The biggest web design myth is that screen graphics should be 72dpi.
December 10, 2001 9:06 AM Subscribe
The biggest web design myth is that screen graphics should be 72dpi. Sorry if this counts as a self-link (it's a long discussion that I sparked on a forum devoted to a vector graphics app called Xara), but I thought it might interest designers & programmers here.
Mostly I use 72 because if I scan something thats 1"x1" the screen equivilant is 72dpi.
Doesnt relaly matter, though...
posted by atom128 at 9:22 AM on December 10, 2001
Doesnt relaly matter, though...
posted by atom128 at 9:22 AM on December 10, 2001
If the filesize for a 72 is smaller than a 1000, but they both look the same on the web, I'll go with 72. I don't create web content to be printed.
There. An answer that doesn't require me to use a calculator.
posted by jragon at 9:25 AM on December 10, 2001
There. An answer that doesn't require me to use a calculator.
posted by jragon at 9:25 AM on December 10, 2001
it is not a myth, it really has little to do with screen presentation which is waht the forum seemed so keen on.
It's all about image-optimization, sure, if you want a big bloated 300dpi image used, be my guest, but the information has to be stored for the extra ppi in there, and depending on which file format you use-- it'll get huge quick.
posted by tj at 9:28 AM on December 10, 2001
It's all about image-optimization, sure, if you want a big bloated 300dpi image used, be my guest, but the information has to be stored for the extra ppi in there, and depending on which file format you use-- it'll get huge quick.
posted by tj at 9:28 AM on December 10, 2001
The filesize is the same. I just checked.
I created a new 72 x 72 pixel image in photoshop at 72 dpi. JPEG filesize = 367 bytes. A new 72 x 72 pixel image in photoshop at 300 dpi = 367 bytes as jpeg. From what I understand based on using Bryce, 300 or 600 or 72 dpi is merely a function of the header at the top of the graphic and how it is interpreted by graphics programs and printers.
Anyway. Why do I use 72 dpi? Because it's standard, prints "normally", and is a 1:1 ratio. And that's why I will continue to use 72 dpi for the web, since the web != print.
posted by xyzzy at 9:29 AM on December 10, 2001
I created a new 72 x 72 pixel image in photoshop at 72 dpi. JPEG filesize = 367 bytes. A new 72 x 72 pixel image in photoshop at 300 dpi = 367 bytes as jpeg. From what I understand based on using Bryce, 300 or 600 or 72 dpi is merely a function of the header at the top of the graphic and how it is interpreted by graphics programs and printers.
Anyway. Why do I use 72 dpi? Because it's standard, prints "normally", and is a 1:1 ratio. And that's why I will continue to use 72 dpi for the web, since the web != print.
posted by xyzzy at 9:29 AM on December 10, 2001
From the thread:
[72 is] the default "screen resolution" of the Mac. And virtually everybody who use a Mac - including tons of graphics people - are brain-dead, because the Mac does not encourage the use of one's brain.
Wow.
posted by jragon at 9:30 AM on December 10, 2001
[72 is] the default "screen resolution" of the Mac. And virtually everybody who use a Mac - including tons of graphics people - are brain-dead, because the Mac does not encourage the use of one's brain.
Wow.
posted by jragon at 9:30 AM on December 10, 2001
Can't really argue with that kind of logic, can you?
posted by dsandl at 9:35 AM on December 10, 2001
posted by dsandl at 9:35 AM on December 10, 2001
Another popular myth is the integrity of the 216 "web safe" colors. Webmonkey has a good article about it.
posted by gazingus at 9:39 AM on December 10, 2001
posted by gazingus at 9:39 AM on December 10, 2001
Having done both print AND web work, dpi is not really an issue for web folks, just keep it at 72 or 96 and you're not going to run into any problems.
Just don't expect to be able to use the same graphic for a print project, because the quality will be unacceptabe.
As far as the web pallette goes, the infamous 216 is the closest thing to a safe pallette to use, unless Pantone has succeeded in their whole "web color" thing they were pushing back in '99
posted by tj at 9:50 AM on December 10, 2001
Just don't expect to be able to use the same graphic for a print project, because the quality will be unacceptabe.
As far as the web pallette goes, the infamous 216 is the closest thing to a safe pallette to use, unless Pantone has succeeded in their whole "web color" thing they were pushing back in '99
posted by tj at 9:50 AM on December 10, 2001
Wow, jragon. I wish that person would come to metafilter so we could spend a while beating the arrogance out of them.
It is true, however, that there were 72 pixels to the inch on the original 9" Macintosh monitor. That has remained the nominal resolution ever since, whether the monitor's actual resolution is higher or lower.
I just measured; the PowerBook I'm using has 78 pixels to the inch. Not bad.
-Mars
posted by Mars Saxman at 9:51 AM on December 10, 2001
It is true, however, that there were 72 pixels to the inch on the original 9" Macintosh monitor. That has remained the nominal resolution ever since, whether the monitor's actual resolution is higher or lower.
I just measured; the PowerBook I'm using has 78 pixels to the inch. Not bad.
-Mars
posted by Mars Saxman at 9:51 AM on December 10, 2001
Tj and Jragon, I disagree (I think you've both fallen for the myth, as I did too for years and years). These discussions have a way of spiralling into unpleasantness (e.g. the mac/braindead comment), so please note I mean no disrespect.
