48

My friend recently linked me a GIF of the entire Shrek movie in 60x60 resolution. At the time of writing this, it has 12 million views on imgur. This got me thinking, could a copyright holder technically sue over the uploading of this GIF? I'm guessing yes, because it's still barely legible. But at what point are they not allowed to do so? Surely they can't claim intellectual property over one pixel shifting to specific colors for 90 minutes, for example? How about 2x2? Etc.

I read What considerations determine copyright infringement? which to me makes it seem like they COULD sue over the one pixel shifting colors for 90 minutes, because I did copy and modify the movie. Is that really true?

I also read How similar should a tune be to constitute theft of someone else's tune? which states that it depends a lot on the case since there are so many variables. Is that also true in this case? I imagine songs work very different from movies.

Looking for both US or EU answers.

MarcusOtter
  • 553
  • 3
  • 9
  • 2
    I assume being a GIF means it doesn't feature the soundtrack, which is protected separately. – Stuart F Aug 29 '22 at 09:45
  • 11
    The fact that it is being referred to as "the entire Shrek movie" might have some bearing on this - that clearly declares that it is intended to be a copy of the entire movie. – Zibbobz Aug 29 '22 at 12:47
  • 9
    I imagine this is like pornography: I may not know how to define it, but I know it when I see it. Largely boils down to understanding intent of both defendant and plaintiff. You can usually tell if someone is trying to circumvent the spirit of the law with loopholes and technicalities. – DKNguyen Aug 29 '22 at 13:51
  • 2
    Odds are quite high that your web browser is playing an MP4, not GIF; does this change the essence of your question? – MonkeyZeus Aug 29 '22 at 17:13
  • 1
    @MonkeyZeus you can download it for yourself from imgur and see that it's a GIF (but maybe that's committing a crime, heh). But no, it would not change the essence of my question, as the question is about resolution, not necessarily the video format used. But it is audio-less, which is being correctly taken into account by the answers so far. – MarcusOtter Aug 29 '22 at 17:27
  • @MarcusOtter Technically, your browser downloads everything which a website presents to it; if it didn't then you couldn't view content such as the logo on this website. Albeit modern cache doesn't usually let you identify individual files easily so having the file in your downloads folder could be a "smoking gun" which reveals intent. It would be trivial for a bad acting website to force your browser to cache illicit/illegal images. If that bad acting website had your location information then it could also report you to the authorities and give you a real bad time. – MonkeyZeus Aug 29 '22 at 17:34
  • MarcusOtter and @MonkeyZeus: When I right-click on the animated rectangle, Chromium pops up a context menu that includes "save video as", and it's a .mp4. That's the context menu for the in-browser video player that's actually playing something to produce the pixels on the page. "Open video in new tab" also opens a .mp4. The download link presumably lets you get the original .gif that someone uploaded, which imgur transcode to an MP4 so it could play in the browser with seek controls. (And so it could stream; not sure if browsers do well with gigantic .gif files.) – Peter Cordes Aug 29 '22 at 18:47
  • The GIF download is 40 MiB (and not easily seekable, e.g. with mpv which can play it locally). The MP4 that imgur shows you is only 1.5 MiB. And clearly transcoded from the GIF. Not that any of this particularly matters to the legal question. – Peter Cordes Aug 29 '22 at 18:55
  • 1
    "But at what point are they not allowed to do so?" -> https://en.wikipedia.org/wiki/Sorites_paradox – Boann Aug 31 '22 at 04:44
  • Don't forget that the audio is copyrighted material too. You could reduce the video to zero pixes and still be in violation. And in any case, your compressed version is clearly a Derivative Work. – keshlam Sep 01 '22 at 05:21

5 Answers5

44

A few years ago, there was a trial in the USA about some short sound on some music CD:

One party claimed that one piece of music on the CD contained a sound of less than one second length which is copied from another CD without the permission of the copyright owner of the other CD.

It could never be found out if this claim was really true.

The court's decision was:

If it is not possible to distinguish between a copy and a work that does not depend on the other work at all, it is not a copyright infringement - even if the sound has been copied from the other CD.

