Do People View All 360°?
blog.vrtigo.io
blog.vrtigo.io
Conventional storytelling in visual media is all about directing the viewer's attention to where you want it.
Back in the 1990s I did some work with VRML and found that with the user interfaces of the time the user would usually turn his back to what he was supposed to see and then not be able to find it.
If you wanted to break the "looking ahead" pattern you'd want to design the scene so that important events happen in different places; one thing you will need is good audio cueing, and "stereoscopic" sound in headphones rendered by HRTF won't be good enough -- you probably want 22.4 sound, but a properly mixed 5.1 track and an optimal speaker system might be enough.
Also I have no idea how you light a 360 degree scene, both in the sense of "everything looks good" and "you don't see the lights".
Even in outdoor scenes, Hollywood-quality work requires the use of lights, reflectors and other tricks -- if you use real actors and don't light them well they'll go to the union and complain.
There is a large precedent in video game storytelling but even there, critical moments in the story are usually in cut scenes or rendered using "Visual Novel" conventions.
It was a natural step from HL1 as players became more comfortable navigating 3D game worlds. I'm wondering if they're waiting for that moment in VR before choosing to make another story-heavy game like that.
It was actually a revolutionary new story telling technique at the time that did "keep the action going" because you could explore the room as the characters talked.
Yeah, HL1 was was revolutionary in that regard.
I did not enjoy HL2 or its successors as games.
Games like Skyrim and GTA I like because most of the time the stuff you find hidden here and there are interesting things happening in every corner you look. But some of the modern takes on CRPGs like Divinity: Original Sin, I have a hard time with. They encourage you in tooltips and loading screens to talk to everyone, try to trade with everyone, because some random person might be super important. But 90% of them are useless. So I waste how much time searching empty chests and pointless NPCs to find maybe one out of 100 that's actually important?
I don't like 100% on-rails stuff, but I do like a little guidance. I don't play games to waste time, I play games to be entertained. I really liked HL2's guidance. It's one of those games where you play it as a cinematic experience, not to explore an vast and complex world. There are plenty of games on both sides of the coin.
In fact, there were parts of the game that told an "ambient story" that were totally skippable during the driving scene along the coastal highway. The only reason you get out of the car is if your curiosity to see what's in that damn house gets the better of you. Otherwise it is total inconsequential.
https://www.youtube.com/watch?v=1vr6n53jPR0&feature=youtu.be...
Headphones are definitely enough.
Games and other similar interactive content seem like a much better use of the technology.
Absolutely agree. There has been exactly one video where I was constantly spinning around 360 and it was the Lion King musical 360 video (1).
The reason it kept my interest was 1. things are happening 360 the entire time, and (more importantly) 2. I got to see a view that I had never seen before (that is, looking from the stage into the audience). If the Lion King video had been shot in an open field, I doubt I would have been spinning around nearly as much.
TBH, I find 360 video tiresome. It gives me a feeling of anxiety (low level annoyance, not full blown panic) that I might be missing something interesting if I'm facing the wrong way. I really don't find the payoff to be worth it. I feel like 360 video is going to end up like 3D TV.
You get higher fidelity content given a smaller space as well. People will choose the higher fidelity option where they are can conveniently view from their couch without turning around. Its a chore. We like to "veg" and just enjoy the movie/game.
Even the entire 180 sphere may be too much. People don't want to have to look straight up or straight down. There will be a sweet spot or "sweet zone" as more content starts to utilize VR devices.
* The play is set in a fancy nightclub, with half the audience sitting on the stage and the actors also moving around the rest of the audience.
* The play is set in a restaurant, and the servers are actors (unknown to the audience until the play starts and your waiter starts participating).
* The play is staged in a large room; the audience moves to different parts of the room, among the set.
* The stage is essentially a long, wide hallway between two long rows of the audience.