Tj, there is NO difference in file size! Try creating a 100px by 100px image at 72dpi and another 100px by 100px image at 300dpi. They will both be the same size! File size is determined by number of pixels, NOT dpi (actually, it's number-of-pixels times color bit depth). Since both files have 10000px (100 times 100), they are both the same size. Since they are both the same size, they will take the same length of time to download. But don't take my word for it (I didn't take anyone's word for it). Try it!
On the other hand, if you create a 1 INCH by 1 INCH file at 72dpi it will be much SMALLER (in file size) than a 1 INCH by 1 INCH 300dpi file. This is because the 300dpi version is cramming many more pixels into the 1 square inch than the 72dpi version. Again, it's number-of-pixels, not dpi that matters.
Even if dpi did matter, what is magic about the number 72? Why not 70 or 50 or 100 or 200?
And I agree with you about the web safe colors, gazingus.
Now I have to say a mea culpa for each student I've taught the myth to...
posted by grumblebee at 9:56 AM on December 10, 2001
Tj, there is NO difference in file size! Try creating a 100px by 100px image at 72dpi and another 100px by 100px image at 300dpi. They will both be the same size! File size is determined by number of pixels, NOT dpi (actually, it's number-of-pixels times color bit depth). Since both files have 10000px (100 times 100), they are both the same size. Since they are both the same size, they will take the same length of time to download. But don't take my word for it (I didn't take anyone's word for it). Try it!
On the other hand, if you create a 1 INCH by 1 INCH file at 72dpi it will be much SMALLER (in file size) than a 1 INCH by 1 INCH 300dpi file. This is because the 300dpi version is cramming many more pixels into the 1 square inch than the 72dpi version. Again, it's number-of-pixels, not dpi that matters.
Even if dpi did matter, what is magic about the number 72? Why not 70 or 50 or 100 or 200?
And I agree with you about the web safe colors, gazingus.
Now I have to say a mea culpa for each student I've taught the myth to...
posted by grumblebee at 9:56 AM on December 10, 2001
Yeah, that guy Klaus is very full of arrogance:
"I am an intransigent atheist, and have been so since kindergarten. Religion is bunk, as is astrology, parapsychology, reincarnation, spoon bending and all other kinds of superstitious garbage – including environmentalism, which is based on the mystical and perverse idea that Nature has a value apart from human needs and goals. In short, I decry all notions which are contrary to reason, logic and sound science."
But it is true that monitor display is completely independent of the concept of inches.
posted by donkeymon at 10:05 AM on December 10, 2001
"I am an intransigent atheist, and have been so since kindergarten. Religion is bunk, as is astrology, parapsychology, reincarnation, spoon bending and all other kinds of superstitious garbage – including environmentalism, which is based on the mystical and perverse idea that Nature has a value apart from human needs and goals. In short, I decry all notions which are contrary to reason, logic and sound science."
But it is true that monitor display is completely independent of the concept of inches.
posted by donkeymon at 10:05 AM on December 10, 2001
Does anyone know of a good resource that goes into the dpi topic WITHOUT buying into the myth?
posted by grumblebee at 10:07 AM on December 10, 2001
posted by grumblebee at 10:07 AM on December 10, 2001
grumblebee, I don't see these "myths" as hard and fast rules, more as guidelines that as long as you folow them, you won't get into any trouble..
When you're creating an image, you are correct (as long as you're using pixels as the dimensions). It is when you are using an image that is pre-existing that this comes into play.
You just can't assume that you're working on an image that was created w/ the web in mind, since print is still pretty big, and images originally intended for it are almost allways specified in real-world dimensions.
posted by tj at 10:08 AM on December 10, 2001
When you're creating an image, you are correct (as long as you're using pixels as the dimensions). It is when you are using an image that is pre-existing that this comes into play.
You just can't assume that you're working on an image that was created w/ the web in mind, since print is still pretty big, and images originally intended for it are almost allways specified in real-world dimensions.
posted by tj at 10:08 AM on December 10, 2001
Klaus may or may not be arrogant. He's certainly forthright about what he believes. The charge of arrogance rests on your interpretation of his second sentence. Does he mean "I think religion is bunk..." or does he mean "religion is bunk and anyone who believes in it is stupid"?
I share most of Klaus's beliefs, but I don't look down on people who are religious, so I don't consider myself arrogant. So I could have written the same bio for myself, though my wording probably would be a bit more humble.
Anyway, back to dpi... Anyone want to explain why 72 is an important number?
posted by grumblebee at 10:11 AM on December 10, 2001
I share most of Klaus's beliefs, but I don't look down on people who are religious, so I don't consider myself arrogant. So I could have written the same bio for myself, though my wording probably would be a bit more humble.
Anyway, back to dpi... Anyone want to explain why 72 is an important number?
posted by grumblebee at 10:11 AM on December 10, 2001
that link...very long discussion for something so simple.
lets not recreate it here...please. i beg you.
pixels is pixels, printing is dpi, lpi etc...there is a book that came out a while ago, Digital PrePress Complete from hayden books, good reference, as is any photoshop book. Read your books people. It isn't mysterious.
posted by th3ph17 at 10:18 AM on December 10, 2001
lets not recreate it here...please. i beg you.
pixels is pixels, printing is dpi, lpi etc...there is a book that came out a while ago, Digital PrePress Complete from hayden books, good reference, as is any photoshop book. Read your books people. It isn't mysterious.
posted by th3ph17 at 10:18 AM on December 10, 2001
I see where you're going, tj. It's a little like saying don't eat cake if you want to lose weight. This doesn't take into account someone who eats a tiny piece of cake each night after a day of fruits and vegetables, but maybe it will keep most people out of trouble.
