A Site for Photographers by Photographers

Featured Equipment Deals

Missing Pages: White Balance Read More

Missing Pages: White Balance

Jon Sienckiewicz offers a juiced-up User Guide for creative people via his "Missing Pages" column on Photo.net. This month covers the topic of White Balance.

Latest Equipment Articles

4 Outdoor & Adventure Photo Packs Read More

4 Outdoor & Adventure Photo Packs

Photo packs have come a long way in the past decade, especially those that are targeted toward outdoor and adventure photographers. Alaska-based adventure photographer Dan Bailey takes a closer look...

Latest Learning Articles

Getting Started in Video Read More

Getting Started in Video

Photographer Ted Kawalerski made the transition from still to motion and has never looked back. Ted takes you through the steps to get started in a medium that will open your photography business to...


Display, Printing, DPI and PPI

by Bob Atkins, 2004


One aspect of digital images which seems to cause a lot of confusion to beginners is the matter of image size. There are three basic measures of image size

  • Pixel count - e.g 3000x2000 pixels
  • Physical size - e.g. 4" x 6"
  • Resolution  - e.g. 72 pixels per inch (ppi)

The confusion seems to arise because people aren't sure of how these are related. They wonder if ppi affects the size of the image you see on a monitor screen. We know that displays are measured in pixels (e.g. a 1024x1280 display), and that screens are a given with (e.g. 15"), it seems logical that the pixels/inch setting should affect the size of the displayed image. Logical maybe, but wrong. Although monitors do have a measurable figure for ppi (pixels per inch), the ppi information in an image is NOT used for monitor display in web browsers.

Digital image files are "tagged" with other information. For example the width and height in pixels is in there, as is the resolution (ppi). Some image files also contain all sorts of information about the image, such as exposure data, focus data, flash data - and this is stored in what is referred to as the EXIF header. EXIF stands for Exchangeable Image File Format, and is a standard for storing interchange information in image files, especially those using JPEG compression. So how is this information used, and what uses it?

Pixels

When you display a digital image on a monitor using a web browser, the only thing that determines the size of the image is the width and height in pixels. All the other data contained in the image file including resolution (ppi data) is  ignored. If your image is a 480Kbyte file which is 800 pixels wide by 600 pixels wide, it will display as a full screen image if you are using an 800x600 display. It doesn't matter if your ppi is set to 1 or 1000  This is 100% true as far as web display goes and as far as any monitor display goes - unless some software intervenes. For example the IE6 browser may take large images and resize them so they fit on the screen. However ppi is still ignored. A few advanced page layout programs and advanced image editors are capable of taking ppi into account when displaying images.

So I'll say this once again. The way you control how large an image appears on someone's monitor screen when viewing your images on the web is by changing the height and width in pixels  If your original image is 1600x1200 pixels it will probably be too large to see all at once on 95% of the video monitors out there. It will also be slow to load since it will be a large file. If you want someone using an 800x600 display to be able to see your image clearly, you need to change the size to, say, 600x400 pixels (remember the browser window is smaller than the full monitor display). You change image size in software. All image editing programs can do this. It's called "resampling". If you make the image smaller it's often called  "downsampling" or "downsizing". if you make it larger it may be called "upsampling" or "upsizing". See your image processing software manual for details on what options your software offers.

For example here's a screen from PhotoShop (and PhotoShop Elements) which allows you to change the pixel width and height of the image.

ps4.gif (7692 bytes)

You can see here we are starting out with an image which is 1000 pixels wide and 640 pixels high. Let's assume we want people who still have 640x480 displays to be able to see all this image at one time. To do that we have to reduce the image size. Let's say we want to make it half as big, 500 pixels wide by 320 pixels high. What do we do? Well, we ignore what's in the document size box, since that ONLY affects the way the file will be printed. The data in the Pixel Dimensions box is all that counts as far as web display is concerned. So what we do is simply type "500" in the width field of the Pixel Dimensions box as shown below, with both the "constrain proportions" and "resample image" boxes checked

ps3.gif (7552 bytes)

Note that since we had the "Constrain Proportions" box checked, changing the width also changed the height in correct proportion. Clicking on "OK" will resize the image to 500 x 320 pixels using a "bicubic" resampling algorithm. The "bicubic" algorithm usually gives the best results but takes most time. There are alternative resampling algorithms called "bilinear" and "nearest neighbor" which are faster but which might not give as good results. On a small file like this, speed doesn't matter, but if you're working on a 180MB file, it might. Resampling is the name for the process which figures out how best to subtract (or add) pixels when the number of pixels in the image is changed.