On a related note, I've seen many plays where the actors engage the audience members directly, asking questions, having them participate; in one we were the class for a pair of teachers. What I've seen has worked very well; it's not corny or hackneyed, but smoothly integrates with and enhances the performance.
I've seen it so much recently that it can't be coincidence; it must be a trend in the theater.
Good theater doing this sort of thing uses it for emphasis. To force attention to shift or change or to make it clear to you that "something has changed".
Likewise, audience participation is there to give the audience the sense of being active participants, but still without demanding too much of them (unless it's really active improv or something). It has particular power, I think, in plays that are somewhat confessional in nature. Drawing the audience in to the conspiracy/shame/excitement is powerful there. But if it's a standard narrative I think it bogs things down a bit.
Agreed. This one and the others are all excellent points.
In the past I found these endeavors to be gimmicky, but recently it seems like someone has found the winning formula. Despite being completely comfortable in the theater, I still find that the recent attempts have eliminated a last bit of reserve and distance in my mind.
I'd love to know what the new theory is; what is the model or idea that all these productions are embracing? It can't be coincidence.
In the same day I watched a short animated film in 360 and the experience was just annoying; it wasn't always clear where to look and at some times 50% of the action or character expression was happening behind you. It did not enhance the experience at all.
Or they were sitting down?
However, we did notice that the majority of videos with lots of action tend to center their content at 0 degrees.
Of course, most of the action still happens in front and to your sides. Not so much behind. That seems like a reasonable compromise with viewer attention, though.
http://store.steampowered.com/app/496190/
There's a also a 2D opera version of the same concept, which you can find on YouTube. The VR version basically puts you in the role of the movie's (main? viewpoint?) character.
I believe the reasons why so few people turn around and stay long enough to look behind them are the same reasons why they don't do so in real life.
Sure, if there are distracting/scary noises behind you you turn around, but otherwise you turn your body and concentrate your attention on something specific. And within that, at the less macro scale you foveate on the specific area of attention (hence their observation about the wider FoV of a headset...which isn't completely well attested IMHO).
So the filmmakers are doing what people do and want.
Not to put too fine a point on it, but we recently created a 360 video of a "Goat Yoga" session. Everyone involved in the yoga was in a circle around the camera with all the goats in the middle. The people watching the video are encouraged to stand at the beginning of the video or follow along with the yoga instruction. There's no narrative and yet we have analytics that show multiple people have viewed the video several times because they are the center of the action and it doesn't matter where they look, they still get a similar, but somehow custom, experience.
People also don't look up or behind them and especially don't look up and behind which is why rides e.g. at Disney leave pretty much all the rigging out in the open above and behind. Which makes the rides extra interesting if you know where to look.
A dataset of where the resting head position is for people from locales which drive on different sides of the road, and of people with differing hand-dominance would be interesting.
I'd expect a lot of VR users to be young gamers who don't yet drive?
That's all to say: the wiring of our optic nerves and visual cortices are seemingly quite symmetric—but there's more to the eye than its connection to the brain.
The thing that bothers me the most is hosts that move... I don't want to move my head to watch the show!
Graphics also bother me.
If I had my choice: Minimal on-screen graphics and if any, they need to be duplicated and actually used in a 360* way ("Look behind you to see x") and no hosts, just narration.
the whole thing of "show the title in 4 different places around you so you won't miss it" seems like a poor hack, and it feels like being able to include a separate "stream" which will be pinned to your view would alleviate a lot of the problems.
If you want to get fancy you could allow some programmatic interaction between the "view" and the "world" to give some kind of stylistic hint on where the user should look, but i'm not really sure how that would work, and at what point do you stop making a movie and start making a videogame?
I think producers have to rethink it from the ground-up rather than trying to force a model for one format into a new format.
I've seen a few 360* videos that are well-made and were obviously made specifically for 360 but others are just like "We got a 360 camera, then stood in front of it and did the same thing."
The 3D scene can be rendered as usual, and any "must see" information can be rendered onto an obstructing-but-transparent 2D surface attached to your avatar's nose.
When I've played with VR at friends' houses, most of the time we've played sitting on the sofa or sat in computer chairs. There's a bit of craning, e.g. when playing flying games, but often you only need to do a bit of left/right head movement.
Looked for a source to corroborate, but I heard it in conversation with Jaunt people.
Another good one is the SNL Seinfeld monologue https://www.youtube.com/watch?v=6HS9h4xFRww. Because of the audience participation you crane your neck around to see who's talking.
I much rather prefer 180 Steroscopic, these are videos that are two videos either Side-by-Side or Over/Under. Then mapped to each eye.
180 fills enough space which is supported by these findings in the post but give you a much better feeling of immersion.
Conceivably Lytro's light field camera could give you stereostopic 360 video, but the amount of data required for separate eye perspectives in every direction is colossal. The camera apparently captures video at something like 300 GB/s.
We also use subconscious cues to let us know important things are happening outside of our FOV and forces us to change it. Has VR reached the point where it can make those cues possible?
It seems to work ok, but I haven't yet done much user testing yet to see if it makes you puke or gives you a sore neck. It's so obvious I wouldn't think of patenting it, and I'd much rather share the idea and benefit from other people's experience.
The jury is still out on whether it sucks or not, and you really have to try it to know what it's like, but I'd appreciate feedback from anyone who's seen or tried anything like that, wants to hack it up and try it themselves, or even links to research, thought experiments or rules of thumb anyone has which might apply.
It's very simple: roll your head to the right to yaw to the right, roll your head to the left to yaw to the left. Of course there is a neutral "dead zone" so you have to roll enough past the threshold to start turning, and the speed you turn ramps up as you roll even more. The ramp can plateau at a maximum speed to keep you from spinning too fast. The range of the dead zone and the slope and plateau of the ramp are configurable parameters, which need to be tuned.
Head roll or bobble isn't a common gesture in my culture, so I thought it would be good to use for turning around, since you wouldn't do it accidentally very often (once it's properly tuned), and it's easily reversible. However the head bobble is more common and has meaning in Indian culture! [2] So I'd like to hear from anyone who bobbles their head on a regular basis!
As well as tuning the ramp parameters, it is possible to filter and debounce the gesture so you didn't turn around accidentally when you bobbled your head expressively.
The roll-to-yaw interface requires more tuning and experimentation of course, but since rolling your head is a much less common gesture than pitching up and down and yawing left and right, it seems useful to use it for turning (yaw) around.
I think it's important to have a way to turn your virtual body around without actually turning your physical body, so you can still comfortably view the entire 360 degrees of immersive video while sitting down or attached to the computer by wires.
Here's one of my favorite papers that has some great ideas about pragmatic issues which apply to all kinds of user interfaces: Buxton, W. (1983). Lexical and Pragmatic Considerations of Input Structures. Computer Graphics, 17 (1), 31-37. [3]
"PRAGMATICS
In examining the two studies discussed above, one quickly recognizes that the effect of the pragmatic level on the user interface, and therefore on the user model, is given very little attention. Moran, for example, points out that the physical component exists and that it is important, but does not discuss it further. Foley and Van Dam bury these issues within the lexical level. Our main thesis is that since the primary level of contact with an interactive system is at the level of pragmatics, this level has one of the strongest effects on the user's perception of the system. Consequently, the models which we adopt in order to specify, design, implement, compare and evaluate interactive systems must be sufficiently rich to capture and communicate the system's properties at this level. This is clearly not the case with most models, and this should be cause for concern. To illustrate this, let us examine a few case studies which relate the effect of pragmatics to: 1) pencil-and-paper tests of query languages 2) ease of use with respect to action language grammars 3) device independence"
[1] https://en.wikipedia.org/wiki/Head_bobble
Where people look in VR depends on the content. That's all there is to it.
That blog post is terrible advertising for your company, as it shows that you will gladly create bullshit content around a trivial fact to boost sales.