You may have read recently that some of the major studios, most prominently Disney, are alleged to have infringed the patent rights in a certain motion-capture system used to make blockbuster films like the multi-billion-dollar Marvel movies. Further, an article like this one in Hollywood Reporter by Eriq Gardner might give a reader the impression that a) the patent infringement allegations are indisputable; and b) that the plaintiff in this case Rearden LLC is claiming to “own” some of the famous CG characters featured in these movies. So what gives?
A lot. But not necessarily what the headlines imply. As for the patent infringement allegations, these are impossible to comment upon at the moment because 1) I know almost nothing about patent law; and 2) the entire backstory as to how Disney and other studios may be implicated is too complex to unravel in this forum. Suffice to say, we’ve got some former technology partners who are no longer friends, some dubious-sounding shenanigans involving the sale (or not) of shared (or not) intellectual property, and an inventor named Steve Perlman (Rearden’s CEO) who seems to have at least a few good reasons to be pissed off at somebody.
Time and the courts will sort all that out, but for the sake of addressing the copyright allegations in this story, we need to assume hypothetically that the studios are implicated in the patent infringement from which the copyright claims stem. And let’s cut to the chase and clarify that Rearden cannot—and is not—claiming any ownership of the underlying pictorial or graphic works we know as Hulk, Deadpool, Iron Man, etc. What Rearden does allege is that by making unlicensed use of its MOVA Contour technology, the studios simultaneously infringe Rearden’s copyright in the outputs of this technology. Hence, cutting through the fog and high drama, the entire copyright claim appears to turn on whether Rearden’s outputs are in fact copyrightable at all.
MOVA Contour’s Outputs
As an example, Mark Ruffalo goes into a motion-capture studio where he grins, shouts, pouts, growls, and raises curiously adorable eyebrows in the mode of his inner Hulk. He may be directed by the film’s director or some other member of the creative team, or he may self-direct to produce all the faces needed for every scene in which Hulk will appear. These performances, fixed in the computer drives during the process, are indisputably the property of Marvel Studios, which is owned by Disney.
The captured data, consisting primarily of multiple two-dimensional images is then input into the MOVA Contour system, where the software renders a pair of three-dimensional outputs: what Rearden calls a “Captured Surface” and a “Tracking Mesh.” These two assets then serve as a highly-detailed, digital armature onto which Marvel-employed animators then composite the CG Hulk, who will then grin, shout, pout, growl, and raise curiously adorable eyebrows in the movie, looking just enough like Mark Ruffalo to make the character transformation seem organic.
Rearden asserts that, although the Captured Surface and Tracking Mesh are never displayed in the final movies, these two outputs are, nevertheless copyrighted works and, therefore, every CG-composited element built upon these outputs is a “derivative work.” So, absent a licensing agreement to use the software and transfer ownership of the outputs to the licensee, the studios are alleged to have infringed Section 106(2) of the copyright act, which protects the derivative works right. Indeed, if this claim has merit, it would be quite staggering to consider the number of infringements implicated. One can almost hear the Hollywood-bashing copyright-haters chomping at the bit; but if that’s the case, they might want to belay the schadenfreude.
I imagine the studios’ motion to dismiss will be denied because the allegations seem to warrant further proceeding. But even if the copyright claim goes far enough to render an opinion by the court, I believe there is ample evidence in both scholarship and case law to conclude that Rearden’s outputs are not copyrightable and, therefore, cannot implicate the derivative works right.
Is MOVA Like Other Software Used to Make Creative Works?
At first blush, one might assume that the MOVA software is akin to using Word or Photoshop, which are used by authors all the time without any fear that Microsoft or Adobe can claim a copyright in the works made with these products—even if the author pirates the software! But, Rearden counters this premise by saying that because their software produces a specific, repeatable output fixed in a medium, and because this output is produced almost entirely by the “labor” of the software, the analogy to Word isn’t quite right.