So basically you're saying, take the high-res image and knock it down to 72. This will reduce the number of pixels to a managable amount for web traffic and web display. Is that right?
In general, I guess this is true. Though you could just as easily look at the number of pixels in the width and the height. If you're designing a 800 x 600 webpage and your image is 1000 x 2000, that tells you pretty quickly there's going to be trouble. If you reduce the number of pixels to 100 x 200, you're fine.
Also, a 1000 x 2000 72dpi image is still really big (just as big as the 300dpi version), so it will load slowly.
Perhaps some of this also comes down to a dislike for magical numbers. Even if you used the "knock it down to 72 to generally get the results you want" approach, the 72 stands out as really odd. Why not say knock it down to 70 or 100?
posted by grumblebee at 10:19 AM on December 10, 2001
So basically you're saying, take the high-res image and knock it down to 72. This will reduce the number of pixels to a managable amount for web traffic and web display. Is that right?
In general, I guess this is true. Though you could just as easily look at the number of pixels in the width and the height. If you're designing a 800 x 600 webpage and your image is 1000 x 2000, that tells you pretty quickly there's going to be trouble. If you reduce the number of pixels to 100 x 200, you're fine.
Also, a 1000 x 2000 72dpi image is still really big (just as big as the 300dpi version), so it will load slowly.
Perhaps some of this also comes down to a dislike for magical numbers. Even if you used the "knock it down to 72 to generally get the results you want" approach, the 72 stands out as really odd. Why not say knock it down to 70 or 100?
posted by grumblebee at 10:19 AM on December 10, 2001
th3ph12, I understand your irritation. But I have read many books on the subject (though not the one you mention, which I'll check out). Most books perpetuate the myth!
posted by grumblebee at 10:20 AM on December 10, 2001
posted by grumblebee at 10:20 AM on December 10, 2001
I mostly use the myth on print designers to prevent them fro dumpgin piles of files which are 3000-5000 pixels wide when they only need to be 300 pixels wide for the website. So I tell them to reduce the image to 72 dpi, the dull click of familiarity goes of in their heads, and my hard drive doesn't get filled up with their huge images. In the end it is a decent approximation based on a very old measurement.
(Please note that I am only making fun of the particular print designer with whom I work, not all print designers)
posted by donkeymon at 10:33 AM on December 10, 2001
(Please note that I am only making fun of the particular print designer with whom I work, not all print designers)
posted by donkeymon at 10:33 AM on December 10, 2001
So basically you're saying, take the high-res image and knock it down to 72. This will reduce the number of pixels to a managable amount for web traffic and web display. Is that right?
exactly grumblebee. this is an issue when you're using images originally intended for print, which you WILL run into sooner or later (most likely sooner).
Though you could just as easily look at the number of pixels in the width and the height. If you're designing a 800 x 600 webpage and your image is 1000 x 2000, that tells you pretty quickly there's going to be trouble. If you reduce the number of pixels to 100 x 200, you're fine.
You're essentially knocking down the dpi (remember - a print term)when you do this
The only reason 72 is used is that it's the lowest dpi that you can display on screen w/out affecting image quality (of course as long as you're not using pixel dimensions)
posted by tj at 10:33 AM on December 10, 2001
exactly grumblebee. this is an issue when you're using images originally intended for print, which you WILL run into sooner or later (most likely sooner).
Though you could just as easily look at the number of pixels in the width and the height. If you're designing a 800 x 600 webpage and your image is 1000 x 2000, that tells you pretty quickly there's going to be trouble. If you reduce the number of pixels to 100 x 200, you're fine.
You're essentially knocking down the dpi (remember - a print term)when you do this
The only reason 72 is used is that it's the lowest dpi that you can display on screen w/out affecting image quality (of course as long as you're not using pixel dimensions)
posted by tj at 10:33 AM on December 10, 2001
Sheesh, that discussion you linked to laid it all out pretty clearly by the third or fourth reply. There's no big mystery to any of this.
On a computer's monitor an image's size depends only on the number of pixels. People talk about dpi mainly because that's what's used in print design, and a lot of the early web designers came straight from print (lots still do).
There's some sense in talking about 72dpi or 96dpi in computer images for those folks who are used to thinking in/working with dpi and want a rough idea how large, physically, a given image will display on "most" users' screens. The 72/96 values are "arbitrary" and based on the history of the technology, but it's still a good ballpark. Current monitor/lcd technology is pushing towards 200ppi, I believe, but it'll be a while before that's available in consumer devices.
Also, aren't "points" as in font size defined as 72 per inch?
posted by cps at 10:41 AM on December 10, 2001
On a computer's monitor an image's size depends only on the number of pixels. People talk about dpi mainly because that's what's used in print design, and a lot of the early web designers came straight from print (lots still do).
There's some sense in talking about 72dpi or 96dpi in computer images for those folks who are used to thinking in/working with dpi and want a rough idea how large, physically, a given image will display on "most" users' screens. The 72/96 values are "arbitrary" and based on the history of the technology, but it's still a good ballpark. Current monitor/lcd technology is pushing towards 200ppi, I believe, but it'll be a while before that's available in consumer devices.
Also, aren't "points" as in font size defined as 72 per inch?
posted by cps at 10:41 AM on December 10, 2001
Anyone want to explain why 72 is an important number?