For this reason, I'd guess that a 4x3 image would not be a copyright infringement, yet, while 60x45 would definitely be one.

Just for reference: The same image as 3x4 and as 45x60:

"Mona Lisa" in different resolutions

Martin Rosenau
  • 852
  • 6
  • 8
  • 30
    One animated pixel shifting colors at, say, 10 fps for 90 minutes would contain data equivalent to (for example) a 180x300 pixel static image, though. It would certainly be possible to tell with near certainty whether it was produced by downscaling a specific video file, simply by repeating the process and observing whether or not the resulting sequence of colors was the same. (On the other hand, such use would nonetheless be highly transformative and almost certainly fair use.) – Ilmari Karonen Aug 29 '22 at 08:29
  • 3
    @IlmariKaronen Counterargument 1) When being sued on court, the defendant would be able to create a completely different movie resulting in exactly the same 4x3 pixel "movie" when being downscaled. Counterargument 2) I'm not sure about US copyright, but according to German copyright some "work" must have been invented by a person. The 4x3 "movie" lacks all information that has been invented by the originator of the movie. – Martin Rosenau Aug 29 '22 at 13:55
  • 3
    @IlmariKaronen (Parenthetical Counterargument :P Being transformative does not—by itself—make something Fair Use, at least in the US. And in this particular case, I’m not at all sure it would be Fair Use, since the copyright holder would have exclusive right to produce derivative works and the rotation of the image in spacetime would definitely be derivative, however much it is transformed from the original.) – KRyan Aug 29 '22 at 13:59
  • 7
    'When being sued on court, the defendant would be able to create a completely different movie resulting in exactly the same 4x3 pixel "movie" when being downscaled.' – Yes, and that "completely different movie" would also be a derivative work of Shrek. It's statistically impossible for a movie to have the same 4x3-pixel downscale as Shrek by sheer coincidence, so if a movie does have the same 4x3-pixel downscale as Shrek, then there are only two possibilities: either the movie is derivative of Shrek, or both the movie and Shrek are derivative of some third work. – Tanner Swett Aug 29 '22 at 16:41
  • What about something like https://i.stack.imgur.com/OZ84c.jpg? – MonkeyZeus Aug 29 '22 at 17:42
  • 3
    @TannerSwett The question would be: Can you prove that the completely different movie was made using the downscaled Shrek movie and that it is not pure incident that the downscaled 4x3 "movie" has similar colors? As far as I understood the court decision correctly, it is not a copyright infringement if no human is able to see a link between both movies! – Martin Rosenau Aug 29 '22 at 18:18
  • 1
    @MartinRosenau It'd be fairly trivial (and convincing) to argue that the probability of two movies having the exact same downsampled pixel values over their runtime is so astronomically small that the movie must, beyond any reasonable doubt, be derived from the Shrek movie. – Birjolaxew Aug 29 '22 at 22:09
  • @Birjolaxew It's not that simple: There are a lot of downscaling algorithms and using different algorithms, you will get different results. Not knowing the algorithm used, it would even be difficult to prove that the 4x3 "movie" is a downscaled version of the original movie at all! – Martin Rosenau Aug 30 '22 at 09:39
  • 1
    @MartinRosenau: I see where you're coming from, but in practice it would be simple to prove, in most if not all practical cases: for any reasonable downscaling algorithm, the average brightness of the downscaled version will—to a great extent—track the average brightness of the original; for any individual frame that is of no use, but the graph of "average frame brightness over time" will give a characteristic fingerprint of the movie that's preserved in the downscaled version. The chances of a close, consistent match from anything other than deriving from the original movie are infinitesimal. – psmears Aug 30 '22 at 13:17
  • @TannerSwett: There is a third possibility: that Shrek is a derivative of my movie that has the same 4x3px downscale. This one may be tricky to prove in court, though. – psmears Aug 30 '22 at 13:19
  • @MartinRosenau can you add a link to the court case in question (or a summary/wiki article/etc) ? – mbrig Aug 30 '22 at 18:20
  • 1
    @mbrig Unfortunately, I only read a German language article about the trial - so the information may be wrong. But the court verdict can be downloaded here. – Martin Rosenau Aug 30 '22 at 18:45
  • @Birjolaxew how is that not equivalent to saying "that puddle is exactly the same size and shape as my melted swan-shaped ice cube, therefor I can conclude that the puddle was produced by a swan-shaped ice cube"? An infinite number of ice cubes of different shapes could have produced that puddle. At some point the information loss is too great, which is what the question and this answer are about. – Jared Smith Aug 31 '22 at 12:00
  • @JaredSmith Because the burden of proof in copyright cases is often "beyond a reasonable doubt". Application of law does not depend on mathematical proofs. There might be an infinite number of possible movies that would produce the result, but if the probability of any one of them being produced organically is so small as to be realistically impossible, then it isn't reasonable to doubt the copyright infringement based on their theoretical existance. – Birjolaxew Aug 31 '22 at 15:16
  • 1
    @JaredSmith A puddle produced by a melting piece of ice generally doesn't contain a significant amount of information about the shape of the original piece of ice. On the other hand, scaling a movie down to 4x3 pixels does preserve a substantial amount of information. If we pessimistically assume that each pixel only preserves 1 bit of information, and that only one frame per second is significant, then shrinking down a 90-minute movie still preserves 12 * 90 * 60 = 64,800 bits of information. – Tanner Swett Aug 31 '22 at 15:40
  • 1
    64,800 bits of information is way more than abundantly plenty to prove that copying occurred, beyond anything vaguely resembling a shadow of a shred of an iota of doubt. Therefore, this answer's claim that "it is not possible to distinguish between a [4x3 pixel] copy and a work that does not depend on the other work at all" is, unfortunately, utterly wrong. – Tanner Swett Aug 31 '22 at 15:48
  • @TannerSwett If the number of bits was the main criterion, all works created after 1934 would be "derived works" of the Champernowne constant... – Martin Rosenau Aug 31 '22 at 17:15
  • @MartinRosenau That's correct, and that demonstrates that merely counting matching bits isn't sufficient to demonstrate that copying occurred. – Tanner Swett Aug 31 '22 at 17:40
26

