First, a disclaimer: As it so happens, I’ve come to know some of the producers of this excellent (my non biased opinion) episodic VR show, and while I do stay true to what the RealVision knowledgebase has stood for, for the past 8 years, I will admit the critique if any, will be written keeping relationships in mind.

To qualify why I think this VR series is excellent, one has to know that producing in 360 (VR) is no trivial matter, especially on an episodic basis. To that end, Invisible has broken new ground – good VFX, the cuts work*, and… there’s story! The story makes you want to go from episode to episode, and although each episode lasts on average about 6 minutes, I won’t be surprised if Invisible becomes the first VR production to initiate VR binge watching.

Now that Hollywood is getting into Cinematic VR, (you do know who’s directed this series, don’t you) what can you do to impress Hollywood and score the next narrative VR gig? Let’s dissect the trailer of Invisible, and see what we can learn:

invisible_no_stereo_ana

Skill 1: To Qualify for the label “Cinematic VR,” shoot in stereoscopic 360:

Otherwise, it’s just QTVR from the 1990’s being watched in an HMD – nothing more. It’s good to know, Invisible is shot in stereoscopic 360. However, the more astute among readers will take a look at the image above and quickly protest – Yes, you’re right – This scene is devoid of stereo. It’s a still frame from the 360 stereoscopic trailer, but for some reason there are mono scenes/shots such as this one, scattered throughout.

In fact, the Invisible trailer is a good example of how Stereoscopic 360 immerses you in a scene (those scenes that have been post cured) and where the 2d 360 scenes with larger than life giant actors towering above you, quickly pull you (me) out of the scene.

If there’s one skill you want to hone; it’s learning that mixing stereo with mono, might “demmerse” a viewer in VR. It’s jarring and noticeable. There are some scenes where 2D 360 does work in VR; those are times when we reach the limits of stereopsis and other depth cues such as motion parallax etc. kick in. Think of a flight over the Grand Canyon, or indeed, the establishing fly over in the Invisible trailer.

Side note: Don’t run the risk of sounding like an amateur by stating aloud that anaglyph is not the way to judge stereo – Stereo veterans use anaglyph to judge a scene quite accurately for depth budget without even wearing those glasses.

invisible_bad_stereo_no_cleanup2_ana

Skill 2: Cure your VR, in Post

Time to get those red-cyan glasses out. You’ll see the many stereo anomalies in the scene. Again, a quick glance without the glasses will tell you some of the bigger ones that can lead to eye strain – the lower torso of the main male actor, the upper left overhang of the wall above the stoves, the patched in monoscopic nadir.

I’m assuming this is the much vaunted Jaunt camera (I’ll correct the article if I’m wrong). But if it is, then, imo – it’s not worthy of being called a Cinematic VR camera if it’s got a huge nadir patch to be filled in, in mono. That’s a sizeable crater right there!

While Nokia’s Ozo does a decent job from top to bottom, it falls flat (well a little snark does no harm) in the rear, and that can be just as disconcerting. As an example – Try watching the recent NBA video in the GearVR, standing up. There’s a few scenes where you find yourself in the middle of the basketball court and if you look behind, the floor slopes so steeply (because the Ozo goes to mono at the rear, by design) it’s bound to give you vertigo or throw you off balance.

invisible_slum_ana

Back to scene – You want to be skilled in Nuke and CaraVR or at the least have a few Nuke ninjas on the team.

Mettle’s SkyBox Studio also helps a lot, when curing already shot stereoscopic 360 footage. In the case of Invisible, leeway should be given, taking into consideration this was produced on an episodic basis, but going forward, It’s crucial that ceilings, nadirs and other large stereo anomalies be cured in post. There’s quite a few scenes in the trailer with stereo anomalies, the picture above is another example. By now, you will have become stereo experts – see if you can spot anomalies without wearing anaglyph glasses, then reconfirm with them on.

invisible_unmotivated_camera_ana

Skill 3: Motivated Camera moves in VR.

You’ll have to see the trailer again to see why the importance on motivated camera placement in VR. If this scene was to show the POV of a robotic stealth camera as used by Special Ops under a doorway – then it makes perfect sense, but, and again, imo – based on the point of view of the next scene where we find ourselves almost level with the girl on the bed… the camera placement didn’t work.

It’s always tempting, to a Director who’s been used to regular Hollywood filmmaking to look for opportune camera angles to tell a story (low angle looking up, to denote power in a scene featuring an antagonist – a fly on the wall looking down to show a birds eye view of the scene or to denote CCcam surveillance etc). Montage over mise en scène and deep staging, is how you paint excitement and depth in 2D. However, all that changes in VR where one of the first things done is to:

  1. Choose if the camera is the audience and the scene is playing out in FP-pov, or
  2. If the camera is third person but yet capable of invading personal space (allows for OTS shots with such realism, that if done well, you can see (feel?) sweat beads trickle down the back of the neck of the actor).