Why is the US standard railroad guage 4ft 8.5 inches? It just came to be that way because of the technology available at the time. In the case of the railroad, it was a horse's ass. In the case of the computer screen, it was an 11.1" monitor with a 640x480 display.
posted by dlewis at 10:42 AM on December 10, 2001
Why is the US standard railroad guage 4ft 8.5 inches? It just came to be that way because of the technology available at the time. In the case of the railroad, it was a horse's ass. In the case of the computer screen, it was an 11.1" monitor with a 640x480 display.
posted by dlewis at 10:42 AM on December 10, 2001
why on earth is this a remotely complicated issue? why is it even possible for a "myth" to exist? there is not a shred of ambiguity here. computers display pixels - rasters.
real world lengths only go into instructions for printers, because they print on real world paper, have resolutions that are relatively continuous (compared to screens), and (because of the near continuous resolution) it makes sense to pass them instructions as vectors with inverse length units.
computers have such fat pixels that if you care about "actual size" on screen (like if you're developing a school lesson, and you want kids to put a ruler to the screen) then you should think in terms of dot pitch - the size of the pixel on the screen (a few tenths of a millimeter) instead of the inverse length units of printer graphics.
posted by badstone at 10:46 AM on December 10, 2001
real world lengths only go into instructions for printers, because they print on real world paper, have resolutions that are relatively continuous (compared to screens), and (because of the near continuous resolution) it makes sense to pass them instructions as vectors with inverse length units.
computers have such fat pixels that if you care about "actual size" on screen (like if you're developing a school lesson, and you want kids to put a ruler to the screen) then you should think in terms of dot pitch - the size of the pixel on the screen (a few tenths of a millimeter) instead of the inverse length units of printer graphics.
posted by badstone at 10:46 AM on December 10, 2001
THANK YOU for publishing on this topic, i've been crowing about this for years!
posted by bug138 at 10:46 AM on December 10, 2001
posted by bug138 at 10:46 AM on December 10, 2001
As an additional comment, let me make a few notes:
People often use 72dpi so that they can use points as a pixel measurement in vector-oriented programs like Freehand, since at 72dpi, one Pont (1/72") equals one pixel.
For the web the idea of the "inch" is pure fiction. Its all about Pixels. You could deliver 100 pixel square images at 50,000dpi and they would still be 100 pixels square on the screen, same size as the 2dpi image.
This does get more complicated when you start dealing with font sizes and how large a given Point Size of a font will render on the screen. Windows uses a 96dpi model internally for font rendering, so a 20 point (20/72") font will be rendered as 26 pixels in height, wheras the Mac uses a 72dpi internal model and will render a 20 point font as 20 pixels. This can vex web developers using points in their CSS code crossplatform.
posted by bug138 at 10:56 AM on December 10, 2001
posted by bug138 at 10:56 AM on December 10, 2001
I'm starting to think all those mac-using webheads *are* idiots. Must...remain...respectful...
bug138: The Windows dpi for fonts can be 72 or 96, I think, based on the "Large Fonts" setting. Moreover it's customizable.
posted by cps at 11:00 AM on December 10, 2001
bug138: The Windows dpi for fonts can be 72 or 96, I think, based on the "Large Fonts" setting. Moreover it's customizable.
posted by cps at 11:00 AM on December 10, 2001
bug138, when you say Windows usess 96dpi internally for font rendering, what does this mean? You translate that to 20points and 20/72", but what if the font is being displayed on a HUGE screen (i.e. cinema or Times Square). Does Windows think internally that its output is a certain size monitor?
posted by grumblebee at 11:02 AM on December 10, 2001
posted by grumblebee at 11:02 AM on December 10, 2001
Well said, bug138. And just in case you're not completely confused at this point: Windows' large font model assumes 120dpi, while UNIX displays are generally one of 75dpi or 100dpi. The great thing about standards is that there are just so many to choose from...
And if that hasn't convinced anyone in web design to stick with pixels, I don't know what will.
posted by dlewis at 11:02 AM on December 10, 2001
And if that hasn't convinced anyone in web design to stick with pixels, I don't know what will.
posted by dlewis at 11:02 AM on December 10, 2001
Go work at a digital service bureau for a while and I guarantee you any confusion on this issue will disappear in the first two hours.
posted by jeremias at 11:05 AM on December 10, 2001
posted by jeremias at 11:05 AM on December 10, 2001
Well, DPI is just a meaningless number unless you're going to print something. Period. If you're not going to print something, ignore the DPI. For Web, Film, DV, NTSC, PAL, etc. you need your pixel count (640x480 PAL, for instance), and your software will pretty much ignore the DPI.
Conversely, the pixel count is irrelevant if you're going to print. THAT is when you set the DPI to something appropriate (150 for magazines, higher for photos) and set the image size in inches, rather than pixels.
I don't see what's so hard to understand about this.
posted by phalkin at 11:10 AM on December 10, 2001
Conversely, the pixel count is irrelevant if you're going to print. THAT is when you set the DPI to something appropriate (150 for magazines, higher for photos) and set the image size in inches, rather than pixels.
I don't see what's so hard to understand about this.
posted by phalkin at 11:10 AM on December 10, 2001
Webmonkey has a good article about it.Slashdot readers thoroughly deconstructed the 22 really websafe colours and get it down to about 16.
posted by holloway at 11:11 AM on December 10, 2001
I'll chime in here on what I know (mainly Windows development). In Windows the basic unit is a twip. A twip is defined as one twentieth of a point. And there are 1440 twips per inch. Divide 1440/20 and we get 72 points per inch. These are device-independent units that you can use to layout your forms (again Windows forms). Windows provides APIs for converting from twips to pixels.
