Photos without EXIF data, is it possible to tell is they are taken by an APS-C or full frame sensor.

<p>Hello I am administrating a photo competition. One of the requirements of the competition is that the photo has been taken with an aps-c sensor or full frame sensor. Some of the entries have been supplied without EXIF data and having contacted the owners they no longer have the original files. As it stands I will have to exclude some of these photos, which is a shame as they are very good photos. <br>

<strong>PLEASE DO NOT POST BACK that this rule is stupid, I have been through this rule with the competition organisers several times and they are not going to change this rule, so posting this will not help me or the competition entrants. </strong></p>

<p>I suppose the question is, do the entrants have to prove that their pictures were taken with certain cameras, or is it up to you to prove that they weren't? You seem to be taking the first route. The other approach is to take the entrants' word for it. And if they are lying, well, so what? Is it worth worrying about?</p>

<p>Cheers</p>

<p>Alan</p>

 

<p><strong>I suppose the question is, do the entrants have to prove that their pictures were taken with certain cameras, or is it up to you to prove that they weren't? You seem to be taking the first route. The other approach is to take the entrants' word for it. And if they are lying, well, so what? Is it worth worrying about?</strong><br>

<strong>Cheers</strong><br>

<strong>Alan</strong><br>

As I have stated in the initial post I have not set the rules and the rules are not open to be changed. Discussion of changing the rules is not helpful to either me or the entrants.<strong><br /></strong></p>

 

<p>Back to the original question ... I am not aware of any technique to determine the type of sensor without the EXIF. Of course, the EXIF could be modified to state it came from the type of sensor.</p>

<p><strong>Back to the original question ... I am not aware of any technique to determine the type of sensor without the EXIF. Of course, the EXIF could be modified to state it came from the type of sensor.</strong><br>

Is there any threshold that a picture taken with an APS-C or full frame sensor would not fall below <strong>?</strong> <strong><br /></strong></p>

<p>For photos such as those for your contest, there is no way to tell. Even the entries that you receive with EXIF data indicating that they were taken with the required sensor type could have had their EXIF data edited.</p>

<p>A very large depth-of-field can be a hint that a shot was taken with a small digicam or cell phone camera. I read somewhere that a digicam renders a DOF like a APS-C (or full-frame?) dSLR at an aperture 5-stops smaller (i.e., f/2 looks like f/11). Of course, some people actually shoot at very small apertures and background blur can be manipulated by post-processing...</p>

<p>So without the original EXIF nothing is certain... But keep in mind that EXIF data can easily be changed, manipulated or faked.</p>

<p>There is just no way to do this. With the EXIF data gone, you can of course look at the resolution of the picture, but that doesn't tell you anything - first you can resize the pictures, and second, there are P&S cameras which create higher resolution files than the first generation of full frame digital cameras did.<br>

The rule is utterly stupid, but as you said....saying that is not going to help one bit.</p>

<p>If the exif data is missing there is no way to tell what sensor was used.<br>

If exif data is present there is no way to confirm what sensor was used as exif data is user editable.<br>

You say the rules can't be changed, fine but rules are open to interpretation. You do not provide the exact language of the rule in question but based on the summary you provide, a strict interpretation would eliminate all Olympus dSLRs, all Canon 1D series dSLRs, all MF digital cameras and probably some other models that don't have APS-C or FF sensors.<br>

You could also take the position that everyone read the rules and is basically honest therefore all entries meet the rule.</p>

<p>From the answers all ready given my advice is to trust the information given by the entrant. A lot of image manipulation programs will strip the EXIF data off and could make the photo invalid if you decide that without this data it can't be used in the competition.</p>

<p><strong>from the answers all ready given my advice is to trust the information given by the entrant.</strong><br>

It's not an issue of trust the competition rules state that you must be able to supply evidence of meaning this requirement. As I mentioned before this is not helping me or the competition entrants.<br>

If there is no other option I will have to exclude these entrants.<br>

Can I confirm one final time. There is no other metric which would supply this information, DPI for example?</p>

 

<p>The OPs original question was simple and the answer is no.</p>

<blockquote>

<p>Can I confirm one final time. There is no other metric which would supply this information, DPI for example?</p>