Of course you can make images larger as well as smaller. You can take an original image which is 500 x 320 pixels and make it 2000 x 1280 pixels. The problem is that you have now created an image with an extra 1.92 million pixels which have to be made up by the software. The larger image will have no more real information in it than the original, so it won't look as sharp. If you must do this, the best way to do it is using bicubic resampling and to do it in small steps. By this I mean you may get a slightly better result by enlarging in 5 or 6 smaller steps than by enlarging in one big step. So for example you might change width (and height in proportion) in steps from 500 - 750 - 1000 -1500 -2000, rather than in one step from 500 - 2000. Sometimes you may see a small difference, sometimes you may not.

Pixels per Inch

You'll note that the width and height in the document size box have changed along with the number of pixels in the image. This makes sense because the resolution (pixels/inch) has not been changed. "Pixels per inch" and is almost exclusively used for printing, not video display. If you take an image that is 1000 pixels wide and 640 pixels high, and you print it with a ppi setting of 180 pixels per inch, the print will be 5.556 inches wide by 3.556 inches high. If you resample the image down to 500 pixels wide by 320 pixels wide and you keep the ppi setting at 180 ppi, the print (document size) will now be 2.778 inches wide by 1.778 inches high.

You can, of course, change the ppi setting. 180 is probably as low as you want to go for quality images. For the very highest quality you might want to go to 360 ppi as shown below.

ps5.gif (7202 bytes)

Note that the resample image box has been unchecked since we don't want to change the number of pixels in the image, just how those pixels are printed. Instead of printing 180 pixels to every inch, we now want to print 360 pixels to every inch. Obviously this will result in an image which is half the size (but with twice the linear pixel density). You will see the new print (document) size is now 1.389 inches wide by 0.889 inches high (pretty small!).

What happens if you only have a 500 x 320 file, but you want a 10" x 6.4" print? Well, you either have to set the resolution to 50 ppi, or you have to set the resolution to something like 200 ppi and upsample the image by a factor of 4 (to 2000 x 1280 pixels). Either way image quality will suffer, showing you can't make good large prints from small image files. There are some programs (e.g. Genuine Fractals) which claim to be able to upsize with less quality loss. Sometimes they give results which are a little better than using bicubic upsampling in PhotoShop, sometimes they don't.

Dots Per Inch

There's another term, dpi, which is associated with printers. DPI stands for "dots per inch" and is a property of a printer and printer driver software, not a digital image. It's a measure of how finely spaced the droplets of ink can be in a print. However the number is a bit misleading since it's not always measured in the way you think it might be! Printer settings of 360dpi, 720dpi, 1440dpi and 2880dpi are often found. However the difference between then is subtle at best. Most people probably couldn't tell the difference and 360dpi usually looks great. Changing DPI does not change the size of the print. ppi controls that. dpi controls print quality (though as I said, over 360dpi you typically don't see much change).

Resizing images for use on Photo.net

Photo.net requests that images attached to forum posts be at most 511 pixels wide, and that images uploaded to the image gallery be at most 800 pixels wide. This is done for speed, bandwidth, and ease of display. So what you'd do here is take the image from your digital camera, click on "Image -> Resize -> Image Size" (in PhotoShop Elements II), or "Image -> Image Size" (in PhotoShop 6). Then you check the "resample image"  and "constrain proportions" boxes and type in either "511" or "800" in the width box of the Pixel Dimensions section and hit "OK". Your image will now be resized and you can save it, preferably as a JPEG using the "File -< Save for Web" dialog. Ideally images should be under 100Kbytes in size and you can adjust this with the JPEG settings in the "Save for Web" dialog box

Right now photo.net will probably further resample your image, even if it's small enough to be within the requested size parameters. This is due to the way the existing software was originally designed. At some point in the future this may be changes so that no further changes are made to submitted images which are within the stated size requirements.

Note that there are some programs which create JPEG files which the photo.net software for attached images doesn't understand. If you attach such an image to a forum post it will display as a link, not as an image, even if it's under 511 pixels wide. One program known to do this is Micrografx Picture Publisher 8.0. There may well be others. The problem seems to be that the photo.net program incorrectly reads the file dimensions, so it doesn't know that it's 511 pixels wide or less. In that case it attaches the file as a link rather than attempting to display it.

