I haven’t paid for a VR movie before; I felt no need to as most experiences were free. But when a big name Hollywood pioneer like Robert Rodriguez jumps in with a VR film, a $5 ‘ticket’ would be worth the price… or so I thought. A 4gb download later, I have an opinion. TL;DR, this won’t be a flattering review.
Those who can’t… criticize.
If I had a dollar for every time I heard the line, I’d be rich enough to well… fund a big budget VR film. There’s no point easing in, so let’s get down to it:
VR Films need VR Cinematography:
You’d think this is a no-brainer. You’re creating for an immersive medium and yet somehow Directors still fall prey to advisers with vested interests, and end up shooting 2D for 3D Scratch that … they end up shooting 2D for VR.
A 2D cinematography mindset is the singular reason why Cinematic VR is lagging.
It was bad enough in regular 3D films, now it’s rearing it’s ugly head in 1/2 VR films too.
2D to 3D conversions in VR, suck:
They suck big time, for exactly the very reason Robert Rodriguez hopes to evangelize VR as an immersive platform, better than Imax (paraphrasing his words in the featurette with the film). In VR, yes, there is no distance between the viewer and the screen. One is wearing the screen on their face and so it’s even more obvious when something’s “off” with the characters in front of you.
I’m not talking about the technical caveats of poor conversion either, as seen in the closeup image above (look at it though anaglyph glasses and see the background pulled into the male characters neck area)
No, that is not the main problem here. The problem is de-immersion. Why?
In VR you can’t cheat…
A lot of Cinematic VR films are actually flat 360 panoramic video; Apple called it QTVR back in the ’90s. In choosing to shoot with a single camera with a fisheye lens and converting the whole thing in post, Robert Rodriguez has fallen victim to the same thinking that plagued 3D filmmakers:
- Pacing for 2D – When you’re wearing an Imax Screen on your face, *YOU* are a camera strapped to a roller coaster. If you have a hard stomach, by all means go for that ride.
- Pitch/Yaw and swish pans work for 2D not VR – I’m all for the “break the rules” mantra, but not at the cost of common sense. If you shoot for 2D, you have no idea what happens to the human vestibular system and senses when viewing in 3D (even if converted) in a VR headset.
- Scale: – When converting in post what you gain is less ‘retinal rivalry’ but at the cost of scale and ‘realism’. At the very least, the conversion could have been tuned for a VR headset. It’s not impossible to sculpt depth in post to maintain a semblance of 1:1 scale for VR and headset. However, it looks like the stereo-post dept. on The Limit, created the 2D-to-3D conversion for traditional cinema screens.
- We are Voyeurs in VR: Once filmmakers understand this powerful “presence” tool [viewers becoming voyeurs in a film], they will kick themselves for not using VR for the medium that it is – A way to truly break the 4th wall and draw the viewer into the story
What is the use of converting flat 2D pictures into “VR” when powerful binocular cues are lost in the process? People ‘connect’ with people when they are face to face. If characters necks and faces are victims of post-processed “depth”, it’s uncanny valley on a whole different scale (even if at a subconscious level)
When wearing a VR headset, the film world becomes the real world…
This film was shot in 180 and while I don’t have anything against 180 films passing off as video based “VR”, some of the lessons, especially the one on “Wobble Cam” outlined in this article would have helped the producers of “The Limit” understand the medium, better.
Sidenote: “Think in 3D” is as relevant to VR filmmaking as it was/is to stereoscopic 3D filmmaking.
2D-3D conversion can work better than native 3D in VR if done properly. We’ve done it on many VR projects seamlessly.
I agree with Thomas. The 180 stereo cameras are bad. Point. The misalignment off image center, the stereo depth budget on closeups and especially the parallax and perspective difference off center is killing the experience. Same with 360 stereo – computational stereo. It is so much more wrong than conversion. It also hurts, you get more retinal rivalry. This depth budget might not have been enough, the separation not clear enough, too much cardboard cutouts and other issues, but it is a step in the right direction. I still have to see a car interior shot natively in stereo VR without giving me a headache. Agreed on some of the movements, but that is a filmmakers choice. The common sense rule is not that common sense if a filmmaker wants to take control over your POV. Nauseating, sure it might be for many, but guiding you to look at the region of interest is part of film making and needs to be enabled to give this medium a chance with filmmakers. More than just audio cues. This is experience clearly enables peripheral look around more than the typical ‘What is behind me’ turn all the way around and miss what is happening in front. It is different.
Thank you for reading and your reply, Tina. You do make very good points, particularly that of perspective skew/stereo fall off on 180 stereo cams.