Even squeezed down to a single pixel, the animation would still technically be a derivative of the original movie. However, using the movie in such a radically transformative manner would almost certainly be considered fair use (or equivalent in other jurisdictions) and thus not actually infringing on the movie's copyright.

In particular, in the , 17 U.S. Code § 107 defines fair use and sets out the four main criteria to be considered when determining whether a use of a copyrighted work is fair:

In determining whether the use made of a work in any particular case is a fair use the factors to be considered shall include—

  1. the purpose and character of the use, including whether such use is of a commercial nature or is for nonprofit educational purposes;
  2. the nature of the copyrighted work;
  3. the amount and substantiality of the portion used in relation to the copyrighted work as a whole; and
  4. the effect of the use upon the potential market for or value of the copyrighted work.

Let us consider the hypothetical 1x1 pixel animated GIF version of the Shrek movie — or the less hypothetical movie barcodes mentioned in the comments — with respect to these criteria (and the additional guidance provided by the U.S. Copyright Office regarding them):

  1. Purpose and character of use: Squeezing a movie down to a single blinking pixel is highly transformative and, as such, more likely to be considered fair:

    Additionally, “transformative” uses are more likely to be considered fair. Transformative uses are those that add something new, with a further purpose or different character, and do not substitute for the original use of the work.

    While in a purely mechanical sense downscaling the video to a single pixel (and removing the audio entirely) adds no new information, the idea of transforming a movie in such a way could certainly be considered something new and creative, and the resulting original work of visual art is certainly substantially different in purpose and character from the movie and cannot possibly substitute for it. Indeed, it is unlikely that a typical person viewing the single animated pixel without additional explanation could even recognize it as being based on a movie, or what movie it was based on.

