Wednesday, February 01, 2017

The VFX Predictinator Was Completely Wrong, Part 1

"Ex Machina" winning the Oscar® for Visual Effects

What is The VFX Predictinator? Start here.

2016 was a year of earth-shattering upsets: the Chicago Cubs won the World Series. Donald J. Trump won the American presidential election. And “Ex Machina” won the Academy Award for Best Visual Effects.

Needless to say, we were shocked by “Ex Machina”’s win at the 88th Academy Awards over here at The VFX Predictinator Headquarters.  This upset didn’t just shake the pillars of the mighty VFX Predictinator, it shattered the entire philosophy into smithereens.

The visual effects category has seen a few upsets in its day. Typically, the award goes to the behemoth of the category (think “Avatar”, “Jurassic Park” and “Forrest Gump”). Most recently, the most memorable upsets include “Hugo”’s win over “Rise of the Planet of the Apes”, “Babe”’s win over “Apollo 13”, and “The Golden Compass”’s win over “Transformers”. As shocking as those upsets were, our VFX Predictinator accurately predicted those races. (Yes, even “Hugo”.) We understand how those films won, since they each earned their victory with just enough points of criteria as dictated by The Predictinator.

We have no idea how “Ex Machina” won. Our formula, which correctly predicted the winner of the Visual Effects Oscar from 1989-2014, was completely wrong.

Its victory upends our philosophy. There is no model to explain what happened. Alex Garland’s film is imaginative, thought-provoking, and beautiful. The quiet, modest and understated visual effects that support the film are of the highest quality; the depiction of Ava (Alicia Vikander) was both technically flawless and aesthetically gorgeous.

I’ll touch on the highlights of last year’s numbers (to read the entire analysis, read The VFX Predictinator, 88th Academy Awards Edition). By all accounts, it was “The Revenant”’s Oscar to win; Alejandro Inarritu’s film was nominated for twelve Oscars (and ultimately won for Director, Actor and Cinematography), and soared with pre-awards Oscar buzz. It was a solid hit, and also earned key points since the primary visual effects for the film consisted of an organic creature. “Mad Max: Fury Road” was close behind in the scores; the film was also a critical darling and ultimately won a staggering six Oscars from its ten nominations. “The Revenant”, “Mad Max: Fury Road” and “The Martian” (another visual effects nominee), were all nominated for Best Picture, as well.

Then there was that universally-loved, gargantuan blockbuster which earned over $2 billion, “The Force Awakens”.

In contrast, “Ex Machina”, while highly respected and a critical darling, was barely on the radar. Yes, it earned two Oscar nominations (Visual Effects and Original Screenplay); in contrast, the other visual effects nominees earned five, seven, ten and twelve nominations. Surprisingly, “Ex Machina” didn’t even earn a single nomination at the Visual Effects Society Awards, which has eight feature film categories. The film earned $25M in the U.S.; the film with the next lowest box office score was “The Revenant”, with $119M.

From a purely statistical point of view, these numbers simply don’t point to a victory for “Ex Machina”. In fact, the film earned the least number of points from The Predictinator of the five nominees. Looking at the scene from a less numbers-driven perspective, the visual effects for “Ex Machina” were beautifully designed and executed, but one could argue the challenges tackled and achieved for the film did not rise to the level of past visual effects Oscar winners, which made its victory surprising to so many industry watchers. Some voices have expressed a similar sentiment for “Babe”, “The Golden Compass” and “Hugo” as well, but those victories are, for the most part, explainable. I have yet to hear a rational explanation of how the Academy’s 6000+ members cast more votes for “Ex Machina” over its competition, probably because I believe(d) so strongly in the philosophy behind our formula.

Let’s look at some of the theories I've heard that attempt to explain the “Ex Machina” win.

This theory states that the front-runners’ votes (in sum totaling a near-majority) were split right down the middle, allowing an underdog to sneak in and grab a victory. Let’s say the front-runners were “Mad Max: Fury Road” and “The Revenant”, and they each earned precisely 32% of the vote. As front-runners, the two films garnering 64% of the vote seems plausible. That leaves “Ex Machina” to swing in with 35% of the vote and win victory, right? But this only works if zero percent of Academy voters voted for “The Force Awakens” or “The Martian”, which isn’t likely.

Okay, so what if “Fury Road”, “Revenant” and “The Force Awakens” split the near-majority vote? For this to work, each of the three films would have to earn precisely 24.9% each, leaving “Ex Machina” with 25.3% of the vote and the win--assuming zero votes for “The Martian”. As a reminder, with five nominees, the absolute minimum percentage of votes required to win is 20.1%.

This theory is based on the idea that actors, the largest single branch of the Academy membership, prefer to vote for performance-heavy films, typically shunning visual-effects driven blockbusters, even in the ‘technical’ categories such as Editing, Cinematography and Visual Effects. (Reminder, the full Academy membership votes for the Oscar winners, not individual branches.) Typically, this would mean the actors’ branch would shun films like “Transformers”, “Poseidon”, and “The Avengers”, which, as the theory goes, wallow in spectacle at the expense of human characters.

Giving some credence to this theory is the fact that Alicia Vikander, star of “Ex Machina”, won an acting Oscar Academy Award last year--but for a different film. She took home a Best Supporting Actress award for her work on “The Danish Girl”. Was “Ex Machina” riding the buzz and momentum that ultimately awarded Vikander an Oscar? This idea becomes far less likely once one is reminded that “The Revenant” featured megastar Leonardo DiCaprio, the actor who took home a Best Actor role for the very same film. It’s more likely that “The Revenant” (starring a man who already had four Oscar nominations and ten Golden Globe nominations under his belt) would benefit more from the ‘halo effect’ of momentum than “Ex Machina”, starring the relative-newcomer Vikander.

In addition, The Predictinator attempts to account for a bit of ‘actor prestige’, with a piece of criteria dedicated to acting: ‘has the lead actor of a film previously won an acting Academy Award?’

This idea takes the point of view that the grand philosophy of The Predicinator is solid, and the Academy voting block, riding an un-predicted massive wave of goodwill toward "Ex Machina", voted for it in overwhelming numbers. There's no denying the visual effects for "Ex Machina" were beautiful and compelling; it rightfully earned its stellar reviews. Similar to Marisa Tomei's acting win for "My Cousin Vinny", or "Shakespeare in Love" winning Best Picture (over "Saving Private Ryan"), sometimes the 6000+ members of the Academy think differently than is expected; typically, the very next year, they go back to voting the way that is typically predicted.

Something has changed in the way visual effects are being evaluated, interpreted and understood. The art and science of digital visual effects have matured to such an extent that, perhaps, a new chapter in the history of filmmaking has begun. Our previous assumptions no longer are true, and therefore, our formula has begun to crumble.

In Part 2, we’ll discuss this idea and illustrate how it affected the results of The VFX Predictinator. Here's Part 2.

No comments: