Based on a screenplay by Julina Tatlock and Oscar-winning screenwriter Melisa Wallack, the mystery series centers on an heiress’ quest to guard her recently deceased grandfather’s estate and the family's supernatural secrets. The story, told over five, roughly six-minute episodes, marks the first attempt at using virtual reality as an episodic-friendly, storytelling format while challenging many of VR's current known truths.
Autodesk M+E is making a VR experience! Follow along on their Journey to VR.
A scripted VR series has never been done, so what were your initial concerns with ‘Invisible'?
(Laughs) I had several. My first thoughts were on frame count and camera count. Jaunt VR works at 60 frames per second and that was something I had resisted. Now, I’m totally sold. 60fps gives you a really smooth playback and smooth tracking to give people the most seamless experience, and one that won’t make them sick. Jaunt also have 24 cameras on their camera system. I thought, ‘Oh my god, it’s going to be so much data’, and in terms of rotoscoping, it is. My thoughts also went to stereo, because working in stereo VR is hard and counterintuitive. Jaunt’s algorithm is pretty good at extracting stereo, and advanced for what it does, but it wasn’t perfect. We had to go in and fix a variety of little things.
Can you elaborate on the difficulties that come with stereo VR?
It takes time to get everyone up to speed: artists need to be trained to think in terms of left eye and right eye, in how to be efficient with roto and in how to extract tracks. If you’re shooting stereoscopic content with a camera as opposed to rendering it in pure CG, there’s a lot of physical realities that come into play. Filmed content has an analog interface to it, where it comes from the real world into a digital one, is going to come in imperfectly, but that’s in contrast to the CG material which of course renders mathematically perfectly right out of the renderer. And when you add to that the spherical projection that is VR, and how to deal with that in stereo, we spent a lot of time working with anaglyph glasses, finding methods for checking and fixing stereo that are efficient at the scale of 3600 frames per minute. We created five episodes for Invisible, at 6+ minutes each, so we had a total running time of over 30 minutes. That’s 100,000 frames of content we had to get out.
Tell us more about what to expect with ‘Invisible.’
You can watch it all at once, or by episode. You can watch it sequentially or out of order. The pieces link together and refer to each other but at the same time, each episode is a self-contained chapter in the story. Hopefully, people will watch them in order and get to build up to the exciting climax at the end. Whether viewers will binge watch it, or whether they’ll want to pace themselves and watch one episode per week more like traditional television, we don’t know yet. That’s part of the experiment. And I don’t say ‘experiment’ to sound as though we had no plan, I simply mean that we’re really curious to see how the public will respond. We want to know if the length works as is, or if it should be longer, more like a feature, or even shorter still? The metrics need to be found out so that we know how to proceed with future projects.
And how does perspective work?
The ruleset for the perspective isn’t strict. It does not say that it’s always from the POV of one person, or that it’s always from a specific height. What it does say is, ‘Here’s the interesting thing you’re supposed to be looking at and here’s how we’re driving the story, between cuts. Whenever you’re cutting in VR you have the additional and crucially important step of rotating the footage on each cut so that, where your eye is left at the end of one shot, your eye is near where we want it to be on the next shot. It sounds simple, but technically, the current tools available can get very slow doing these rotations, so it’s cumbersome right now. Using that technique, though, you can successfully cut, cut, cut, and know more or less where people are looking and direct their vision in that way.
What are you hoping ‘Invisible’ will accomplish and/or teach others about VR?
We’ve seen, two, five, or 9 minute pieces in VR so far – but what happens when you’re in a world for a full half hour? What transformation takes place in the viewer? Do they feel like they live in the world? Do they want to be passive and watch, or do they want to be active? Do they become exhausted, will their eyes tire viewing this VR space? This was an experiment that we’re putting out in the world, to find out what appetite exists around this format. We hope Invisible will answer many of those questions, and that will be very valuable information. We’re not slavishly focused on simply spewing VR content, we’re wanting to generate the right kinds of experiences and in finding something that successfully appeals to the appetite of viewers.
Have you come out of this with a solid grasp of what it takes to tell stories in VR?
Oh, definitely, we have much more of a sense of what works. For instance, our shooting ratio on Invisible was 10:1. We did a lot of takes and did a lot of scenes that didn’t make the cut. It’s easy to look back at those now and see why they didn’t. We learned a lot about pacing and about cuts. Establishing shots, for example, don’t work very well, unless it’s maybe a very stark, strong contrast between the two scenes. We had to find techniques that would allow the viewer to identify what location they’re in.
How did you handle all the action, being that we’re told a lot of cutting should be avoided?
We did an interesting action sequence where you have to swivel around to see everything, and you have to watch it multiple times to catch every single piece – it’s for creative effect. It’s not supposed to be clear, measured or spoon-fed to the viewer. Invisible proves that there are compelling reasons to cut quickly and direct the viewer’s gaze with action and dialogue and let the viewer look where they’re going to look. Speeding things up is definitely something we explored successfully.
Did working with action director, Doug Liman, mean many of VR’s current rules were challenged?
He definitely challenged them but not in a defiant way. He’s shooting what he knows, that’s where he’s coming from. He knows what works in non-VR so he figured, why not explore it in VR? Admittedly, you have to find it in the edit, it doesn’t come for free. You have to find how to rotate the camera, how quickly to cut. In this case, a fair amount of visual effects were used. We fixed a lot of things to make cuts, and there was a lot of timing work. Not every production has the intent to have visual effects, so we had that working for us.
What were your biggest takeaways from your ‘Invisible’ experience?
That cutting quickly works – action works. Of course, shooting stunt sequences for VR is more complicated and takes even longer than for feature films. There is so much of the world to clear out, you don’t have a great preview system to see what you’re shooting, and there’s a lot to think about in terms of safety. If it takes three hours to do one take, you’ll only get three or four done in a day. We found out that there are many pain points with that. However, what we wanted to do was completely in contrast to the current mantra of what to do and what not to do in VR. We wanted to get into a world like that of, The Bourne Identity, Edge of Tomorrow, or something as full of action and as awesome to view as those were, and I think we were successful. We learned a lot about the medium in pursuing that.
Autodesk’s Maya, Arnold and Shotgun were crucial parts of the workflow for ‘Invisible.’
Watch all episodes of ‘Invisible’ on any device over at Jaunt.