  2. Nature of the copyrighted work: This factor does not particularly weigh in favor of fair use, as the copyrighted work in this case is a creative work of fiction and visual art rather than, say, a technical manual or a news report.

  3. Amount and substantiality of the portion used: While the entire duration of the movie is used, each frame of the movie is transformed and reduced to a vanishingly small fraction of its size and image content (less than one millionth, assuming the original was in full HD resolution) in a manner that does not permit reconstruction of any recognizable portion of the original movie. Furthermore, the audio part of the movie is omitted entirely.

  4. Effect upon potential market or value: As already noted in the analysis for the first factor above, a blinking pixel cannot possibly be a market substitute for an animated feature film. Thus, this factor weighs heavily in favor of fair use.

With three out of the four points weighing strongly in favor of fair use, it seems likely that at least the 1x1 pixel version of the movie would be found sufficiently transformative and distinct from the original to be considered fair use. And the same would likely be true of, say, a 2x3 pixel version too, or indeed any version with so few pixels that no recognizable image can be seen in them.


What about the 60x60 pixel version mentioned in your question, then? Here, we're getting into somewhat murkier territory.

For one thing, now the original movie is clearly recognizable from the downscaled version. Indeed, it could be argued that the downscaled animation — even with just 60x60 pixels and no sound — remains marginally viewable as a movie, and thus could in some sense substitute for the original movie. Whether anyone would actually want to watch a 90 minute 60x60 pixel GIF animation, even for free, instead of proper movie is certainly doubtful. But one could conceivably come up with some imaginary situation where someone might view the two as potential substitutes.

Furthermore, the very fact that the movie is recognizable from the animation somewhat makes the transformativity argument slightly weaker, or at least not quite as obvious: a single blinking pixel is obviously not a movie, but a low-resolution silent movie still is a movie, even if not a very good one. And, whereas one could easily argue that transforming a movie into a single blinking light is an act of artistic expression that creates a completely new viewing experience entirely unlike watching the movie in a normal way, it seems a bit harder to make that same claim about a "thumbnail" version where the characters and events of the original movie are still perceptible, however crudely.

That said, if the matter were to come before a court, it certainly seems possible that a competent lawyer could successfully argue in favor of even the 60x60 pixel version being fair use. But it also seems possible to me that such an argument might not succeed. And, in practice, a lot could depend on the specific context, manner and purpose for which the downscaled version was exhibited and/or distributed.

(For example, as noted in the comments below, thumbnailing images for preview, in the context of an otherwise legitimate index providing legal access to the original images via hyperlinks, has been held to be fair use in several cases, such as Kelly v. Arriba Soft and Perfect 10 v. Google. Yet it's also obvious that merely scaling down the resolution of an image or a video somewhat does not automatically make an otherwise infringing use fair, e.g. if the intent is to provide a free pirate substitute for the commercially sold original work.)


And, of course, for other jurisdictions all bets are off again. While, in general, sufficient transformation of the original work — at least, as noted in Martin Rosenau's answer, to the point where it's impossible to determine whether copying actually occurred or not — should be a defense against copyright infringement everywhere, the specific manner in which this is codified into law and precedent varies a lot between jurisdictions, as consequently does the threshold of what is considered transformative enough.

