22

I have two cameras, a 6mpix Panasonic FZ8, and 14mpix Canon A2200.

When I compare shots from two cameras, one of differences I notice is
dpi number that camera reports in JPG info:

  • FZ8 reports dpi=72
  • A2200 reports dpi=180

What physical meaning can these numbers have? I am at loss for guesses.

I am fairly familiar with notion of dpi in scanning and printing. I can calculate density of pixels on the sensor of the camera. But then, the linear density of pixels on the sensor will be hundreds times larger than number above. So what, if anything, does it mean?

21

The values written in JPEG files are arbitrary and essentially meaningless. They don't relate to anything about the camera, its sensor, or the resulting images. They certainly don't relate to image quality or acceptable resolution for printing. Really, they mostly serve to confuse people.

The EXIF standard seems to imply that if the tag is missing, 72 is the (still-meaningless) default. However, it is apparently mandatory for the TIFF standard, from which the JPEG/EXIF format basically inherits everything. So maybe it has to have some value to properly comply with the standard.

Others have noted that some desktop publishing or word processing software reads this value and will use it for default scaling on the page. So, I guess in that sense, there is "meaning", but I'd argue that this is really misapplication, because the original value doesn't have meaning. Garbage in, garbage out, as the saying goes.

Now, maybe in an alternate universe the standard could relate to a standard print size. Or, the camera could estimate real detail in the image and give a recommended maximum print size. But none of that is the case. In practice, these values are meaningless and you should ignore them, even if some software makes assumptions based on them when opening files.

  • 6
    I realize now that for images coming from scanner, dpi field carries real meaning, so that's why that field is there. – Andrei Apr 22 '11 at 19:55
  • 2
    I think the 72 dpi number comes from the CRT days. AFAIR, this was the average dot pitch of CRT monitors back then, so it is not totally meaningless. – ysap Jul 1 '11 at 15:32
  • 3
    Sure, not totally arbitrary in origin, but I'm going to stick to saying that it's meaningless in this context. – mattdm Jul 1 '11 at 15:41
  • 1
    For a historical perspective -- in print layout workflows a dpi number is used (or is supposed to be used) as a guide to how large the image can be on the printed page. If you set a 200 dpi value (common for newspapers) for a 2000x1000 pixel image the person placing the image in the page layout software (who is normally not a photographer) would be told that the image is 10x5 inches. The same image with a dpi set to 72 would seem to be about 28x14 inches -- but if the editor actually used it at that size the image would appear pixelated. – David Rouse Jul 1 '11 at 17:27
  • 2
    As far as today's digital cameras go and the workflow of most photographers -- right, the numbers are random and bogus. I was just pointing out the use of the dpi values in the print publishing industry to show why they exist at all. Unless I'm terribly mistaken metadata attached to digital images pre-dates DSLR cameras and the layout of the original metadata was designed for news services, like the Associated Press. – David Rouse Jul 1 '11 at 18:47
4

In my experience those numbers have no special meaning. The camera makers just pick one and use it in all their firmware.

  • I just noticed that the Olympus OM-D E-M5 has a menu option to select your own arbitrary value. Given everything else I'm not sure that's really a useful option, but there it is! – mattdm Jun 22 '12 at 21:05
4

The number is just a random filler. It has no significance since the camera does not know how big you will print.

Most cameras default to 72 which according to the EXIF standard is the default value. Some cameras let you set it yourself. Then again, it has little meaning unless you will print all your images without cropping exactly at the same size.

  • Print, or view on-screen for that matter. – mattdm Apr 22 '11 at 19:14
  • I don't think printing is where the DPI should (theoretically) resolve from. This is a photo, so the dimensions of the field-of-view of the camera, divided by the resolution, would be the actual dpi of the photo. – Fake Name Apr 23 '11 at 11:44
  • Therefore, it seems to me that if you know the angle of view of the camera's lens, and the camera had some manner of ranging facility (ultrasonic?) to measure the distance to what it is pointing at, the values could be populated with a (rough, perhaps) meaningful value. – Fake Name Apr 23 '11 at 11:45
  • @Fake Name - How about just using the focus-distance? As a matter of fact, you can probably write a small program that does this after the fact if your camera stores the focus-distance in the EXIF (only some do unfortunately). – Itai Apr 23 '11 at 13:02
  • @Fake Name - but in a typical image, each pixel records an object at a different depth. So, the conversion will be correct only at a small amount of pixels? Do you suggest that for the same lens-sensor configuration (FoV), two images of a person in different distances will produce different dpi values? I think the best that you can do following your idea is to state the Dots-per-Radian (DPR) or -Degree (DPD) which is the angular spatial resolution of your system. – ysap Jul 1 '11 at 15:39
2