What if you don't have PhotoShop?

Not everyone can afford PhotoShop - and indeed not everyone needs or wants it.ust about every image editor ever made allows you to adjust the size of images and just about every camera or scanner sold comes with one or other of them. They all have different menus so it's not possible to give definative instructions for every single program. If for some reason you don't have any image editing software, take a look at IrfanView ( www.irfanview.com). It's easy to use and free for non-commercial use.You can adjust image size via the "Image -> resize/resample" menus. GIMP is another alternative. It's a program originally developed for UNIX and ported to windows. It's not as easy to install or to use as IrfanView. If you're a novice, I'd recommend thinking twice about GIMP.


© Copyright 2004 Robert M. Atkins All Rights Reserved

Article created 2004

Readers' Comments


Add a comment



Lu Yin , January 19, 2004; 11:54 P.M.

For Windows XP users:

One easy way to resize picture is by using XP Powertoys that allows you to right-click on the image icon to resize. It's fast and easy. And it is free. It is available from Microsoft's website.

ranjeet utikar , January 20, 2004; 01:23 A.M.

Another option is powerful imagemagick software. For example to make a thumbnail (from the docs) use the convert program like ..

convert -size 120x120 cockatoo.jpg -resize 120x120 +profile "*" thumbnail.jpg

In this example, '-size 120x120' gives a hint to the JPEG decoder that the image is going to be downscaled to 120x120, allowing it to run faster by avoiding returning full-resolution images to ImageMagick for the subsequent resizing operation. The '-resize 120x120' specifies the desired dimensions of the output image. It will be scaled so its largest dimension is 120 pixels. The '+profile "*"' removes any ICM, EXIF, IPTC, or other profiles that might be present in the input and aren't needed in the thumbnail.

for more info check out http://www.imagemagick.org/

Chris Fuller , January 20, 2004; 01:52 A.M.

There are a variety of resizing tools available. The best known is Genuine Fractals (http://www.lizardtech.com/solutions/gf/). There is also PhotoZoom (formerly S-Spline, http://www.trulyphotomagic.com/shortcut/customer/home.php?cat=3&sub=serie.info} and Extensis pxl SmartScale (http://www.extensis.com/pxlsmartscale/index.html?ref=hp). Many people use Fred Miranda's Stair Inerpolation action or plugin (http://www.fredmiranda.com/software/). I do all of my printing through Qimage (http://www.ddisoftware.com/qimage/) which resizes the print as it spools it to the printer. Therefore, I do not have to resize the file before printing (great for file storage). Qimage comes with a host of other features. Some people do not like its interface. But it is only $44.95 and it comes with free upgrades for life. I recently compared each of these methods by resizing an image to 13x19. I cropped a 5x7 segment and printed a sample of each. I found Qimage to offer the best resizing of the five options. Chris Fuller

Qiang Lin , January 21, 2004; 01:26 P.M.

Another tool MS Photo Edit, which comes with MS-Office, can be used to resize the images too, although it's not a great tool. One comment regarding dpi for the printers, it's dots of inks, not the pixles of the images. It usually takes a lot of dots to get one pixle with perticular tone. Therefore, you do see better result if you use photo paper and higher dpi setting for the printer. However, this doesn't mean you are getting higher pxi in your print.

Walter Alberto Aprile , January 21, 2004; 04:35 P.M.

Another important operation besides resizing is image rotation, typically done just after pulling the images off your digicam, prior to archiving them. The JPEG format can be rotated without decompressing it, which means that the operation is fast, and there is no quality loss.

But not all programs are capable of doing that: many will decompress the JPEG to whatever internal format they use, rotate it, and then recompress it. One program that can do lossless JPEG rotation is IrfanView (the quick access key is J).

James Burger , January 22, 2004; 01:04 P.M.

Another way to resize an image in Photoshop, which will help when sizing a series of images uniformly, is to use the crop tool. Select the crop tool and then, below the main menu, input the dimensions and dpi you desire. Then, you can simply crop the image by selecting the entire canvas and pressing enter. This will automatically resize your image, plus, until you clear or change the parameters, your crop tool will perform the same exact function on any other images you are working with.

