17
\$\begingroup\$

A competition I want to participate in demands that they want unedited photos. The submission should not even have basic edits like brightness, saturation, white balance etc

They claim they'll check this using meta data, is it possible that they'll know if one has edited the photo?

I think some people might edit and submit and will get an unfair advantage...

\$\endgroup\$
29
  • 25
    \$\begingroup\$ Impossible. A RAW image must be edited to save it as a recognisable picture, so anyone shooting RAW must have edited in order to submit. \$\endgroup\$
    – Tetsujin
    May 17, 2021 at 16:58
  • 16
    \$\begingroup\$ Sure - but a) how would you be able to tell the difference? b) most devices that shoot to jpg have already made several computing decisions as to how to render the jpg, from simple sharpness, white balance etc right up to full HDR & out of focus blur enhancement. Where do you stop? \$\endgroup\$
    – Tetsujin
    May 17, 2021 at 17:15
  • 30
    \$\begingroup\$ I got this requirement removed from a competition I used to be a judge for every couple of years. I shot an image in raw+jpeg and edited the raw image 4 ways, then cloned the original exif data to all 5 images (script took less than 10 minutes to write) and challenged the organizers to tell me which was the original out of camera version. They could not come to a consensus, so the requirement was removed. They did maintain a rule against "compositing and hyperrealistic editing" which accomplishes the goal of judging the photography skills instead of the photoshop skills. \$\endgroup\$ May 17, 2021 at 20:07
  • 19
    \$\begingroup\$ There's really no such thing as an "unmanipulated" or "unedited" photo. I've never looked at a photograph and thought I was looking at the real world instead. The information coming off a camera's sensor has to be processed to look anything like what our eyes expect to see of a scene that we just photographed. With "straight out of camera" JPEGs, we've either allowed the makers of the particular camera we're using to make all of the decisions regarding processing, or we've modified them slightly as much as the camera allows before shooting. \$\endgroup\$
    – Michael C
    May 18, 2021 at 0:39
  • 3
    \$\begingroup\$ This is like telling you to shoot film but you're not allowed to use a darkroom for anything. \$\endgroup\$
    – J...
    May 19, 2021 at 10:36

4 Answers 4

43
\$\begingroup\$

Being slightly harsh, competition rules like that show that the organisers don't really understand how modern cameras work. A very high level and simplified view of how a camera makes a image (JPEG):

  • Light hits the sensor. Every pixel on the sensor produces an electrical reading which corresponds to the amount of light hitting it.
  • The camera converts those readings into an image. At this point, the camera makes a whole load of decisions around white balance, brightness, saturation and a number of other things. There are no "default" decisions here, the camera chooses - and on just about every camera, you have control over those decisions, whether they're "basic" decisions like white balance, or "complicated" decisions like adding a filter.
  • The camera might or might not write any of those decisions in the metadata, or they might just write "auto" for all the settings. In any case, you can delete the metadata, or replace it with other metadata - there's no way to reliably detect from metadata whether an image has been edited.
\$\endgroup\$
11
  • 22
    \$\begingroup\$ "Competition rules like that show that the organizers don't really understand how modern cameras work." This. If underexposing film when shooting and then pushing development was OK in the film days, why isn't the digital reverse equivalent (exposing to the right, then pulling exposure back in post) OK in the digital age? Or underexposing slightly to protect highlights and then pushing the mids and crushing the blacks? These are both legitimate techniques in various situations to get an end result image that actually looks more like what the eye might have seen when the scene was shot. \$\endgroup\$
    – Michael C
    May 18, 2021 at 3:20
  • 9
    \$\begingroup\$ I know you provided a high-level description but I feel the need to add: "Every pixel on the sensor produces a non-linear, not entirely deterministic electrical reading which corresponds to the amount of light hitting it which is algorithmically smoothed and corrected to produce a numerical value corresponding to a preceived brightness..." \$\endgroup\$ May 18, 2021 at 14:06
  • 1
    \$\begingroup\$ I think this answer has pedantically over-analyzed the pragmatic rules which were laid out. In other words "Submit your photos as they were taken by the camera; jpeg format assumed." \$\endgroup\$
    – MonkeyZeus
    May 19, 2021 at 12:22
  • 6
    \$\begingroup\$ @MonkeyZeus Am I allowed to change white balance in camera? Am I allowed to up the saturation in camera? Particularly for a static scene like a landscape, what's then the difference between that and editing it out of the camera? \$\endgroup\$
    – Philip Kendall
    May 19, 2021 at 12:31
  • 1
    \$\begingroup\$ It may be interesting to note that I had a project to see what we could extract from EXIF metadata. The result? Not much—many sources don’t bother to write any to begin with, a lot of processing software doesn’t bother to preserve it, and even when these things do, they’re done only kinda-sorta to any kind of standard, so good luck guessing what it means. Plus all of it is extremely prone to corruption. We abandoned that project. \$\endgroup\$
    – KRyan
    May 20, 2021 at 3:59
17
\$\begingroup\$

Submissions for these types of contests, and even some news agencies, require photos to be submitted as straight-from-camera JPEG files, not as exports of RAW images. This is usually enough to satisfy the submission requirements.

