“These are not natural events”
Ariel’s Technodramaturgy in the Royal Shakespeare Company’s The Tempest
A Signpost to the Future
In one of the weirder annals of theatre history, the playwright and notorious Shakespeare iconoclast Sir George Bernard Shaw was solicited in 1949 by the Malvern Marionette Theatre to write a ten-minute comedy for two of the company’s most eminent puppets—Shakespeare and Shaw himself. The result is the profoundly silly Shakes Versus Shav, wherein the puppet Bard and Bardolator engage in a battle of wits (and fists) over their respective theatrical legacies. Rest assured—it’s not his most incisive work.
Nevertheless, in his introduction to this farcical agon, Shaw makes a curiously prescient remark. Bemoaning the dying medium of puppetry at the hands of “cinematic effects,” Shaw imagines a future of “puppets simulating live performers so perfectly that the spectators will be completely illuded.” This uncanny spectacle, Shaw argues, would result in “the death of puppetry for it would lose its charm with its magic.”
That future, it seems, is here. Digital puppetry, or motion capture (hereafter mo-cap) is ever approaching photorealism. It’s long since become a staple of the film and video game industry, and has filtered all the way down into smartphone applications, as recently unveiled by Apple with their “animojis.” Yet defying Shaw’s augury, this golden age of digital puppetry has conversely allowed the medium to return to its theatrical roots. Earlier this year, the latest in live mo-cap was spectacularly displayed in Gregory Doran's production of The Tempest at the RSC, in which the character of Ariel (played with spritely magnetism by Mark Quarterly) was accompanied in real time by his computer-generated double.
Now, as someone who works in CGI and projection design, let me assure you that this production is a big deal technically and creates a very promising precedent for live mo-cap (for reasons I'll explain in geekier detail). Yet for the purposes of this article, I'm more interested in exploring the production’s technodramaturgy—a phrase I’ve admittedly coined, but which simply refers to the dramaturgies that arise out of the use of a given technology.
In doing so, I’m partly on a mission to defend the use of mo-cap in classical work. For while Doran’s production was widely lauded, commentators generally perceived an incompatibility between the technology and the play itself. Across the board, critics agreed that the use of mo-cap felt more like “a one-off experiment rather than a signpost to the future,” and that Ariel’s CG avatar, while undeniably breathtaking, was “ultimately gimmicky” and had a tendency to “upstage the actors.”
I imagine Shaw would have been similarly unimpressed; reviewing a production in 1837, he remarked that “The Tempest is so magical that it would make the scenery of the modern theatre ridiculous.” Like the detractors of the 2017 production, Shaw suggests a tactlessness, even impudence, in using modern stage machinery to realize Shakespeare’s fantastical play. My goal in this essay is therefore to explore the dramaturgical impetus for the production’s use of mo-cap, and how the construction of an embodied and artificial Ariel may account for some of the critical skepticism around the use of this digital avatar.
A Quaint Device
Unlike the bulk of the Shakespearean canon, The Tempest actually has a fairly straightforward textual history, being first published in the Folio of 1623. Resultantly, our “authoritative” version of The Tempest is accompanied by remarkably coherent and detailed stage directions. Despite debate over whether these were written by the Bard himself—J. Dover Wilson wrote that they “possess a beauty and elaborateness…[that] bear the unmistakable impression of the master’s hand”—the contemporary consensus seems to be they were added by the editors of the volume, John Heminges and Henry Condell, for increased legibility.
Regardless of their authorship, the stage directions of Tempest can provide a useful framework when thinking about how one might stage this play, necessitating as it does a number of spectacular stage effects. In an essay from 1938, the Shakespearean scholar John C. Adams makes a particularly bold claim about a stage direction from Act 3, Scene 3, in which Ariel appears as the Harpy, and conjures a richly laid banquet with his ministers, before “with a quient device the Banquet vanishes.” While this effect would normally suggest the use of a conventional Elizabethan trapdoor, Adams argues that the phrase “quaint device” here suggests “a new and ingenious property…unlike anything normally used in the playhouse.” Taking the inference a step further, Adams remarks:
Have our discoveries any bearing on Shakespeare’s choice of harpy costume for Ariel? …When Ariel “clasps his wings vpon the Table” their outstretched expanse would wholly conceal the banquet.
While this deduction may seem a little presumptuous, Adams touches on an interesting idea —that the technical capacities of Elizabethan stage machinery may have impacted some of Shakespeare’s dramaturgical considerations. After all, the relationship between theatrical meaning and stage machinery is rarely a one-way street; more often than not, the two feed each other in a kind of iterative loop—or as Finn Ross, the video designer for the 2017 production described it, as the result of an “opinion matrix.”
When I spoke to Ross, he described how the production team aspired to a “nuanced” rather than a “slavish” use of their own quaint device. When the RSC’s partners, Intel and Imaginarium Studios, first became involved in the process, the team settled on a basic rule for the use of mo-cap. “When Ariel was there as himself as a character, sort of speaking directly with Prospero, then he would just be himself, as Mark,” Ross told me, “and then whenever Ariel was conjuring something…when he was a magical manifestation…then he would appear as a sort of avatar.”
Acts of magic, conducted as part of the spirit’s servitude, were rendered by his digital counterpart. By contrast, in moments when Ariel presented as his embodied self, exercising his own agency, he appeared as essentially human.