</blockquote>

<p>The answer to this follow up is also no, dpi would not reveal anything.<br>

If they are submitting the original capture at full size, and are required to do so, and that capture comes from a camera that is the only one on the market that creates files with those pixel dimensions (not dpi), that would be one way you could determine the camera and therefore the sensor.</p>

 

<blockquote>

<p>but based on the summary you provide, a strict interpretation would eliminate all Olympus dSLRs, all Canon 1D series dSLRs</p>

 

</blockquote>

<p>This poster has a good point. I assume the rule is there to limit it to more professional-type images, but if you don't include aps-h it's a big oversight.</p>

<p>Steve, with all respect I don't think people are missing your point, but they have added an insight that affects what you want, which is to say: not only can the non-EXIF people not validate the sensor they used, you cannot validate ANYONE's sensor that they used as it is easily modified. The logic of this is that if you are trusting the participant in ALL entries- not verifying anything yourself. I understand you may not be interested in that, that's fine, but the point raised seems essential to what you are trying to determine.<br>

Good luck, sounds like a bummer situation.</p>

<p>IMO the rule is stupid, but as you say, that's beside the point. I think this would be a situation similar to trying to establish whether a photo had been altered. It all comes down to showing the original file or negative. The entrant should be able to do that. (One NEVER throws away negatives or original files!) If they can't do that, then they're out of luck, probably as it should be.</p>

<p>... but I hate these sorts of rules.</p>

<p><strong>When there is a will, there is a way.</strong><br>

<strong> </strong><br>

<strong>Choose a place like a garden, zoo, a farm, a river bank,.. whatever. Every contestant will be there from 6am to 6pm and registers his/her camera and memory cards with the organizing people. There will be people watching them for cheating</strong><br>

<strong> </strong><br>

<strong>PLEASE DO NOT say that this way is stupid, I have been through several competitions. Like in a violin (piano, guitar,...) competition, will they accept recorded CDs or tapes? NOO, every contestant has to perform right there LIVE</strong></p>

<p>In the old days we used to try to 'scan' a binary file, like an executable file *.exe by loading it up into simple word processors like WordPad (Start => Programs => Accessories => WordPad). All the binary pieces of data show up like 'gobbly gook' (ie, 'cartoon cusswords').But any readable pieces of text show up as well. It was easy then to peruse down through the binary contents looking at all the internal text strings, which sometimes gave you useful information.</p>

<p>You can try that same technique using a picture file. Just make a scratch copy of the picture file you want to examine. (Working off a scratch copy will preserve the original.) Get on a Windows box and run WordPad. Then use the ordinary File => Open menu to locate and load up the picture file. Again, all the binary data will look like garbage. But there should be some text strings imbedded in the picture file that will show up like readable text. At the very least you should see the text fields for the Make and Model of the camera, and possibly the firmware level.</p>

<p>But you would still have to know which cameras employ the 'correct' sensor types and which ones do not.</p>

<p>The answer to your question is....NO</p>

<p>Without EXIF data there is no way to tell. Actually with EXIF data there is no way to tell either since EXIF data can be changed if you know what you are doing.</p>

<p>If I were in your situation I'd go with Sarah Fox's suggestion. If they can't provide the original file then you don't have anything to rule with or against. If they do provide the original file and they did indeed edit the exif on the original file to fit what you're asking then they slip by.</p>

<p>I think the bigger issue here is this. If you don't have the original file, how do you as administrator of this competition know that these are actually that photographers images? You do ask for submissions from the actual photographers don't you? You may want to run a few suspect images through http://www.tineye.com but if they can't provide original files then that means you can't verify that they are the photographer of said images.</p>

<p>No exif, no original, no way of verifying anything, no entry.</p>

<p>The short answer is no, but the long answer is, probably.<br>

Each sensor is different. Sensitivity to color is different. The dynamic ranges are different. The pixel size and number of pixels are different. Even the pixel arrangements are different. Big sensors are better at gathering light, and are better with small apertures (less diffraction).<br>

Each lens is different. The color casts are different, the bokeh is different, the focal length and apertures are different, the optical formulas are different, and these all affect the picture.<br>