Also, one trick that works reasonably well for upsizing an image without losing significant resolution is to use the Image Size dialog box, shown above, and change your measurement to percentage. Then, upsize the image to 110%. If you do this several times, you will notice that the resolution of the now larger image remains surprisingly consistent. I have used this and it works, however, no one seems to be able to explain why. Use 109% or 111% and it doesn’t work nearly as well... 110 seems to be a magic number in Photoshop. In fact, the book, Photoshop for Digital Photographers claims that you can upsize a thumbnail to a poster without ruing the image. I’m not sure you can take this trick that far, but it does work for at least doubling the size of an image, or making a web image of hi-enough resolution for printing.

jb

Henry Wolf , January 25, 2004; 01:20 P.M.

I'm not so sure about the dpi, and I would be glad to hear what people here say about my view.

IMHO: out of my workflow and considerations comes the one and only file that contains all the information available to me, as TIFF, or PNG at about half the size. For web use, no question that I have to resample it to the intended ppi screen size. But for printing, why worry? I feel that it would be best to leave it entirely to the designers of the driver, ie. printer. They have to put up with all sorts of files. If I pick some dpi by guesswork, the driver will have to re-process it either way all over again, with suboptimal results. My advantage of "no dpi" would be that I have the one and only original file ready for all imaginable purposes.

You see what I mean. Comments?

Peter Heritage , January 25, 2004; 02:30 P.M.

"Note that there are some programs which create JPEG files which the photo.net software for attached images doesn't understand. If you attach such an image to a forum post it will display as a link, not as an image, even if it's under 511 pixels wide. One program known to do this is Micrografx Picture Publisher 8.0. There may well be others."

There are - Picture Window Pro is one of them. It took me some time to figure out, but now I use PS Elements to resize on those rare occasions when I need to include in-line attachments.

Gordon Richardson , January 26, 2004; 11:29 P.M.

There have been several threads on photo.net about this, but it is good to have a single article which clarifies much of the confusion surrounding pixels and image size.

The concept that increasing size by increments of 110% has been raised before, but there is by no means agreement that it actually provides any benefit. See: Can enlarge w- Photoshop 110% with great results (http://www.photo.net/bboard/q-and-a-fetch-msg?msg_id=005d5j).

Bob Atkins , January 27, 2004; 03:24 P.M.

I agree that the difference using lots of small increments vs. one big increment makes very little difference in most cases. I've seen all sorts of "tricks" suggested. For example I've seen statements, as above, that upsizing by 110% is "magic" and 109% and 111% don't work as well.

I'm not sure I believe most of these statements. My own experience suggest that the program you use and the algorithm you pick makes significantly more difference than the step size you use.

There's certainly an advantage in using bicubic over linear interpolation for photographic images for example. Going to an algorithm like Lanczos can improve things a little, but really all of the upsizing algorithms have problems. If the information isn't there in the first place, it's essentially impossible to accurately recreate it.

Genuine fractals is works in a somewhat different way. Sometimes it works, sometimes it creates something that looks more like a watercolor painting than a photograph when you look closely at an image.

The bottom line is that there is no single "best" (or least worse)solution for upsizing images.

Robert Nancarrow , February 06, 2004; 12:56 P.M.

This thread has answered a lot of my questions about re-sizing (I'm new to digital imaging). Could someone tell me what happens to the original image when you re-size? Is it preserved ?, does the re-sized file ask for a new name? Can I always go back to the original large (maybe RAW) file for same image quality even if I've altered it from re-sizing and other changes? Please give examples, they help a lot. I currently use Elements but will soon be getting PS. Thanks, Bob.

Claudio Cortes , February 24, 2004; 09:56 A.M.

I've tried Irfranview for resize images but my experience is color loss specially when going from 1800 to 768 or 192 pixels. I' ve got better results using Photoshop 6.0 and taking advantage of automated tools (batch and scripting) and saving images as web. In this link is La Papa

James Burger , February 25, 2004; 03:01 P.M.

On Robert Nancarrow’s question about saving your resized image: you can always revert to your raw image, but you will have to reapply any processing parameters you used to make the tiff or jpg. Raw files are your “negatives.” You cannot save changes to them; you can only export a copy with the applied changes in a different file format. The Raw file itself will not be changed.

When you resize or make any other alterations to an image in Photoshop (or another program) it’s advisable to save those changes under a new file name. Once you overwrite a file with new information, your old info is gone (or at least very difficult to recover). In other words, if I am in Photoshop working on a file called abc.jpg and I increase its size by 200 percent then save it, abc.jpg will always appear 200 percent larger than it was before the changes were made. So, when I know that I am about to alter an image, I’ll resave it with a different name – in this example, I would save it as “abc_2.jpg.” (I usually “Save As,” before I do anything else, because otherwise you run the risk of hitting “Save” by accident and overwriting the original). Then, only when I am sure that the original is less desirable than the new version, will I permanently delete or alter a file.