All windows graphics are in pixels though, so any GDI (Graphics Device Interface) calls are in pixels.
And there is a third unit that all dialog templates are in. DLU (Dialog Units). A horizantal DLU is defined as the average width of the current font / 4, while a vertical DLU is defined as the average font height/8. So when Microsoft defines the standard layout of wizards, property pages, et al. The units are given in DLU.
Now on my system 19" monitor/1024x768 resolution, I get 15 twips per pixel (or so windows tells me). This gives me 1024*15=15360 twips divide by 1440 to get 10.67 inches across. This is obviously not correct. This is however printing resolution I believe.
Or I could be a moron.....
posted by patrickje at 11:14 AM on December 10, 2001
All windows graphics are in pixels though, so any GDI (Graphics Device Interface) calls are in pixels.
And there is a third unit that all dialog templates are in. DLU (Dialog Units). A horizantal DLU is defined as the average width of the current font / 4, while a vertical DLU is defined as the average font height/8. So when Microsoft defines the standard layout of wizards, property pages, et al. The units are given in DLU.
Now on my system 19" monitor/1024x768 resolution, I get 15 twips per pixel (or so windows tells me). This gives me 1024*15=15360 twips divide by 1440 to get 10.67 inches across. This is obviously not correct. This is however printing resolution I believe.
Or I could be a moron.....
posted by patrickje at 11:14 AM on December 10, 2001
i thought i had a pixel
it turned into a dot
i tried to print as pixels
but pixels they were not
pixels they are square
unless you make them video
and dots are round when made
though its in a square cell they print
the dot gain fills the square up
the line screen makes the tone
aye, what handy knowledge,
at work or while at home
and photoshop i do recall
it used to try and save
at 72, or 75--oh boy,
those were the days.
so webbers keep your pixels
and printers keep your dots
and gamers keep your pants on,
its frame rate that you want.
and lest you think you've been forgot,
i will here mention twips
the math of which i oft forget
while writing rhyming quips.
posted by th3ph17 at 11:20 AM on December 10, 2001 [1 favorite]
it turned into a dot
i tried to print as pixels
but pixels they were not
pixels they are square
unless you make them video
and dots are round when made
though its in a square cell they print
the dot gain fills the square up
the line screen makes the tone
aye, what handy knowledge,
at work or while at home
and photoshop i do recall
it used to try and save
at 72, or 75--oh boy,
those were the days.
so webbers keep your pixels
and printers keep your dots
and gamers keep your pants on,
its frame rate that you want.
and lest you think you've been forgot,
i will here mention twips
the math of which i oft forget
while writing rhyming quips.
posted by th3ph17 at 11:20 AM on December 10, 2001 [1 favorite]
There's a reason for 72dpi: The number of "lines" per inch, in print, is 72. Also, there are 12 points in a pica, and approximately 6 picas in an inch. These measurements are used for print, so naturally, 72dpi is optimal for translating print dimensions into pixel dimensions.
The rest of you guys covered the topic well enough.
posted by Down10 at 11:22 AM on December 10, 2001
The rest of you guys covered the topic well enough.
posted by Down10 at 11:22 AM on December 10, 2001
I made test page with a 200 pixel square saved as 72 DPI and 600 DPI GIFs, JPEGs, and PNGs. If you look at the file sizes in bytes the 600 DPI versions are 3–5 times larger than the 72 DPI versions.
Photoshop's Save for Web reduced the resolution of all of the images to 72 DPI regardless of the initial setting; using Save As retained the initial DPI settings. The JPEGs and PNGs shifted colors. I used the Macintosh version of Photoshop 6.0.1 under the Classic mode of OS X.
posted by kirkaracha at 11:34 AM on December 10, 2001
Photoshop's Save for Web reduced the resolution of all of the images to 72 DPI regardless of the initial setting; using Save As retained the initial DPI settings. The JPEGs and PNGs shifted colors. I used the Macintosh version of Photoshop 6.0.1 under the Classic mode of OS X.
posted by kirkaracha at 11:34 AM on December 10, 2001
And here I thought that the biggest web design myth was that people really do like splashy colors and animation and big advertisements to load first when they're trying to read the news/opinion/article on a site while using a slow connection. Silly me.
posted by yesster at 11:59 AM on December 10, 2001
posted by yesster at 11:59 AM on December 10, 2001
yesster, the big myth was that despite nobody liking that, they could still make money off it somehow.
posted by cps at 12:12 PM on December 10, 2001
posted by cps at 12:12 PM on December 10, 2001
kirkaracha, this doesn't make sense. I did the same test, and my files came out equal in size regardless of dpi (I DIDN'T use save for web). what are we doing differently?
posted by grumblebee at 12:13 PM on December 10, 2001
posted by grumblebee at 12:13 PM on December 10, 2001
Is it just my imagination or does the linked forum sound an awful lot like an infomercial for some vague graphics software product?
"ANOTHER myth shattered! Wow, Klaus! Can I really cut through that tin can with this amazing pocket fisherman tool??"
Sheesh.
posted by 40 Watt at 12:14 PM on December 10, 2001
"ANOTHER myth shattered! Wow, Klaus! Can I really cut through that tin can with this amazing pocket fisherman tool??"
Sheesh.
posted by 40 Watt at 12:14 PM on December 10, 2001
kirkaracha...
those 600 dpi images are 0.333 inches wide. 200 pixels.
exactly the same as 72 dpi--2.778 inches wide, 200 pixels.
if the 2.778 inch wide pic were 600 dpi, it would show as 1667 pixels on screen.