Ilmari Karonen
  • 1,222
  • 8
  • 11
  • 1
    Why would "the fact that the movie is recognizable from the animation" undermine the transformativity? An example of transformative use listed in https://en.wikipedia.org/wiki/Transformative_use is Google's use of thumbnails (which are clearly recognizable as a downscaled version of an image). The GIF seems to be transformative to me because the purpose is different. I don't really know what the intended purpose is of the 60x60 GIF version, but it doesn't seem to be the same purpose as to watch the Film in a theatre, on a DVD, on Netflix, etc. – Brandin Aug 29 '22 at 10:21
  • @Brandin: Perhaps that was poor phrasing on my part. I'll try to think of a better way to say what I meant. Certainly a derived work can be transformative even if it closely resembles the original in some ways. But a derived work that doesn't at all resemble the original in any recognizable way is obviously highly transformative in a way that a more recognizable derivative isn't, at least not as obviously. – Ilmari Karonen Aug 29 '22 at 11:56
  • @Brandin I think the reasoning there is the “amount and substantiality” prong. If you can see everything that happened, you’re copied much more of the substance of the work than if you’ve only shown the color palette of each frame. – Davislor Aug 30 '22 at 00:20
  • @Davislor Courts have found that copying an entire image at low resolution counts as a lesser amount. I'm not sure in what proportion, but this 60x60 GIF is so ridiculously low res, it has reduced colours, it has no sound, etc. If I had to bet, I think the only purpose of this GIF is to attract clicks, views and comments. It does not replace a licensed version of the Shrek film and won't negatively impact the current or future market. For example, I really doubt if the copyright holders could produce their own 60x60 GIFs such as these and then charge money for viewing those on the market. – Brandin Aug 30 '22 at 06:49
  • What about that ASCII Star Wars movie? – htmlcoderexe Aug 30 '22 at 10:00
  • I remember “Eye of the Tiger” played on an old needle printer :-) – gnasher729 Aug 30 '22 at 11:26
6

At no point is it not copyright infringement. What is protected is the "expression", and not the physical characteristics of that expression. Thus a Harry Potter novel produced on a typewriter by an author is protected, and remains protected if someone turns it into a PNG file, and likewise an OCR's version created from that, which that is turned into a Word 93 document, and then later an ODT file (we can then print it, re-scan and so on).

Clearly, somebody copied. Copyright means "the right to make copies", and clearly making unauthorized copies is an infringement on that right. Copyright does not mean "the right to make exactly identical copies using the same technology.

user6726
  • 214,947
  • 11
  • 343
  • 576
  • 25
    …but if someone were to shred the Harry Potter novel, turn it into papier-mâché, make a sculpture out of it and then distribute photos of the sculpture, that would pretty clearly not be copyright infringement. Or at least there would be a very strong fair use defense based on the transformative nature of the use. And there is at least an argument to be made that e.g. squeezing a movie down into a single blinking pixel would be closer to this than to your example. – Ilmari Karonen Aug 29 '22 at 07:42
  • 6
    @IlmariKaronen You mean something like this? – Arthur Aug 29 '22 at 08:02
  • 4
    I think the case hinges on the definition of "copy". Obviously, the lawmakers target the typical case: Copying a book makes typically only sense if the copy is legible etc. That is, the typical case presumes that the essence of the work is preserved. Whether "copies" that destroy this protected essence (the text, the image, the film plot and artistic expression etc.) are still violations is exactly the question here. Exactness or identical technology is of course not required; but one may argue that the preservation of the essential contents is. – Peter - Reinstate Monica Aug 29 '22 at 10:32
  • 8
    I don't find this example very pertinent, as it's about creating clearly identifiable copies that differ only in their technical aspects while still representing identical expressive aspects. A 1x1 downscaling of a movie simply does not have the same expressive aspects as the original. A closer example might be writing a text that's every 100th word from Harry Potter - likely a less clear-cut case than just re-typing the book in its entirety on a typewriter. It might infringe if it's still recognizable as a story about a boy wizard, but likely not if it's just a jumble of prepositions. – Nuclear Hoagie Aug 29 '22 at 13:58
  • @NuclearHoagie: It quickly becomes clear copyright infringement once your friend does the same thing with every 100th word starting from the second, and then some random other person does the same thing starting with every 100th word from the third, and then... – R.. GitHub STOP HELPING ICE Aug 30 '22 at 19:16
  • 1
    @R..GitHubSTOPHELPINGICE I don't necessarily agree with that, unless you and your friends are working together to actually copy the whole book. Ten thousand independent book reviews could, in aggregate, quote HP in its entirely without any copyright infringement, so long as they each individually quote a limited enough passage to be considered "fair use". – Nuclear Hoagie Aug 31 '22 at 16:37
  • @NuclearHoagie: No, I don't think they could - there's plenty of "filler" material that could not plausibly end up quoted in a review for a legitimate purpose. But if it did, there is that legitimate purpose - being a review - that makes it arguably fair use. If you just copy a mechanical subset according to a pattern, such that it's in a readily usable form for reconstructing the original mechanically, it's almost certainly infringing. – R.. GitHub STOP HELPING ICE Sep 01 '22 at 02:32
3

This got me thinking, could a copyright holder technically sue over the uploading of this GIF?