This voyeuristic capability that VR affords the audience, is where it’s at.

If you agree with the above, guess what – you’ve just developed a very powerful skill in VR storytelling: You’re leveraging a shortcoming of VR, and turning it around, to your advantage. What’s the shortcoming? – The fact that on the best of today’s VR displays – even Medium Closeup (MCUs) look like mud as far as resolution is concerned. CUs is where the immersion is!

invisible_unmotivated_camera_fly_ana

Now, this is not to say you can’t have an omniscient POV in VR. Indeed, a fly-on-the-wall camera placement is cool in VR, especially when you want the audience to take in all the action at one go. But, Camera motivation is key. If there were for instance, a few security cams visible in the room, then one of them could have been “us” looking down. That would be ‘motivation’ for capturing the POV as shown in the image above.

Below, is another one – This would look extremely stylish in a tent-pole blockbuster movie. In the Invisible trailer, because I feel I belong in the scene as a bystander seeing the protagonist rush by, it keeps me wondering if I’m manning a periscope view from a man-hole. (It just might well be, in which case, I sit, corrected!)

invisible_unmotivated_camera2_ana

Again, this is not a rule written in stone – but usually in VR where the camera is “us” and we are an Invisible voyeur in the movie, alternating from being david and then goliath without drinking any magic potion is a little disconcerting. Working out blocking, motivation and a depth budget and Depth Continuity in stereoscopic 360 is an asset that will allow one to score high in Hollywood VR filmmaking.

invisible_wasted_audio_immersion_ana

Skill 4: Arouse your audience through audio:

In fact, if you have the liberty – stage the scene so the immersiveness – the presence – is heightened via the use of surround audio. In the scene above, if we were at the POV of the actor – say if the actor were running and fell down to the mud, the heightened sound (ear-to-the-ground) of tyres coming to a halt on gravel or wet earth – Cue foley excitement- would lend a sense of greater immersion. Seeing dirt (anything organic) in stereoscopic VR is something that seems to tickle the primitive brain.

In VR, the devil and presence is in the details.

Side note: The Jaunt ‘crater’ is very present in this scene. Wear your red-cyan 3d glasses to see it, or see the whole trailer in stereoscopic 3D (needs to be downloaded not streamed for 3D)** on the Samsung VR app in the GearVR. There was also one scene which looks to be the interior of a refrigerator with leaves and a few testtubes in it, that was shot in monoscopic 360, making the test-tubes look like huge radioactive fuel rods. This leads me to believe the Jaunt Camera couldn’t be placed due to size, which means there is an industry for small form factor stereoscopic 360 cameras.

Over all, Invisible looks to be the first Cinematic VR series done well. It can rightfully lay claim to that title, and the story is very engaging. I’ve not seen all the episodes yet, and waiting to download them to binge watch in VR!

* The cuts, for the most part work. However, with the scenes playing out in 360, if you decide to turn your head (for example in one scene where the invisible man goes thru ‘you’ in the hospital scene) then subsequent ‘main action’ plays out behind you, forcing you to keep your torso twisted. A solution, going forward would be 1) Samsung updating their Samsung VR app to recognize a quick tap on the trackpad to reorient “center” or 2) combing through a script and initiating an “eyeblink” cut, back to center, (so people can untwist/un-rubber neck) thus re-framing the main action in the scene.

** I’ve been told by Samsung (thanks Andrew) that it’s no longer a requirement to download to experience a video is Stereoscopic 360 in the Samsung VR app for GearVR, but for the best experience ( non varying streams ) downloading makes sure it stays at the best quality.

 

  • DennisonBertram

    how are we supposed to see stereo at the zenith/nadir?

    • If you really want to, this link has good info. http://www.vrcolony.net/tutorialsblog/2015/9/19/3-camera-zenith-for-better-3d
      Or, you use Nuke+CaraVR with a good VR camera rig and Nuke’s slitscan renderer will give you the next best output for stereo towards the zenith/nadir, short of realtime engines rendering perfect zenith/nadir stereo.

      However, in the article, the point was about using a camera system without a huge “mono” crater. If you look at the third image in the article (anglyph of the girl on the stone steps) you can see right from the bottom of the backpack it goes mono, and the tops of the Huts to zenith are mono.
      This leaves only a slice in the middle showing stereo 360.
      My opinion is that the stitching algorithms/software is not up to the mark. Computational photography or not.