File differences aren't going to be based on resolution, but on compression and file type. Anything else is impossible. Or am i going insane? Someone hold me. This thread is turning into that linked one.
posted by th3ph17 at 12:15 PM on December 10, 2001
those 600 dpi images are 0.333 inches wide. 200 pixels.
exactly the same as 72 dpi--2.778 inches wide, 200 pixels.
if the 2.778 inch wide pic were 600 dpi, it would show as 1667 pixels on screen.
File differences aren't going to be based on resolution, but on compression and file type. Anything else is impossible. Or am i going insane? Someone hold me. This thread is turning into that linked one.
posted by th3ph17 at 12:15 PM on December 10, 2001
Does anyone know of a good resource that goes into the dpi topic WITHOUT buying into the myth?
Are you trying to win an argument at work or something? For God's sake, you started this exact same discussion on another board three days ago.
posted by KLAX at 12:16 PM on December 10, 2001
Are you trying to win an argument at work or something? For God's sake, you started this exact same discussion on another board three days ago.
posted by KLAX at 12:16 PM on December 10, 2001
KLAX, no I'm not trying to win an argument. I'm trying to understand something. I am also a teacher who is disturbed that he's been spreading false knowledge for years. Can you understand that?
40 Watt, that forum isn't an infomercial. It's a fan/tutorial site for users of Xara. It's only commercial in the sense that Xara, ltd. hosts the forum (which also includes discussions re: Photoshop, Flash, etc.). But negative comments about the tool are as welcome there as positive ones.
Yes, Xara is obscure and for good reason. It's such a good vector graphics app (for the PC only, alas), that a few years ago Corel bought it, renamed it CorelXara and proptly burried it. Now Xara, ltd is out of their contractual obligations with Corel, so they can again market their app independently. I'm in no way affiliated with Xara, ltd., but having used Illustrator, Freehand, CorelDraw and Flash for years, I'd pick Xara any day, hands down. Its feature-set is rich, its price is low, its footprint is tiny (written totally in assemnly), and it rarely crashes.
I DID start a discussion on the Xara board and then moved it here. I did this because I know a lot of smart webheads hang out here. I also know that there are some people here who love to flame, insult, mock, etc. This is too bad. If you don't like reading something, why don't you move on?
posted by grumblebee at 12:36 PM on December 10, 2001
40 Watt, that forum isn't an infomercial. It's a fan/tutorial site for users of Xara. It's only commercial in the sense that Xara, ltd. hosts the forum (which also includes discussions re: Photoshop, Flash, etc.). But negative comments about the tool are as welcome there as positive ones.
Yes, Xara is obscure and for good reason. It's such a good vector graphics app (for the PC only, alas), that a few years ago Corel bought it, renamed it CorelXara and proptly burried it. Now Xara, ltd is out of their contractual obligations with Corel, so they can again market their app independently. I'm in no way affiliated with Xara, ltd., but having used Illustrator, Freehand, CorelDraw and Flash for years, I'd pick Xara any day, hands down. Its feature-set is rich, its price is low, its footprint is tiny (written totally in assemnly), and it rarely crashes.
I DID start a discussion on the Xara board and then moved it here. I did this because I know a lot of smart webheads hang out here. I also know that there are some people here who love to flame, insult, mock, etc. This is too bad. If you don't like reading something, why don't you move on?
posted by grumblebee at 12:36 PM on December 10, 2001
Is there any point to this, other than possibly having web images look nicer when printed?
grumblebee (excellent name, btw): OS X's Show Info lied to me. When I check the files in the Terminal, the file sizes for the high- and low-res GIFs are the same, same with the PNGs. (I could have predicited this, but I hadn't had any caffeine yet.) I corrected my test page.
th3ph17: I don''t what to tell you; I just doublechecked the Image Size and the inch measurements are what you said, but both images are 200 x 200 pixels.
When I made the images I created a new 200 pixel x 200 pixel file in Photoshop and set the DPI to either 72 or 600. I didn't do anything with inches.
posted by kirkaracha at 12:39 PM on December 10, 2001
grumblebee (excellent name, btw): OS X's Show Info lied to me. When I check the files in the Terminal, the file sizes for the high- and low-res GIFs are the same, same with the PNGs. (I could have predicited this, but I hadn't had any caffeine yet.) I corrected my test page.
th3ph17: I don''t what to tell you; I just doublechecked the Image Size and the inch measurements are what you said, but both images are 200 x 200 pixels.
When I made the images I created a new 200 pixel x 200 pixel file in Photoshop and set the DPI to either 72 or 600. I didn't do anything with inches.
posted by kirkaracha at 12:39 PM on December 10, 2001
[Ah, Metafilter — a bit more like Slashdot every day (but with fewer lawyers).]
posted by retrofut at 1:04 PM on December 10, 2001
posted by retrofut at 1:04 PM on December 10, 2001
In the case of the computer screen, it was an 11.1" monitor with a 640x480 display.
Worse, actually - it was a 9" monitor with a 512x342 display.
-Mars
posted by Mars Saxman at 1:30 PM on December 10, 2001
Worse, actually - it was a 9" monitor with a 512x342 display.
-Mars
posted by Mars Saxman at 1:30 PM on December 10, 2001
Wow. This is a truly amazing example of people taking a rather simple issue and beating the crap out of it until it pleads for mercy and promises to be really complicated if they'll just leave it alone...
