In Defense of "Gravity" and Emmanuel Lubezki

As much of the world now knows, the Academy of Motion Picture Arts & Sciences chose to award Emmanuel Lubezki the Best Cinematography Oscar for Gravity. Despite the fact that his nomination was dependent on his occupation-specific peers within the Academy, Lubezki's win has angered many people in the cinematography world because he committed what appears to be the supreme sin of realizing his image creation almost entirely inside a computer.

The 86th Academy Awards were the third in a row to honor a 3D movie with the Best Cinematography statue, and the fourth overall since Avatar's win at the 82nd Awards. Like Lubezki this year, Maurio Fiore and Claudio Miranda were subject to tut-tutting and side-eye after their wins for the same reason: a gut reaction that "virtual" cinematography is a lesser achievement than employing real glass and material image planes. Robert Richardson was spared the animosity because, apparently, an acceptable proportion of Hugo was physically lensed.

The abreaction to honoring virtual cinematography appears to flow from some sort of sense that executing cinematography primarily with software is "cheating" or "heartless" in comparison to creating images largely in, shall we say, meatspace. This appears to me to be essentially a moral objection, and one that I will argue has no rational basis.

What, exactly, is being honored with a Best Cinematography Oscar? The answer seems obvious, given the straightforward name of the award, but let's unpack it a little.

To start with the elusive, qualitative and less relevant word: "best". For me, the "best" cinematography is that which succeeds to the fullest in serving the story, tone and emotion of the movie in question. "Best" does not have to be the "most beautiful", though the Academy generally awards very beautiful movies. "Best" does not have to be "most complicated", though the Academy often honors very complicated or technical movies. And "best" certainly does not have to be "most realistic"; it's hardly the case that realistic naturalism is what most modern narrative cinematographers are striving for, and nobody objects to matte paintings, forced perspective, model photography, expressionistic color timing, or even an undefined amount of digital visual effects. The fact that "best" is entirely subjective is why the award is bestowed after a voting process. It's the kind of thing that, if impossible to decide objectively, can at least be approximated with a large enough sample size of opinions. As far as the Academy is concerned, the "best" is just what the majority of voters claim to like the most out of the pool of nominees.

So, then, what is cinematography? The answer is too short to be useful: the practice of creating motion pictures. More to the point is exploring what a cinematographer, in the capacity of Director of Photography for a narrative feature film, does.

A cinematographer makes decisions.

A cinematographer must make, or at least contribute to, every single decision about executing the desired look of the movie. The cinematographer's mind is the organizing engine which condenses a myriad of creative, technical and human resources decisions into a cohesive, coherent and aesthetically appropriate look for the movie.

A cinematographer chooses how to use light, contrast, color, camera placement, camera movement, lenses, filtration, focus, framerate and an image capture medium, among other things, to best represent the emotionality, character and story of the movie. A cinematographer marshals a crew of artists and technicians to execute these decisions.

While it's accepted as given that a cinematographer makes these choices during production in the real physical world, every single decision that a cinematographer makes can be, and is, made in the virtual digital world as well.

For every shot in Gravity, Lubezki decided how to use virtual light, lenses (even going as far as simulating the chromatic aberrations and geometric distortions of physical lenses) and filters to tell the story. He chose where to place the virtual camera and how to move it, where to place and how to change the sharp focus of the image, and how to use color and framerate. He chose an appropriate capture medium for the needs of Gravity; the ARRI Alexa's sensor for the live-action photography and a simulated film emulsion for the digitally created bulk of the imagery. He hired (or consulted on hiring) and oversaw the artists and technicians for both the live-action and virtual portions of the shoot.

What, then, is insufficient in Lubezki's work on Gravity that prevents it being honored as cinematography? If, with the Best Cinematography Oscar, we are not honoring the successful execution of the cinematographer's choices then what are we honoring?