Technically, a copyright holder could sue over uploading a GIF that is completely unrelated to their copyrighted work :).

So I assume you are actually talking about at what point the lawsuit might actually have some level of merit. Maybe asking if it could survive a summary judgement motion? Or simply avoid sanctions for being frivolous?

I'm afraid there is no clear answer. Copyright protects expression. Actual copying or transformation of a protected work is nearly universally recognized as a copy, and therefore would be infringing if unauthorized. In the U.S. there are defenses for fair use. Software companies sometimes use a process of 'black box' re-implementation of software where one team has access to software and creates a specification that another team can follow to create software that does the same thing, and can show that the second team has no communication with the first team or access to the original software so they aren't 'copying', although the resulting software does the same thing.

In your '1-pixel' example, that ends up being about 130,000 pixels over the course of a 90 minute movie, enough for a 360x360 image. The problem is though that you are admitting that you are copying. The only thing that makes your GIF special is that fact that it has another work as the source. While there is some ingenuity in the idea of taking a single pixel from each frame of a movie, the actual expression of your idea is in fact simply a copy. The same method could be done with any of thousands of movies and the only differentiation would be the source material. That's something like coming up with a new medium for art such as oil painting, but the method of creation is not copyrightable, only the expression used in creation.

You could argue the defense of fair use, arguing that the use was de minimis, but the fact that you had no original creative expression would probably preclude that.

Jason Goemaat
  • 711
  • 4
  • 6
  • I think the point of characterizing a use as "de minimis" in this context is that it renders the other considerations, like the amount of creativity in the "derived work" irrelevant. Further, "try something arbitrary and see if one finds the result aesthetically pleasing, and if not keep trying other arbitrary things" can be a creative process even if one lucks out and the first try yields something good. – supercat Aug 31 '22 at 14:45
0

As others have stated above, there is no point that creating a single pixel-sized animated gif is not a derivative work. So, prima facie it is a copyright infringement.

In UK law the test of infringement is whether a substantial portion of the original work is contained in the alleged derivative work. This is not a quantitive test per se, it is more of a qualitative test. For example, it is arguable that an infringement can occur where the “idea” has been taken, rather than some identifiable part of a work (see “Temple Island Collections Ltd v New English Teas Ltd & another [2012] EWPCC 1”, discussion here: https://www.copyrightuser.org/educate/the-game-is-on/episode-1-case-file-1/).

Arguably this crosses the line in the idea-expression dichotomy, but it has not been overturned.

The USA doctrine of “fair use” does not exist in the UK, its somewhat equivalent is “fair dealing”, which allows some copying for certain purposes. Principally, this amounts to review, where you would be able to show the Red Bus photo(s), in order to show/describe the differences between them, or perhaps a short excerpt from a film by way of a review.

The proposed use does not appear to fall into the fair dealing defence. However, you may be able to develop a novel argument that indicates a single pixel gif of “Shrek” is a form of review (“Look how powerful this expression is, it still survives down to a single pixel”).

I am not indicating that such an argument would be successful.

There is no doctrine of transformation in UK law. In fact, the opposite is the case, in the sense that there is specific legislation that says, for example, that creating a 2D version of a 3D work, or vice versa, is an infringement.

I believe that the case is that same across the EU, which is moving in the direction that the protected element of copyright is the product of the author’s mind. This tends to imply that copying, with whatever transformation you like, will always amount to an infringement.

The outsider’s view, looking at the USA’s doctrine of transformation (and fair use, for that matter), is that it tends to eviscerate copyright.

Whether an author/rights holder would ever take legal action over a 1*1 pixel gif is another matter.

typonaut
  • 161
  • 1