Of course, this rule necessitates some dramaturgical wheedling. Ariel is something of an enigma in the canon, a slippery Protean figure with no clear textual precursors and seemingly infinite elemental powers. As such, to draw a line between Ariel as “magical manifestation” and embodied self is a little tricky in practice. One could arguably distinguish between Ariel and his presentation of other figures—a water nymph, the Harpy and the goddess Ceres. Yet Doran’s production extended the use of Ariel’s avatar to scenes earlier in the play, when the spirit describes how he “flamed amazement” aboard the King’s ship, and as Prospero recalls Ariel being knotted up in a cloven pine.
In short, the mo-cap avatar established a schism between Ariel’s “art” and his authentic self. Acts of magic, conducted as part of the spirit’s servitude, were rendered by his digital counterpart. By contrast, in moments when Ariel presented as his embodied self, exercising his own agency, he appeared as essentially human. At first, this presents itself as a simple dichotomy: simulation as artifice and embodiment as authenticity. Yet as Michael Billington noted in his review for the Guardian, the simultaneous presentation of these two Ariels, subject and agent, produced the uncanny effect of watching a “Double Ariel,” in which it was unclear whether the digital or embodied Ariel was the dominant spirit.
A Touch of a Button
Here’s where I have to nerd out a little. One of Imaginarium’s major technical innovations for this production was the minimization of latency in streaming a live mo-cap feed. The data captured from Quarterly’s costume (a modified mo-cap suit with 336 individual joints) was sent to Unreal Engine (a gaming engine which applies the motion data to the digital avatar), which in turn sent out a video feed via SDI to d3 media servers, which maps the content in real time to the geometry of the surface, then the signal is sent over fibre cable to the projectors, all with a latency of only two to three frames. And if that’s making your head spin—all it means is that there was essentially no visible disjuncture between Quarterly’s movement and that of his giant avatar (with an interesting exception I’ll mention shortly).
Not only that, because Unreal simply renders out a conventional HD video feed, Ross could place Ariel’s avatar anywhere on the stage, which he mapped with a whopping twenty-seven projectors. The scenic design included two suspended gauzy structures—the Vortex and the Cloud—onto which Ariel’s avatar could be projected, with the effect that the avatar was floating mid-air. This spatial flexibility amplified the physical distinction between the embodied and digital Ariel; as Ross noted, “reassigning the mapping is just a touch of a button,” while Quarterly’s movement was largely restricted to a horizontal plane upstage and the balcony.
Interestingly enough, a lot of this blocking was reconfigured later in the process. Doran had initially blocked Ariel center-stage, interacting with Prospero directly, but it quickly became apparent in technical rehearsals that Quarterly’s movement needed “a significant amount of…reconception” in order to clear the visual field for the mo-cap avatar. The reconfigured blocking resulted in a spatial oscillation between puppet and puppeteer, with Quarterly and his avatar alternating in their physical dominance over the stage.
As I talked to Ross, it became apparent to me that—as Adams suggests—Ariel’s “doubled” characterization not only informed the mo-cap, but was informed by it. One excellent example of this was the use of a facial mo-cap rig when Ariel appears as the Harpy. Now, anyone who has ever worked with facial mo-cap can tell you that is a significantly more challenging endeavor than body mo-cap. Unlike body mo-cap, where the movement of each joint is essentially captured by a glorified accelerometer, measuring the joint’s pitch/yaw/roll, the points of articulation in a face are often only a few millimeters apart, requiring extremely high resolution capture and fidelity.
[T]he technical accomplishment of using mo-cap in live performance is the ability to display to an audience that the data is being captured and rendered simultaneously. …However, the simultaneity of the digital or embodied Ariel also produces the uncanny effect that neither the airy or material spirit is really in control.
Accordingly, Ariel’s facial rig produced a visible lag, with the team resorting to “heavily amplified and affected” microphone effects for Quarterly’s vocal performance, in order to create an artificial delay that would synchronize his voice with the facial capture. For the rest of the production, the avatar merely “recalled” pre-captured facial expressions—happy, sad, angry. These were intricately choreographed to mirror Quarterly’s performance—again to produce the effect of total simultaneity.
All this evidences the team’s extraordinary efforts to reduce, and even conceal, any perceivable latency between Quarterly and his avatar. After all, what some would call the “gimmick,” but I would call the technical accomplishment of using mo-cap in live performance is the ability to display to an audience that the data is being captured and rendered simultaneously. Thus the “quaint device” that is live mo-cap necessarily blurs the line between puppet and puppeteer. However, the simultaneity of the digital or embodied Ariel also produces the uncanny effect that neither the airy or material spirit is really in control.
I would argue that some of the critical antagonism is due to this dramaturgical complication. The production clearly distinguishes between the digital Ariel conjured as part of Prospero’s designs, and the analog Ariel protesting his own vassalage. Yet mo-cap blurs the line between the dominant and dominated Ariel, thus ultimately obscuring the character’s “authentic” motives. However, I would argue that this actually makes a strong case for the use of a digital puppet in this play, insofar as this dramaturgical issue arises from the play itself. The avatar simply augments Ariel’s slippery textual status, establishing Ariel as a bifurcated spirit fluctuating between two motivational poles.
In Ariel’s final scene, Ross described how “it’s quite…a lovely moment, when Prospero sets Ariel free, there’s absolutely no technology at that point.” This was part of the larger design concept; Ross eliminated all video elements after Prospero breaks his staff, to illustrate that “the magic layer’s stripped away.” This climatic moment clearly established the victory of the embodied Ariel over his magical counterpart. Yet for an audience equally enchanted by his digital avatar, I imagine Ariel’s attainment of freedom feels as bittersweet as it does to Prospero, when he declares “Now I want/Spirits to enforce, art to enchant,/And my ending is despair.”