# picture dpi's

**URL:** <https://boards.straightdope.com/t/picture-dpis/641447>\
**Category:** Factual Questions\
**Created:** [November 18, 2012, 10:10pm UTC](https://boards.straightdope.com/t/picture-dpis/641447 "2012-11-18T22:10:26Z")\
**Posts on this page:** 15\
**Page:** 2

<div class="post-metadata">

**Author:** ![Colophon](https://avatars.discourse-cdn.com/v4/letter/c/f05b48/32.png) [@Colophon](https://boards.straightdope.com/u/Colophon)\
**Post date:** [November 19, 2012, 3:33pm UTC](https://boards.straightdope.com/t/picture-dpis/641447/21 "2012-11-19T15:33:00Z")

</div>

As (almost) everyone else said, DPI is meaningless for a digital file. The file just has pixel dimensions. If you display an 1800 x 1200 file on a monitor measuring 36in x 24in then it’s 50dpi (or strictly ppi). If you display it on an arena-sized jumbotron it might be 10dpi or less.

Similarly, you could print out that same image at 300dpi and get a 6in x 4in print.

The dpi tag doesn’t change the resolution at all. Open the file in a graphics program, double the dpi and you’ll see the linear size halve. If you want to print bigger at a given dpi then you’ll have to resize (and therefore resample) the file.

Someone who works on a magazine really should know this (and should know that 72dpi AT PRINTED SIZE is pretty crappy resolution).

---

<div class="post-metadata">

**Author:** ![pulykamell](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/pulykamell/32/3166_2.png) [@pulykamell](https://boards.straightdope.com/u/pulykamell)\
**Post date:** [November 19, 2012, 4:27pm UTC](https://boards.straightdope.com/t/picture-dpis/641447/22 "2012-11-19T16:27:30Z")

</div>

Oh, and I didn’t realize that an update from the OP said this was for a printed magazine. That makes the 72 dpi (which is a bit of an outdated monitor standard) request even weirder. I’ve had magazines ask for 200/240/300 dpi at a specific size way back when I still did magazine work (and this was pre-digital capture, but while film was scanned and transmitted digitally), but never 72 dpi for print usage.

---

<div class="post-metadata">

**Author:** ![Ximenean](https://avatars.discourse-cdn.com/v4/letter/x/aca169/32.png) [@Ximenean](https://boards.straightdope.com/u/Ximenean)\
**Post date:** [November 19, 2012, 4:41pm UTC](https://boards.straightdope.com/t/picture-dpis/641447/23 "2012-11-19T16:41:29Z")

</div>

So you guys are saying that DPI doesn’t mean anything, in relation to JPEGs?

---

<div class="post-metadata">

**Author:** ![billfish678](https://avatars.discourse-cdn.com/v4/letter/b/7bcc69/32.png) [@billfish678](https://boards.straightdope.com/u/billfish678)\
**Post date:** [November 19, 2012, 4:45pm UTC](https://boards.straightdope.com/t/picture-dpis/641447/24 "2012-11-19T16:45:37Z")

</div>

Not unless the definition of DPI when it comes to JPEGS (or any other format for that matter) includes some universally understood printed or displayed size (like 8 by 10 perhaps?).

I need you to get me some gas for my trip. I get 25 miles to the gallon.

---

<div class="post-metadata">

**Author:** ![Musicat](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/musicat/32/20189_2.png) [@Musicat](https://boards.straightdope.com/u/Musicat)\
**Post date:** [November 19, 2012, 5:44pm UTC](https://boards.straightdope.com/t/picture-dpis/641447/25 "2012-11-19T17:44:29Z")

</div>

> [@Ximenean](#):
>
> So you guys are saying that DPI doesn’t mean anything, in relation to JPEGs?

DPI is going to affect simple screen display (although browsers nowadays can reduce and enlarge pretty well), but have no effect at all on printing. The print layout is set by inches or centimeters, and the pixels are either squeezed or expanded to fit. You have a 100 pixel-wide image and you print it in 5 inches, it will be expanded (and look crappy, since the effective DPI is 20). If you put the same 100 pixel-wide image in 1/2 inch space, it will have an effective 200DPI, and look pretty good. It matters not a bit what the original scanning or storage setting was.