Furthermore, where is the line between an acceptable amount of virtual cinematography and an unacceptable amount? The vast majority of major features released in this era have at least some visual effects work, much of it meant to be invisible, and many features have a huge quantity of entirely computer-generated scenes, right down to digital avatars of real actors. For those who suggest dividing the Oscar cinematography award into Physical and Virtual categories, where is the cutoff? 50% of shots? 33% of shots? 75% of shots? Any specific quantification seems absurd, but some kind of choice would be necessary.

I sympathize with the gut emotional response against virtual cinematography. For the artists and technicians of our industry, the transition into digital filmmaking has at many times felt like merely an excuse for less of everything; fewer workers, less time, less money, a less pleasing final image. It makes sense that going another step and leaving behind real sets and real cameras feels like an existential threat.

But, ultimately, the creation of art is about the transmission of thought, emotion, and intent from one mind, or several, to other minds. A slavish devotion to extremely specific, material objects or methods becomes parochial and small, and reduces the achievement of creating art to the mere ability to push bits of matter around.

I choose to honor the mind.

Jurassic Park: 3D IMAX re-release

Hey, folks! It's been a long time. I haven't written about Step Up Revolution or The Life of Pi or Oz the Great and Powerful and at this point I probably won't. I'd expected to write a lot about Step Up Revolution, given that I was responsible for a fair portion of the 3D in that movie, but when I saw it it was (happily) exactly what I'd expected to see from what it all looked like on set, so I wasn't sure what to say other than "hey! We did a really good job!" Since complaining about stuff seems to make for better blogging, here's my experience last night seeing the dimensionalized re-release of Jurassic Park.

I went with several friends (including my fiancee, who'd never seen Jurassic Park) to the Loews Lincoln Square cinema in Manhattan which features the only true "full size" (as in, meant for 15 perf 70mm film projection) IMAX screen in all of New York City, the others being so-called "fauxMAX" standard cinema screens with IMAX-branded digital projection. The main problem with this theater is a side effect of their assigned seating ticketing policy: the only good seats for 3D are in the center of the back row, and these get sold out really quickly. Pretty much every other seat in the theater is either too close to the screen or too far off to the side [see my post below about seeing Prometheus in the same screening room]. My friends and I, sadly, wound up in the center of the fifth row, which is way too close to give a totally fair assessment of the 3D in this movie, but I've decided to just go with it and not to be totally fair.

Problem #1- Dimensionalization as an overall strategy is *still* visually unsatisfying. It's obviously a necessary tool for occasional fix-ups of shots in otherwise natively shot 3D movies, but in the case of Jurassic Park 3D pretty much the whole movie felt like a diorama with unsatisfying depth on faces and most objects even when the roto artists were able to create a wide variety of planes in z-space for objects to live in. I can only hope ILM did a better job on the forthcoming Pacific Rim (the sequel to which, supposedly, Guillermo del Toro wants to shoot natively in stereo). The overall sense of depth *may* have felt better if I'd been sitting in the sweet spot of the theater, but I wasn't. So there.

Problem #2- Long lens compositions. Savvy cinematographers have learned by now how the perspective compression caused by using long lenses destroys any sense of proper depth and scale in stereo images. But this knowledge doesn't travel back in time to DPs shooting 2D movies eventually bound for stereo conversion. As such, so many of Dean Cundey, ASC's fine long-lens framings wound up looking insanely silly when dimensionalized, with characters and objects in the backgrounds not only feeling flat but also being completely out of scale with characters and objects in the foreground. You'd frequently get the feeling that people further away in a long-lens shot were actually *larger* than people in the foreground and it was really unsettling.