If you consider the JPEG file as a document, the divide the pixel count in the image by the DPI number and you get the Print Size of the image. This of course results meaningless as many printing systems re-size the image arbitrarily. However, there still are printing mechanisms in which the print size is given by such parameters. If you have Photoshop, go to the resize image dialog and uncheck the re-sample option. You'll observe that changing any size of the image will change the dpi number. Curiously I'm in the case where I have to print some images and the lab providing the service won't do ANY resizing for me, so I have to re-sample the images to a specific resolution and DPI number. Another use of this number is that it will allow a better approximation of how big you can print an image without distorting too much for your working parameters. So, in your case, your cameras are just setting a default print size for your images. But unless your printing service relies on this for the size, I see no other use for it and it says nothing about your sensor, so a comparison there is meaningless.

1

What physical meaning can these numbers have?

The EXIF standard doesn't actually have a single dpi field, but it does have XResolution and YResolution fields, and these are where the dpi value displayed by most software comes from. The standard doesn't say specifically whether those values apply to the image source, e.g. a scanned document, or destination, e.g. a printed copy of the image, but since the entity that writes the metadata knows where the image comes from but can't know how it will be used, the only reasonable interpretation is that those fields relate to the image source.

The standard does say that if the value for the XResolution and YResolution tags is unknown, then 72 should be used. One could reasonably argue that this was a poor choice, since 72 is a valid value whose meaning shouldn't be overloaded. It probably would've made more sense to choose a value that couldn't possibly represent a real resolution, such as 0 or -1, to mean "unknown." But that's not what the standard says, and so if your image has 72 for the resolution fields, you simply can't know whether the metadata writer knew what resolution to use or not.

The problem is compounded by the fact that some cameras set a value other than 72 when they can't really know what resolution the subject was recorded at, since that varies depending on how far away the camera was from he subject. Those camera manufacturers might have been trying to use the resolution fields to suggest a reasonable print resolution, but this only confuses the meaning of the resolution fields.

If you're trying to actually learn something about an image, you'll be better off using the FocalPlaneXResolution and FocalPlaneYResolution fields, which record the sensor resolution at which the image was recorded. And of course the ImageWidth and ImageLength fields will tell you the dimensions of the image in pixels.

-2

Does the dpi number reported by camera in JPG have any meaning?

Yes. It means the jpeg image is compliant with the International Standards agreed upon by camera makers and others with an interest in the production and use of digital images.

In most current digital environments the DPI reported by the camera in the image metadata has no real meaning.

The applications we use to look at a photo on our screens will almost always scale an image to fit a particular space on the screen.

The printers most of us use will almost always automatically scale an image to fit on a specified size of paper.

But there is still one notable exception: Desktop publishing. Page setting software such as InDesign and even most word processing programs still "think" in actual physical dimensions. That's why such applications usually show a scale at the top and/or sides of the page that shows the width and length of the page in inches or centimeters.

If we have two version of an image that is 720 pixels wide and the only difference between the two is the "PPI" reported in the image metadata, here's what happens when we insert both into a document that measures things in real inches/centimeters:

enter image description here

Remember, the actual pixels in both images are identical, the only difference is in the "DPI" value reported by the images metadata. In this case the reported "PPI" does affect how large the image appears on a "page" in the document or on an actual page when the document printed.

  • 2
    the values written in the field by the camera are still nonsensical and apparently completely arbitrary (comparing the two from the OP) considering that the camera can't know how large you'd like to print the image. – ths Jun 14 '18 at 10:47
  • The fact that your desktop publishing software tries to interpret the resolution tag in a useful way doesn't mean that the meaning it chooses corresponds to the intent of whatever wrote the image. Furthermore, it's entirely possible that two different desktop publishing programs could choose different interpretations of the value for the very same photo. You may not agree that the meaning of the XResolution and YResolution tags are meaningless, but you have to admit that the tags are at least ambiguous, and that requiring them is a flaw in the EXIF standard. – Caleb Jun 14 '18 at 11:52
  • @Calab You can believe that they are ambiguous, or that requiring them is a flaw in the EXIF standard, but you still must admit that they are required by that standard as well as the fact that if no value is present there's nothing for applications (designed specifically to work with compliant files) to interpret. – Michael C Jun 14 '18 at 22:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.