Friday, July 01, 2016

Mo Cap Performance

Maybe it's not "the devil's rotoscope"?

'The BFG' Visual Effects Whiz Joe Letteri Makes the Case for Motion-Capture Performances

The film combines live action with a performance capture giant played by the Oscar-winning Mark Rylance.

... “Each character comes with its own challenges and you apply what you’ve learned to each new project so we’re constantly improving the quality level,” Letteri told The Hollywood Reporter. “Our recent work on the BFG required us to deliver a 24-foot giant whose performance was so close to what Mark Rylance did on set that you’re almost asking the audience to see through the digital character and emotionally connect with Mark.

“Mark used extraordinary subtlety in his performance of the BFG and it was up to us to understand how his acting manifested itself visually so we could make our giant convey the same thing.” ...

The thing about motion capture is, when it's used in conjunction with live-action, it tends to be reasonably effective. Planet of the Apes and Avatar were all movies audiences embraced. The mo cap was well-integrated with live-action characters and environments, and the movie-going public beat a path to the movie-makers big-budget features.

With wall-to-wall mo cap, specimens like A Christmas Carol or Tin-tin don't perform very well. There is a deadness in the eyes, a stiffness in the performance, and more than a hint of "The Uncanny Valley" that makes theater-goers a wee bit uncomfortable. It's not pure animation and it's not live-action, but for most people it seems a trifle weird.

How Spielberg's BFG ultimately performs is still up in the air, but the creators were smart enough to blond motion capture with real people, so there will be fewer creeped-out audience members. Uncanny valleys will be kept to the lowest possible minimums.

Add On: What's often overlooked in "mo cap" movies, where publicity departments (not to mention the directors of these films) always tout the actors and their performances, while ignoring the people at the computers enhancing and enriching images on the screen, including the histrionics.


Unknown said...

As a Motion Editor (body animator) who has worked on Avatar, Dawn of the Planet of the Apes, AND The BFG, we need to clarify what "The Uncanny Valley" truly is.

Mocap is present in most large features and goes widely unnoticed - which is a good thing - but is still generally misunderstood. Typically, Mocap is still not used as best as it could/should be.

When properly used (editing/additional animation), motion capture is the best way to deliver a life like performance to a CG character. Occasionally we don't see that because the artists aren't familiar enough in working with the data.

Most artists working with Mocap data are Animators working in Maya - Motionbuilder if they're lucky, Nuance if they're ultra lucky - and have never worked for a Mocap vendor prior to their current show. So sometimes the data we see on screen has to be killed in order for the artist to have simplified curves that are easier to work with in the toolset they're using (usually Maya). This makes the fidelity far less than the original motion, possibly contributing to the "stiffness" look. The successful films you mentioned don't have that because seasoned Motion Editors working in Nuance delivered unkilled curves, performance preserved and enhanced data.

I've been working in Mocap for 10 years and have grown tired of hearing people blame Mocap for a films demise. "Dead eyes" or "The Uncanny Valley" are specifically referring to the facial performance - which regardless of facial capture technique - is still heavily based on blend shapes and facial animation.

Body capture has been developed over 20 years ago, facial capture is still fairly new. The industry is constantly coming up with new ways to re-target a performers face onto a CG characters face, but it hasn't had the same amount of resources, testing and time that body capture has had. Also, we look at faces EVERYDAY, so when the slightest thing is off, we will 100% notice.

The face will always be the most difficult thing to reproduce in a photo real way. We've made HUGE leaps, but there's still work to be done. Some of the high level Hero facial rigs have over 1000 shapes to dial in the most perfect and accurate performance, which we've seen on the films you've mentioned.

Industry wide we're pretty dam close, but we also have to take into account HOW this tool is used to help the story - does it help or is it just a gimmick? Does the director fully understand it's use? Do the performers understand that they are puppeteering a CG character? Did the artists have the right tools for the job?

My point is that there are a variety of issues that come into play with Mocap films that can contribute to the "Uncanny", not just the "Mocap".

Thanks for the Add On message as well - most of what we see on screen the artists never get credit for. It takes hundreds of people to make the Golems, the Kongs, the Apes, the Nav'i, the Giants, the superheroes, the epic battles etc etc...


Site Meter