And “in relation to JPGs” has no meaning. JPG is a bitmapped storage format just like many others. What I am saying here relates to any bitmapped format.

---

<div class="post-metadata">

**Author:** ![Duckster](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/duckster/32/1244_2.png) [@Duckster](https://boards.straightdope.com/u/Duckster)\
**Post date:** [November 19, 2012, 7:48pm UTC](https://boards.straightdope.com/t/picture-dpis/641447/26 "2012-11-19T19:48:30Z")

</div>

> [@pulykamell](#):
>
> 36MP (not megabyte) is not meaningless without context. In my case, it refers to 7360 x 4912 pixels.

Me bad. I thought I saw MB. :smack:  
For the OP, the magazine should have provided to you their technical requirements for digital images. As an example here are the [National Geographic requirements](http://ngm.nationalgeographic.com/your-shot/rules) for people desiring to submit their own images. DPI isn’t mentioned at all. OTOH, here are the tech requirements for [Our USA Magazine](http://ourusamagazine.com/faq/photo_requirements) (picked in a random Google search).

---

<div class="post-metadata">

**Author:** ![Blakeyrat](https://avatars.discourse-cdn.com/v4/letter/b/ecd19e/32.png) [@Blakeyrat](https://boards.straightdope.com/u/Blakeyrat)\
**Post date:** [November 19, 2012, 10:36pm UTC](https://boards.straightdope.com/t/picture-dpis/641447/27 "2012-11-19T22:36:18Z")

</div>

> [@Ximenean](#):
>
> So you guys are saying that DPI doesn’t mean anything, in relation to JPEGs?

I got the joke, even if nobody else did.

---

<div class="post-metadata">

**Author:** ![yohan\_go](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/yohan_go/32/3620_2.png) [@yohan\_go](https://boards.straightdope.com/u/yohan_go)\
**Post date:** [November 20, 2012, 11:40am UTC](https://boards.straightdope.com/t/picture-dpis/641447/28 "2012-11-20T11:40:10Z")

</div>

@ OP. Easiest way (if you use win 7) is right click on jpg file - \> Properties - \> Details. You do not need special software for that. You could increase your DPI output with Photshop or free Irfanview, but I doubt that you will notice improvement in visual quality as imput quality is so low.

---

<div class="post-metadata">

**Author:** ![Musicat](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/musicat/32/20189_2.png) [@Musicat](https://boards.straightdope.com/u/Musicat)\
**Post date:** [November 20, 2012, 11:47am UTC](https://boards.straightdope.com/t/picture-dpis/641447/29 "2012-11-20T11:47:57Z")

</div>

> [@yohan\_go](#):
>
> You could increase your DPI output with Phot[o]shop or free Irfanview, but I doubt that you will notice improvement in visual quality as imput quality is so low.

What does “increase your DPI output” mean? Boost the number of pixels? Or just recompute the relationship between DPI and inches? Neither will have any positive effect on the outcome for printing and resampling may degrade the image and increase the file size.

---

<div class="post-metadata">

**Author:** ![Reply](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/reply/32/15952_2.png) [@Reply](https://boards.straightdope.com/u/Reply)\
**Post date:** [November 20, 2012, 12:05pm UTC](https://boards.straightdope.com/t/picture-dpis/641447/30 "2012-11-20T12:05:52Z")

</div>

> [@yohan\_go](#):
>
> @ OP. Easiest way (if you use win 7) is right click on jpg file - \> Properties - \> Details. You do not need special software for that. You could increase your DPI output with Photshop or free Irfanview, but I doubt that you will notice improvement in visual quality as imput quality is so low.

No, no, no! Once again, you physically _cannot_ have an improvement in visual quality by artificially raising the DPI. It’s a mathematical impossibility.

There is some sort of widespread misunderstanding about what DPI is and how it actually works. The DPI tag without a specified print size is a meaningless _lie_, put there by Adobe asshats to confuse new graphics people and then later copied by digital camera manufacturers because they are colluding idiots.

To repeat: **DPI only makes sense when you are comparing pixels to real-world dimensions**. The DPI tag in a JPEG is a arbitrary default, a suggestion, a lie, and totally meaningless without a specified size.