I feel the best way to shoot for 180 (stereo) is actually a custom custom “triad” config stereo rig to maintain stereo at the periphery.
This opens up a lot of leeway for stereo windows as the realestate is not full 360. I did like the near 200 degree enveloping view “The Limit” allowed for.
Regarding well done interior vehicle shots, I’d point to F&P studios “Space walk” when you’re in the lunar/mars rover with the astronaut guide besides you. as well as I remember a few others in cramped spaces (lebron james golf cart…)
Perhaps a tradeoff could be conversions for such cramped spaces and the front 180 muxed in (captured in stereo).
There’s just too much fidelity/subtleties lost on human faces with conversions. Again I’d point to scenes of people captured up-close in any F&P Studios productions versus converted humans – In VR with the screen up-close, the eyes/brains can tell sculpted depth on torsos/faces/hair and it is quite de-immersing.
Everyday people can’t tell these technical differences, but when compounded with regular “2D cinematography” styles, it’s little wonder, people take off a headset with either a ‘meh’ or “it was ok” response.
There is no wow factor.
Today, it looks like in the Hollywood Budget arena, there’s less than a handful of Studios who understand cinematic VR filmmaking.
I know one in the UK, the other in Canada.
Others will shoot flat 360 / 180 and convert, making it worse because they’ll only see the final results of the ‘cinematography’ after it’s been converted.
Thank you Clyde! I can not even fuse the ‘dashboard’ in the described scene of Space explorer. The parallax is beyond what should be used when strapping an hmd in front of you. Please take a screenshot of that interior and analyze it. The upper dashboard area is misaligned. It hurts me to watch this.
We have always done a hybrid model which makes sense, but converting the things you can easily shoot in stereo is usually easier.
Using 3 cameras doesn’t change the major size difference you get when shooting objects to the side of frame and nothing can change that, even modern computational algorithms, nor a good vfx artist. It looks worse.
Shooting native has been replaced with post conversion in hollywood for many reasons. Every big blockbuster, Marvel and Sony movie is post converted. One of the reasons is that you have a better control of the depth budget, which goes against the point you are making. I feel that is your opinion, when saying nobody but two production houses know cinematic VR. I cannot make it through Space Explorers without a headache, but I can make it through The Limit. Sure, I get squeezy at times but at least it doesn’t give me a headache due to bad stereo. That alone is a better cinematic experience. Everyday people can feel that as well.
I tried extracting a frame from Space Explorers (mistyped it as space walk earlier) but I think F&P is using a protected flavor of mp4 or my ffmeg filters are clashing. Either ways, I can’t respond to the defend the claim until F&P send a frame. I’ll ask over twitter.
I think terminology like “computational stereo” is the reason why the medium suffered even more (cinematic VR). Jaunt have bailed after that less than stellar attempt at a VR camera. They’ve now moved to AR.
The “Jaunt Crater” and sliver of stereo was a prophecy of things to come. Here’s one review: https://realvision.ae/blog/2017/03/on-being-vrs-keeper-a-k-a-hollywoods-still-learning-vr/
On Computational Stereo, the cringe worthy marketing spiel (chicken wire/mesh/ DNA/patents/Lightfield etc.) in this interview (scroll to about 3:50 into the NextVR youtube interview) here https://realvision.ae/blog/2015/10/is-cinematic-vr-really-virtual-reality/ …These are where the setbacks to video based VR are stem(ming) from.
>> “Using 3 cameras doesn’t change the major size difference you get when shooting objects to the side of frame and nothing can change that, even modern computational algorithms, nor a good vfx artist. It looks worse…”
On the triad config, I think you’ve not visualized it clearly. It’s three pairs of cameras arranged in a triangle config – specifically for shooting periphery filling rounded stereo (not for 360 3D coverage). You don’t hit the falloff on the fisheyes when processed in proper stereo aware software. I won’t name names as I don’t get paid.
You don’t need VFX artists to clean up anything, save dust/flares if present. Far faster than converting an entire 200 degree Field of view where even non humans (did you notice the “sketchup” models effect on the gizmos and control panels in “The Limit”?)
I want to “feel” I’m in that villain’s lair with the rubber/steel texture of those knobs/dials…what I got was 2D textured geometry primitives.
We’ll certainly end up in a long thread of opposing points of view, but it’s a healthy discussion and weblogs are showing people from some AAA studios reading it. (Sony Ent, Lucas film, MPC)… so it’s all good, if it advances the planning for upcoming video based Cinematic VR.