for that size monitor.
Mac
Mac,
I wasn’t saying that CRT technology was based on this bit of history. What I’ve been told is that the Print connection is why that became a standard of choice for onscreen image resolution. And besides, it is an interesting bit of history. ;o)
Rich
Chuck,
Re post #13, it is very erudite and reinforces my understanding of the topic. Thanks. Ken
Thanks for the various comments, especially those that kept people awake. I have read them, but will need to reread them more carefully. Quantitative stuff doesn’t come easily for me, but I don’t think I can avoid it in this case. I have learned things already (e.g., I hadn’t thought about the different effects of changing the number in the Resolution box, depending on whether Resample Image is checked.) I would guess that for people who are good at math, statistics, and possibly playing chess, grasping all that has been said would be much faster affair. It is true that you can use Photoshop Elements to pretty good effect without learning these details, but when you want to get that extra 20% out of your photos, you got to be willing to invest an extra 80% of your time learning non-experimentally. Most people simply don’t have the time or interest to do that. But for those who do, I see know other way by the learn the ‘theory’. Someday, all this may be fruifully hidden from us (remember the stuff you had to know in the early days of the Internet). But for now, there doesn’t seem to be an easy way.
–Paul (Bullen)
Thanks for the various comments, especially those that kept people awake. I have read them, but will need to reread them more carefully. Quantitative stuff doesn’t come easily for me, but I don’t think I can avoid it in this case. I have learned things already (e.g., I hadn’t thought about the different effects of changing the number in the Resolution box, depending on whether Resample Image is checked.) I would guess that for people who are good at math, statistics, and possibly playing chess, grasping all that has been said would be much faster affair. It is true that you can use Photoshop Elements to pretty good effect without learning these details, but when you want to get that extra 20% out of your photos, you got to be willing to invest an extra 80% of your time learning non-experimentally. Most people simply don’t have the time or interest to do that. But for those who do, I see no other way than by learning the ‘theory’. Someday, all this may be thankfully hidden from us (remember the stuff you had to know in the early days of the Internet). But for now, there doesn’t seem to be an easy way.
–Paul (Bullen)
Jim Hess, I understand what you’re saying. Knowledge can be gained in many ways, and experimentation is one of the most exciting and rewarding ways. On the other hand, having a little grounding in theory may make the experiment go better or faster or cheaper – if that’s desired. So for those who want the theory and the math, we offer it here to be taken or left.
Chuck