If you want a better-looking picture, you need to run it through a fancy resampling algorithm (which is what Photoshop’s resizing function does). You can’t just randomly tell Windows to raise the DPI because that’s akin to saying “I want to blow my passport picture up to 50 square miles!” It doesn’t work like that; there’s just not enough detail in that picture and it’s going to come out as a bunch of unidentifiable scattered dots even if you tried.

**You use DPI combined with a print size as a way of saying “This digital photo will look best printed at this print size” (for a given display medium and a given viewing distance). It does NOT tell you how much resolution is in the digital photo to begin with (but you could calculate it if you had _both_ the DPI and the print size). It it is NOT a number you can edit to magically improve image quality. And it does not, not, NOT mean ANYTHING when it’s embedded in a JPEG by itself with no physical dimensions specified.** Please, please, please understand this if you ever work in graphics or photography.

> [@](#):
>
> From [Say No to 72 dpi - Continued](http://www.scantips.com/no72dpib.html)  
> Dpi is only used for printing on paper (pixels per inch), and the camera has absolutely no clue what size we may print the image on paper. So the early thinking was that there is no point of their JPG files guessing some dpi value. So they didn’t, it would be pointless. But then Adobe tells us 72 dpi again (until we change it). And today, images are so large, so now this Adobe 72 dpi value causes the printed size to be shown as maybe 5x3 feet (at 72 dpi), which is totally unreal. So to prevent that, newer cameras started saving a phoney dpi guess, maybe 180 to 300 dpi, merely to show a more reasonable reasonable size, maybe one foot, instead of 4 or 5 feet at 72 dpi, which only caused bewilderment. But the camera still has no clue what size we may print it, and digital images are dimensioned in pixels, not inches. It can be any dpi number at will, the meaning of which is “pixels per inch”, which will compute some printed size in inches, but it simply does not matter what it is until we go to print it.

---

<div class="post-metadata">

**Author:** ![yohan\_go](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/yohan_go/32/3620_2.png) [@yohan\_go](https://boards.straightdope.com/u/yohan_go)\
**Post date:** [November 20, 2012, 3:40pm UTC](https://boards.straightdope.com/t/picture-dpis/641447/31 "2012-11-20T15:40:39Z")

</div>

Yes, I know that, you know that, but some editors don’t. In fact I once did just that (increased DPI from 100 to 300) just to get one such person off my dick (he sent me standard screen captured PNG file, he wanted converted to 300 dpi jpeg, and he insisted). I just say it is possible, not that it is useful.

---

<div class="post-metadata">

**Author:** ![billfish678](https://avatars.discourse-cdn.com/v4/letter/b/7bcc69/32.png) [@billfish678](https://boards.straightdope.com/u/billfish678)\
**Post date:** [November 20, 2012, 4:12pm UTC](https://boards.straightdope.com/t/picture-dpis/641447/32 "2012-11-20T16:12:17Z")

</div>

One related note. In the real world, unlike those CSI kinda shows, the number of pixels (or “DPI”) you got is all you are gonna get.

So, it can be the case that your pixel count is a bit too low and or you are trying to print/display the image a bit too big and the pixels will show.

Now, if it is the only picture you’ve got and you need to make it work there is one thing you can do that sorta helps. You “resample” your pictures into a larger picture. You map one pixel into 4 or 9 or 16 pixels say. Then, you take that larger image and play around with the smoothing and sharpening filters on your image editor. When doing this I find it best to take a small part of the original image and play with it first to make it look best. Like someones eye or a leaf or other small detail where you know what it is supposed to look like. And I do that with the image significantly bigger than what the final image will be displayed at. Once you figure out what to do to make that look best, then you go back and do that process to the whole image

Of course the resolution of this new image will not be any better than the old image. However, it only be somewhat poor in resolution but smooth, rather than somewhat poor in resolution and pixelly (which IMO is way worse).

I’ve salvaged a couple of marginal but otherwise good images that way.