Each camera firmware is different. Some adjust contrast and color for you automatically, and the ones that do use different formulas. The firmware that turns the raw data into JPEG are different.<br>

All of this information forms a unique fingerprint for each camera. It should be possible to identify not just whether it was a DSLR, but exactly which make and model took the picture, or at least narrow the field down significantly.<br>

You might ask the FBI. They may have digital forensics tools that could help identify the camera.</p>

About using EXIF data in original files (raw, I suppose) for verification, well EXIF in those can be modified too. For example, exiftool currently cannot edit data in Sony raw files but can others such as CR2, CRW, DNG, etc.

<blockquote>

<p>If there is no other option I will have to exclude these entrants.</p>

</blockquote>

<p>Since, as several of us have already said, EXIF data can be edited, it seems that you would have to exclude all the entrants, since none have "proof" that they used the appropriate type of camera. As you repeatedly insisted, you have no flexibility of interpretation, so just throw them all out.</p>

<p>I suppose the only kind of image which would stand a chance of being accepted would be one such as the one I attached, that includes the photographer, his name tag, and the camera appearing in a mirror in the right-hand corner. However, you have only my word that I didn't add the camera and me in Photoshop.</p><div></div>

<p>Hector, it looks like you are the winner, because you are the only one get accepted and proved not only by which camera the picture was taken but also by whom. However, do you have a proof that your picture has not been doctored?</p>

<p>As of me, I'm not cleared about the rule yet, is an Olympus 4/3 a full-frame camera? is a Canon EOS 1D an APS-C or full frame camera? is the M9 accepted because it is full frame but not SLR?</p>

<p>@Derek - Once the image is converted to a JPG, all mosaic-related information (ie, about the layout and other aspects of the sensor) is gone, so this is not a fruitful avenue for validation. Similarly, determining a set of color sensitivity curves for a particular camera is difficult enough if you have the camera in your hands and can shoot well defined test targets all day long. OTOH, if all one has is a few random images and nothing to compare them to, it would be essentially impossible to determine the sensitivity curves, bokeh patterns, etc. that you suggested might be possible. FWIW, I was funded by the FBI to try to do the same sort of thing on old (analog) cassette tape recordings. It was impossible to determine a "match". About the best one could do is exclude some classes of recorders and individual units.</p>

<p>@Steve J - You have received some very good information / suggestions. As you said in your original post, and as other folks have suggested, your best option seems to be: If an entrant can't produce an original file, then they must be excluded, as harsh as this may seem. For both this contest and future ones, it is your duty to apprise the contest organizers of the difficulty of validating image files, as summarized by the posts in this thread. Hopefully, the rules will be less silly, and very clear in future contests. Good luck.</p>

<p>Tom M</p>

<p>'It's not an issue of trust the competition rules state that you must be able to supply evidence of meaning this requirement.'<br>

In this case, it's your clear duty to exclude all images that were taken with a camera that does not support forensic image authentication, or had this feature switched off at the time the photo was taken. The competition organizers will also need to purchase the camera manufacturer's expensive specialised data verification software to make this determination, e.g. for Nikon:<br>

http://imaging.nikon.com/products/imaging/lineup/software/img_auth/<br>

or for Canon:<br>

http://cpn.canon-europe.com/content/education/infobank/image_verification/canon_data_verification_system.do<br>

Unfortunately, although image authentication is quite a common feature in recent dSLRs, it's usually switched off by default, so it's rather unlikely that many (or any) of the submitted images will comply. However, this isn't your fault. The Rules are the Rules!</p>

<p>If the organizers are prepared to accept a somewhat lower standard of evidence, simply point the offending photographers towards Phil Harvey's excellent ExifTool package:<br>

http://www.sno.phy.queensu.ca/~phil/exiftool/<br>

A command like:<br>

exiftool -Model='Nikon D900' MyWinningEntry.jpg<br>

ought to be sufficient to allow an entrant to 'find' an 'original' image file with full 'evidence' of 'authenticity' (at least by the standards of the current interpretation of the rules).</p>

<p>Jeez, lighten up people</p>