Most of the time, people trying to skirt such rules by faking EXIF data, tend to make a mistake somewhere, that is a tell-tale clue of manipulation. Often those clues are discrepancies in the various time/date stamps in the EXIF data and/or the filesystem metadata. Sometimes there are extra EXIF fields modified or added by their manipulation tool, that they weren't aware of.

To the point that even straight-out-of-camera JPEGs are manipulated by "decisions" in the camera's RAW processing algorithms, that ignores intent: a CPU or camera has no intent or goal; it is merely a complex machine following its programming. Only a person has intent, so implicit in the requirements for your contest are "no intentional edits" to be made.


But in a strict sense, there is no technical way to be absolutely sure an image wasn't manipulated. Because at some point along the way, we're out of the realm of technology, and into the realm of trust.

For smaller competitions, the stakes aren't worth it to ensure a high degree of trust. That is, the contest generally trusts (but attempts to verify) submissions were made according to the rules. But they're not going to employ expensive time-consuming techniques to try to disprove the trust. They will use simple, point-and-click -style tools to pass a "good enough" test. That's all they can do. As the stakes of the contest increase (such as substantial monetary rewards, etc.), the techniques and tools to spot alterations can be more sophisticated and more expensive.

For some news agencies, they actually extend the trust even quicker that submitters don't alter their images. This is because the photojournalists have a reputation to stake (as do the news agencies themselves), coupled with contractual submission agreements. Those agreements are backed with the possibility of termination or severance of relationship if undisclosed manipulations were made. See What are the editing restrictions for sports/photo journalism?

For other purposes where image integrity is absolutely important, knowing that technological means cannot guarantee image integrity, the trust is placed in processes and procedures ensuring the chain of custody, file handling, etc., is maintained, so that the opportunity to alter the image is reduced or eliminated. Things such as file checksumming, append-only data stores (i.e., blockchains), can certainly help, but they are not guaranteed to ensure unmanipulated data if there is no knowledge of the chain of custody of the data before those tools were employed.

\$\endgroup\$
5
  • \$\begingroup\$ "append-only data stores (i.e., blockchains)" you mean e.g. (example) not i.e. (that is). \$\endgroup\$
    – Nobody
    May 18, 2021 at 9:12
  • 1
    \$\begingroup\$ There is no reason to put "decisions" in scare-quotes when talking about in-camera JPEG rendering; every camera more expensive than a child's toy will have a number of settings which explicitly allow the photographer to affect that rendering process. Changing the white balance on a dial on the camera is exactly equivalent to changing it in a post-processing program like Lightroom. \$\endgroup\$
    – IMSoP
    May 18, 2021 at 9:29
  • \$\begingroup\$ @IMSoP Disagreed, because the camera is not making "decisions", it's a human making "decisions" to set the dial. The camera has no intention or will of its own. It is merely following a script, a set of rules. It is making no artistic judgement or preferences. Those are human choices, intentions. \$\endgroup\$
    – scottbb
    May 18, 2021 at 13:55
  • 4
    \$\begingroup\$ @scottbb No, you missed my point, I'm saying that even straight from the camera the human has made decisions about how to "develop" the raw sensor data into an image, so asking for an image with "no intentional edits" is meaningless. An image with white balance tweaked in Lightroom is significantly less "manipulated" than one where the photographer selected "simulate tilt shift" in the camera's menu, but both images have been manipulated by a computer based on the artistic decisions of a human. \$\endgroup\$
    – IMSoP
    May 18, 2021 at 14:22
  • 3
    \$\begingroup\$ @IMSoP For the most part I don't disagree with you then. But there is a meaningful distinction between intentionally editing after seeing what was taken, vs. deciding to accept what is produced before pressing the shutter button. Yes, the "I'm taking what I get regardless" is pretty well deliberate and informed by settings beforehand. But there's still a difference between the two, and that difference is what the spirit of such contests are about. \$\endgroup\$
    – scottbb
    May 18, 2021 at 14:32
4
\$\begingroup\$

An additional thing to take into account is the code that creates the jpeg has an identifiable signature. The way Photoshop writes the jpeg will be different than the way Corel will write it. Exiftool, for example, has a JPEGDigest tag, which is defined as:

an MD5 digest of the JPEG quantization tables is combined with the component sub-sampling values to generate the value of this tag. The result is compared to known values in an attempt to deduce the originating software based only on the JPEG image data.

JpegSnoop goes into more details on it's Identifying Edited Photos.

\$\endgroup\$
2
\$\begingroup\$

By looking at the exif data, it's possible to know if the image has been developed in lightroom, or comes straight form the camera as a jpeg. This can also be faked.

Now, if you take a picture, then look at the result on the camera LCD, you can check the exposure, histogram, white balance, etc. They you can adjust settings and take another picture. Technically, it is not edited, but the result would be the same as if it was. You could also use bracketing.

I think this rule somewhat gives an advantage to photos of subjects that won't go away, like landscape or portraits, since these subjects allow several attempts. If the intent of the rule was to make the photos more spontaneous and candid, it may achieve the opposite.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

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