---

<div class="post-metadata">

**Author:** ![septimus](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/septimus/32/410_2.png) [@septimus](https://boards.straightdope.com/u/septimus)\
**Post date:** [November 20, 2012, 5:05pm UTC](https://boards.straightdope.com/t/picture-dpis/641447/33 "2012-11-20T17:05:18Z")

</div>

> [@billfish678](#):
>
> One related note. In the real world, unlike those CSI kinda shows, the number of pixels (or “DPI”) you got is all you are gonna get.
> 
> … You “resample” your pictures into a larger picture. You map one pixel into 4 or 9 or 16 pixels say. Then, you take that larger image and play around with the smoothing and sharpening filters on your image editor…

I think this is essentially correct, but slightly misleading. The effective resolution of a digital image _may_ be increased with special procedures, as **billfish678** ’s own example implies. IIRC, an expert once told me that doubling the resolution was about the best one could hope for with a typical image. Bigger increases may be possible in special cases, though of course nothing like the absurdities seen in CSI.

I once attended lectures by the imaging professor who served as expert witness in the retrial of the Rodney King-kicking cops. Blurriness in the home video may have been part of the reason for acquittal in the first trial; the feds definitely wanted conviction in the retrial so hired this imaging expert to enhance the video. Histogram equalization was one technique he used. (I always wondered if an expert counter-witness could have convinced the jury that such techniques were as hokey as reading tea leaves. :dubious: :rolleyes: )

---

<div class="post-metadata">

**Author:** ![Reply](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/reply/32/15952_2.png) [@Reply](https://boards.straightdope.com/u/Reply)\
**Post date:** [November 20, 2012, 9:25pm UTC](https://boards.straightdope.com/t/picture-dpis/641447/34 "2012-11-20T21:25:07Z")

</div>

> [@septimus](#):
>
> I think this is essentially correct, but slightly misleading. The effective resolution of a digital image _may_ be increased with special procedures, as **billfish678** ’s own example implies. IIRC, an expert once told me that doubling the resolution was about the best one could hope for with a typical image. Bigger increases may be possible in special cases, though of course nothing like the absurdities seen in CSI.
> 
> I once attended lectures by the imaging professor who served as expert witness in the retrial of the Rodney King-kicking cops. Blurriness in the home video may have been part of the reason for acquittal in the first trial; the feds definitely wanted conviction in the retrial so hired this imaging expert to enhance the video. Histogram equalization was one technique he used. (I always wondered if an expert counter-witness could have convinced the jury that such techniques were as hokey as reading tea leaves. :dubious: :rolleyes: )

There is a little more leeway with video. If you have sufficient similarities between frames, you can try techniques like [video super-resolution](http://xcorr.net/2011/09/15/video-super-resolution-coming-to-consumer-software/) by extrapolating and combining information from redundant frames. “Histogram equalization” sounds like a way to normalize the contrast (between dark and light areas of an image) through multiple frames so that you get roughly similar brightness through the whole clip OR less extreme contrast within one frame. IOW it’s probably a way of brightening dark parts of the image to make them clearer, but it’s not related to the magical zoom function in CSI – unless there’s some other meaning to histogram equalization that I don’t know about.

On the other hand, it’s hard to see how you might increase the actual resolution of a single digital photo because the information in it is all you got. You can certainly do tricks to make it _look_ clearer to the human eye, but that’s not increasing information density, that’s adapting color, geometry, and contrast to better suit viewers – it works on the brain, not the picture.

---

<div class="post-metadata">

**Author:** ![curlcoat](https://avatars.discourse-cdn.com/v4/letter/c/ea5d25/32.png) [@curlcoat](https://boards.straightdope.com/u/curlcoat)\
**Post date:** [November 21, 2012, 1:44am UTC](https://boards.straightdope.com/t/picture-dpis/641447/35 "2012-11-21T01:44:27Z")

</div>

> [@yohan\_go](#):
>
> @ OP. Easiest way (if you use win 7)

Perish the thought! _has vapors_ I’m Mac all the way, with PSE and something called Image Browser that came with the camera.

> [@](#):
>
> is right click on jpg file - \> Properties - \> Details. You do not need special software for that. You could increase your DPI output with Photshop or free Irfanview, but I doubt that you will notice improvement in visual quality as imput quality is so low.

Since I was getting all the info on what the publisher needed second hand, I just emailed the full size images (about 2800 x 2000). I figure they must have the software on hand to do whatever they want.

[Previous page](https://boards.straightdope.com/t/picture-dpis/641447.md?page=1)