I mean, it confused me a bit, at first, but it wasn't too hard to figure out: "Image is defined by pixels. Image shows up on screen as pixels. To print, we must give each pixel a size on paper, thus dpi."
posted by whatnotever at 1:50 PM on December 10, 2001
I mean, it confused me a bit, at first, but it wasn't too hard to figure out: "Image is defined by pixels. Image shows up on screen as pixels. To print, we must give each pixel a size on paper, thus dpi."
posted by whatnotever at 1:50 PM on December 10, 2001
I don't know, it seems the most of the posters on that forum were missing the point just a bit.
Certainly 72 is an arbitrary magic number, but it is a pretty useful one when you have to go from print to screen or from screen to print. Scanning at 72 DPI makes an 11 inch image a little less than 800 pixels. (792 to be precise) By all means if you are an uber-designer you calculate the space you have on the screen, then pull out your ruler to measure the width of your image, and then calculate your custom pixel to inch ratio to get the maximum desired effect (although being the standards Luddite did I am, I tend to politely suggest to novice designers that if they want pixel by pixel control of their content, life would be much easier on them if they just used a multimedia authoring system to begin with.)
Most users however really can't be bothered, and just want to know which of the scanner presets will give them a reasonable-sized image on the screen.
posted by KirkJobSluder at 2:03 PM on December 10, 2001
Certainly 72 is an arbitrary magic number, but it is a pretty useful one when you have to go from print to screen or from screen to print. Scanning at 72 DPI makes an 11 inch image a little less than 800 pixels. (792 to be precise) By all means if you are an uber-designer you calculate the space you have on the screen, then pull out your ruler to measure the width of your image, and then calculate your custom pixel to inch ratio to get the maximum desired effect (although being the standards Luddite did I am, I tend to politely suggest to novice designers that if they want pixel by pixel control of their content, life would be much easier on them if they just used a multimedia authoring system to begin with.)
Most users however really can't be bothered, and just want to know which of the scanner presets will give them a reasonable-sized image on the screen.
posted by KirkJobSluder at 2:03 PM on December 10, 2001
Exactly. On the Web, DPI is irrelevant; only the dimension of the image in pixels matters. DPI is for converting inches to pixels and vice versa. On the Web you're not dealing with inches, so the concept of DPI is completely superfluous.
posted by kindall at 2:06 PM on December 10, 2001
posted by kindall at 2:06 PM on December 10, 2001
The funny thing is that Apple was so obsessed with wysiwyg (what you see is what you get), that they didn't support multiscan monitors until, like '93 or so. "We'll tell you what resolution your monitor can use, thankyouverymuch."
posted by electro at 2:07 PM on December 10, 2001
posted by electro at 2:07 PM on December 10, 2001
Certainly 72 is an arbitrary magic number, but it is a pretty useful one when you have to go from print to screen or from screen to print.
Microsoft thought 96 was the magic number. At this dot resolution, what they considered the 640 pixel minimum display size, was equivalent to "about 6.5 inches". According to their description of The Logical Inch Problem, this is "precisely the width of text that prints on 8.5-inch-wide paper when you use the standard margins of an inch each side." (Nice mixture of the qualifiers "about" and "precisely" in this argument). It's all hocus pocus.
posted by dlewis at 2:36 PM on December 10, 2001
Microsoft thought 96 was the magic number. At this dot resolution, what they considered the 640 pixel minimum display size, was equivalent to "about 6.5 inches". According to their description of The Logical Inch Problem, this is "precisely the width of text that prints on 8.5-inch-wide paper when you use the standard margins of an inch each side." (Nice mixture of the qualifiers "about" and "precisely" in this argument). It's all hocus pocus.
posted by dlewis at 2:36 PM on December 10, 2001
It's really fun to see design-oriented Mac people and structure-oriented PC people duke it out. Like watching Architects and engineers fight about designs. Let's not hijack the thread and return to the point at hand.
When it comes down to it, *all* forms of measurement are arbitrary, but that's a whole other discussion.
grumblebee, you're not perpetuating a myth, you're just using old standards. Until a new standard is developed (as we've been learning here, there are several different ones floating around) 72ppi (pixels per inch) is fine.
posted by me3dia at 2:54 PM on December 10, 2001
When it comes down to it, *all* forms of measurement are arbitrary, but that's a whole other discussion.
grumblebee, you're not perpetuating a myth, you're just using old standards. Until a new standard is developed (as we've been learning here, there are several different ones floating around) 72ppi (pixels per inch) is fine.
posted by me3dia at 2:54 PM on December 10, 2001
I AM perpetuating a myth if I say "you should make images for the web 72dpi because otherwise their filesize will be too big and they will download slowly" since this isn't true.
I think it's okay to give in to trends when one is contemplating what clothes to wear or what sort of car to drive. But it seems a little silly when one is talking about measuring file size.
posted by grumblebee at 3:50 PM on December 10, 2001
I think it's okay to give in to trends when one is contemplating what clothes to wear or what sort of car to drive. But it seems a little silly when one is talking about measuring file size.
posted by grumblebee at 3:50 PM on December 10, 2001
One reason for the 72dpi might be for printing websites. I haven't tested it, but I imagine that your nice little 100x100 pixel clip-art is going to be the wrong size if saved at 300dpi.
I'm going to check this out...
posted by sauril at 4:10 PM on December 10, 2001
I'm going to check this out...
posted by sauril at 4:10 PM on December 10, 2001
My 2cents. Late, I know.