The difference they claim relies heavily on the case TorahSoft v. Drosnin (2001) in which defendant Michael Drosnin published a book based on outputs (effectively search results) that were exclusively the products of plaintiff ToraSoft’s algorithm. Rearden appears to be clinging to language in this opinion that supports the possibility that the output from computer software could be copyrightable; but the lion’s share of the opinion actually provides reasons why TorahSoft’s output was not copyrightable. And because much of the court’s reasoning in that case seems applicable to Rearden’s claim, it may be a risky citation.
In response to the assertion that MOVA is comparable to Word or Photoshop, Rearden states,“…in neither case does their work provide input to software that synthesizes an original expression that is distinct from the author’s or artist’s input.” I wonder if that is, or should be, the standard. After all, between my writing these words and you’re reading them, there are several software authors whose work renders my expression into intermediary “works” that, if they were indeed copyrightable, would mean that everything created with digital tools is a derivative work of underlying works belonging to the tech companies of the world. I don’t think the courts would view this outcome as compatible with the purpose of copyright.
Are MOVA’s Outputs Expressions or Copies?
Curiously enough, the “sub-millimeter precision” with which MOVA Capture does its job may be the evidence which most disfavors a finding that its outputs are copyrightable. The standards for protection are “originality” and a “modicum of creativity.” Originality literally means that the work is not a copy, and so far, “creativity” has been limited to works produced by humans. We may yet address copyright protection for works that are substantially produced by AIs, etc., but in this case, it seems Rearden is arguing that the human expression embodied in the software is tantamount to authorship of the software’s output, which it claims is separately “creative.”
While this may be possible in some circumstance, as TorahSoft seems to imply, I suspect Rearden’s argument will find slippery ground because the courts will view the purpose of MOVA software as one of making copies of performances that are the property of the defendants. In this context, I believe Rearden’s software, especially as a component of the entire system, may be compared to patented devices that capture images which would strain to meet the standards for protection.
Copyright scholar Justin Hughes, whose paper on photography I have cited in other posts, asserts that a vast quantity of images—security camera footage, satellite photos, Google Street captures, even photos of public-domain paintings sold as postcards—should not stand up to claims of copyright because they lack the modicum of human creativity required for protection. Assuming Hughes is correct, Rearden’s claim of copyright in its outputs would seem to fail under the same analysis.
More specifically, the case that comes to mind in answer to Rearden’s claim is Meshwerks v Toyota, decided in 2008 by the Tenth Circuit Court of Appeals. In that situation, Meshwerks asserted copyright protection in the outputs from its capture technology used to produce 2D, digital models of Toyota cars for advertising purposes. Supplier Meshwerks actually performed a considerable amount of creative labor—in collaboration with software—to produce the outputs. This stands in contrast to Rearden, whose claim relies solely on the products of the software. Nevertheless, and with stated respect for the amount of work involved, the court held that Meshwerks outputs were not copyrightable because their purpose was to create the best copies possible of designs belonging to Toyota. From the opinion, which partly quotes Nimmer on Copyright:
It is certainly true that what Meshwerks accomplished was a peculiar kind of copying. It did not seek to recreate Toyota vehicles outright-steel, rubber, and all; instead, it sought to depict Toyota’s three-dimensional physical objects in a two-dimensional digital medium. But we hold, as many before us have already suggested, that, standing alone, “[t]he fact that a work in one medium has been copied from a work in another medium does not render it any the less a ‘copy.’ ”
So, if we compare this to Rearden’s claim and the Hulk example, Disney certainly did not design Mark Ruffalo, but they do employ him to perform the only creative work in this entire process—that of making faces as Hulk. And I would argue that, akin to Meshwerks, the function of MOVA software is to copy that performance (in this case, from 2D to 3D). In fact, it is the triumph of the system’s capacity to deliver such precise copies which disfavors a conclusion that its outputs are “original” expressions of “creativity.” The process is no more creative than a CAT scan—far less creative, it seems, than the work performed by Meshwerks, which was held not to meet the standard of protection under copyright.
It’ll be interesting to see if any of the usual anti-copyright suspects come out to play on this one. Presumably, they should disfavor Rearden’s claim of copyright for many of the reasons cited here, but siding with Hollywood studios may be more than they can bear.
Image by kentoh.