Hope that helps

jb

Traverse Davies , April 22, 2004; 04:01 P.M.

A couple of things. first PPI: Many browsers only understand 72ppi images, they will balk at all other resolutions and the image won't display, so if you are formatting for the web it is best to assume that at least some people won't be able to view other resolutions and save as 72ppi, Second Photoshop: Use save for web under the file menu if saving in photoshop 7.0 or better and intending to post to a web site, that automatically formats for 72ppi. The other reason for doing this is simply that photoshop puts extra info in the XML meta info for jpg images, save for web doesn't do that and many browsers balk at the extra info. I just tried the 110% and managed to get an image up to almost double, after that it was unusable.

Bob Atkins , July 24, 2004; 01:20 A.M.

As far as I know most, if not all, browsers and operating systems ignore the dpi setting, so whether it's set at 7 dpi, 72dpi or 720 dpi - or even 7200dpi - makes absolutely no difference at all.

Scott Jones , November 06, 2004; 08:20 P.M.

Thank God that the author puts to rest this silly 72dpi figure that people bandy about, not realizing that for screen viewing, it makes no difference. The scan tips web site also makes this clear. Thank you for making this clear...

Haris Taufiq , December 15, 2005; 01:32 A.M.

First time that I ever understood to some extent what these terms mean.

Geoff Doane , November 16, 2007; 12:45 P.M.

I'm late to the party on this, but thought I should add an observation I made a few weeks ago. When dropping images into a MS Outlook message, they appeared larger (and fuzzier) on the screen than they had in Photoshop Elements. In this case, the DPI setting did make a difference (I realize Outlook isn't exactly "web", but many people might assume it is). Elements defaulted to 72 DPI, but Outlook wants to see images at 96 DPI in order to maintain the same size. Changing the DPI to 96 before resizing the original images gave me the results I was looking for.

The other parameter that is affected by DPI in Photoshop is the size of any text you add to the image. It seems that the text size in points is tied in to DPI setting for the image. I imagine that once you save the image, everything is locked together,and the relative size of the text can't change.

Barbara Yasuhara , July 31, 2010; 10:13 A.M.

Hey everybody!  Not sure I am posting this in the right spot, but you all seem pretty knowledgeable  in dpi, pixels, etc.

Does anyone out there know of any tutorials or articles on Photography vs printing?  What I mean is that I am a little challenged in printing exactly what I see in my final photos.  I get alot of cropping etc.  

I know I can fix this problem in PS with the crop tool for say 8x10, but are there any tricks for getting it perfect "in camera" so it's ready to print, without cropping?

Carole Walsh , January 26, 2011; 10:04 A.M.


At the correct resolution, individual pixels will not be visible.

This was an excellent article. In addition to the questions about resolution for web vs. resolution for print, one also needs to pay attention to the color space used. The color space for web is RGB, while the color space for print is CMYK. You can read more about that in the article "Everything You Always Needed to Know About Color, But Didn't Know What to Ask" at http://www.carolewalsh.com/2010/08/pmscmykrgb.

There's also an article on resolution ("The Confusion About Resolution") at http://www.carolewalsh.com/2011/01/resolutionconfusion, which includes the formula for figuring out the size (in inches) and resolution of an image from the size in pixels.

Ken L , July 06, 2011; 04:48 P.M.

Another very late, but highly relevant question:

When you check the file properties of a photo using windows explorer or any photo editor, it will give you the resolution of the photo in PPI (pixels per inch). How does that make sense? According to this article, PPI takes any meaning when you print a photo. But the file properties of any photo show the resolution in PPI as if it was inherent to the picture regardless of how you'd want to print the photo.

Can someone explain please?

Prashant Kumar , April 12, 2012; 07:40 A.M.

how to claim back ppi

Use Adobe Photoshop CS5. Its better and the latest.

---------------------------------------------------------------

Prashant.

 Payment protection insurance (PPI), is an add on coverage that is designed to insure against the loss of payment for monthly installments of loans and credit cards.Claim it when you need it. Read the details in the below link on how to claim ppi.

http://www.howtoclaimbackppi.org.uk


Add a comment



Notify me of comments