DPI means nothing to the screen (I think we've worked that out). I wouldn't have a clue what DPI most of my web images are tagged with. Probably 300dpi because that's what they've been scanned at. However, I've been designing some labels that are being printed out at 200dpi and need to be a particular size, so I tell Photoshop the DPI and the dimensions and it gives me a new document the right size.
From memory, all fixed resolution Mac monitors at 72dpi. whether it's a "two-page" 21" monitor, or a little Classic, they're 72dpi. The PC with it's multisync monitors makes a mockery of screen-based DPI. Take a 14" monitor at 1024x768 and of course the DPI is different from when it's at 800x600.
Other points:
"Safe" colours are somewhat meaningless in an age of JPEGs and 16/24/32 bit displays. I've been using at least a 15-bit video card since '91, so excuse me if I develop web content for more than 256 colour displays.
PAL and NTSC don't techincally have a width in pixels, they just have a line count. You can use the 4:3 ratio and the line count to pretend there's a width, but it's all analogue. With the right graphics adapter you could double the horizontal pixel count and get a much nicer looking picture. Also, the number of lines per frame and the number of lines on an interlaced TV are different. The latter is half the former. Also, there's "overscan" and "video legal". Per field, NTSC really has about 240-256 lines and PAL has 270-312.
Just to top it all off, I have a 21" monitor running at 1024x768. It therefore has a DPI of 64.
posted by krisjohn at 4:12 PM on December 10, 2001
DPI means nothing to the screen (I think we've worked that out). I wouldn't have a clue what DPI most of my web images are tagged with. Probably 300dpi because that's what they've been scanned at. However, I've been designing some labels that are being printed out at 200dpi and need to be a particular size, so I tell Photoshop the DPI and the dimensions and it gives me a new document the right size.
From memory, all fixed resolution Mac monitors at 72dpi. whether it's a "two-page" 21" monitor, or a little Classic, they're 72dpi. The PC with it's multisync monitors makes a mockery of screen-based DPI. Take a 14" monitor at 1024x768 and of course the DPI is different from when it's at 800x600.
Other points:
"Safe" colours are somewhat meaningless in an age of JPEGs and 16/24/32 bit displays. I've been using at least a 15-bit video card since '91, so excuse me if I develop web content for more than 256 colour displays.
PAL and NTSC don't techincally have a width in pixels, they just have a line count. You can use the 4:3 ratio and the line count to pretend there's a width, but it's all analogue. With the right graphics adapter you could double the horizontal pixel count and get a much nicer looking picture. Also, the number of lines per frame and the number of lines on an interlaced TV are different. The latter is half the former. Also, there's "overscan" and "video legal". Per field, NTSC really has about 240-256 lines and PAL has 270-312.
Just to top it all off, I have a 21" monitor running at 1024x768. It therefore has a DPI of 64.
posted by krisjohn at 4:12 PM on December 10, 2001
One more thing. A favourite trick of mine is to create, lets say, a 200x200 pixel image, but specify it's size on a web page as 100x100. When you print the page, the picture looks a lot smoother than if it was only 100x100. Also, if it's a JPEG you can reduce the quality further than normal and it still looks better on both the screen and the printed page.
posted by krisjohn at 4:16 PM on December 10, 2001
posted by krisjohn at 4:16 PM on December 10, 2001
Hmmm, my photoshop can only change ppi, not dpi, so I guess my test isn't valid. Anyone else?
Maybe the "entrenched myth" is actually ppi? Then that would make sense.
posted by sauril at 4:17 PM on December 10, 2001
Maybe the "entrenched myth" is actually ppi? Then that would make sense.
posted by sauril at 4:17 PM on December 10, 2001
Most of you will probably think I'm beating a dead horse, but I was a little disturbed by kirkaracha's experiment. So I repeated it and got results in keeping with the "dpi isn't important" notion. If you're interested, check here:
http://www.grumblebee.com/Waste/dpi/dpi.html
I know some of you have been royally irritated by this thread, but it's been a real learning experince for me and I thank all involved.
Sauril, ppi is just Adobe's term for dpi. They think it's more appropriate since while in their app you're dealing with pixels, not dots. Whatever.
posted by grumblebee at 5:16 PM on December 10, 2001
http://www.grumblebee.com/Waste/dpi/dpi.html
I know some of you have been royally irritated by this thread, but it's been a real learning experince for me and I thank all involved.
Sauril, ppi is just Adobe's term for dpi. They think it's more appropriate since while in their app you're dealing with pixels, not dots. Whatever.
posted by grumblebee at 5:16 PM on December 10, 2001
Wow
I'm just shocked that people could believe that DPI has anything to do the way they look on the screen. I mean it really is amazing, how stupid some people can be.
DPI has zero effect on the way something looks on screen. It never has.
It also has no effect on the way graphics look when printed out. At least, not in IE6, anyway.
posted by delmoi at 8:26 PM on December 10, 2001
I'm just shocked that people could believe that DPI has anything to do the way they look on the screen. I mean it really is amazing, how stupid some people can be.
DPI has zero effect on the way something looks on screen. It never has.
It also has no effect on the way graphics look when printed out. At least, not in IE6, anyway.
posted by delmoi at 8:26 PM on December 10, 2001
Apple's supported multisync (and SVGA) for more than half a decade, and the default ppi has been creeping up to about 86 or 92, iirc, on the newest iBooks.
posted by retrofut at 12:53 PM on December 11, 2001
posted by retrofut at 12:53 PM on December 11, 2001
« Older Cancer Research Project | Travel Makes You Stupid. Newer »
This thread has been archived and is closed to new comments
posted by grumblebee at 9:08 AM on December 10, 2001