Problem #3- Terrible judder and temporal artifacting. Hard to say how much of this is left over from the 2D version (i.e., panning speeds inappropriate for 24fps), how much was caused by IMAX 3D dual projection, and how much worse it may have seemed due to the next problem:

Problem #4- The awful apparent crop from 1.85:1 to either 1.66:1 or 1.44:1 or 1.33:1. This is likely only a problem for screenings on 15/70 IMAX screens, as the "fauxMAX" screenings will probably be 1.85. Again, I was too close to determine the actual aspect ratio but what I *could* tell was that there were a lot of awkward framings. The scene where Hammond meets Grant and Sattler in their trailer was an empty center frame with the characters jammed awkwardly into the edges, the shot from behind Grant which tilts up to see the plesiosaur was really just a shot of Sam Neill's back, the wonderful shot of the T-rex in the visitor center roaring as the banner reading "When Dinosaurs Ruled The Earth" falls past lost all it's drama in a squarer shape, etc. Also, the crop required some pan-and-scanning, which was just horrible in 3D with maybe the worst offender being one of the shots of the helicopter flying a serpentine through the lush mountains. Blech.

In the end I think the re-release of Jurassic Park in dimensionalized 3D is a cautionary tale about the desirability of dipping into a studio's back catalog and dimensionalizing big hits in an attempt to wring some more dollars out of the fans. I, for one, was unimpressed.

Katy Perry: Part of Me 3D

"In 3D, a mo-vie, heard it's beautiful".

First thing's first; a huge amount of this movie is 2D, so if that's likely to disappoint you after paying your cineplex's 3D upcharge then maybe you should just stick to the flat version.

But that would be a shame, because both the native stereo footage (which is mostly reserved for the live performance segments) and the dimensionalized conversions of stills and archival video are really quite lovely. StereoD turned their expert hand to crafting sculptures out of old photographs, making it clear that dimensionalization can certainly shine under the right conditions. While in this case it's perhaps a backhanded compliment to say that the "right conditions" are still photographs, please recall that I've praised their motion work in the past, and if anybody's going to finally crack the nut of motion dimensionalization it will be StereoD. The Thor conversion looked pretty decent, after all.

I've had concerns about juxtaposing a lot of 2D and 3D footage, wondering if it would be too jarring or annoying to the audience, or feel like a lack of added value. I recently acted as stereographer on a short film called (for now) King Theodore and Other Primitive Earthly Pleasures which is a blend of 2D and 3D and motivates its 3D through a specific character's presence in a scene. During shooting I was skeptical about how it may all edit together, but after seeing Katy Perry: Part of Me 3D I am no longer worried. 2D and 3D can live together harmoniously and, as with all things in cinema, buying into a stylistic choice ultimately comes down to motivation. The native 3D in KP:PoM3D is, naturally, motivated by Ms. Perry taking the stage to delight her audience, and it makes perfect sense to separate the dream world of the stage show from the somewhat less dreamy world of life on a major pop tour.

That native stereo was ably handled by a man I've mentioned before here, Matthew Blute, employing RED Epic bodies on 3ality Technica TS-2 and TS-5 rigs using Arri Ultra Prime and Optimo DP zoom lenses (essentially the same package as a major feature to be imminently released of which I have intimate, personal knowledge regarding the stereo capture process... STAY TUNED). I believe Mr. Blute may have been slightly conservative with the stereo but with the chaos of a live shooting scenario I can fully understand the desire to be safer rather than sorrier. There's decent roundness throughout, and certainly no feeling of diorama-like planes, but for certain shots (like an early establishing shot of the arena) I was really wishing they'd gone bigger on the depth. I suppose a more conservative depth style may also have been a choice intended to keep the jumps between 2D and 3D more manageable, but in the end I don't think Blute needed to worry about it so much.

In any event the movie went over really well with the small crowd of young girls and their guardians who happily sang along and waved their arms in all the right places, so mission accomplished in terms of the movie reaching its audience.

Which is totally not me at all, of course, no. Not me at all.

Sports Illustrated Swimsuit 2011: The 3D Experience

Okay, yes, ha ha, the SI Swimsuit cheesecake video, but let's accept that if there's anything 3D should be good at it's capturing the human form. As evidence I submit the vast catalog of statuary from the classical era. While this, uh, "documentary" is ostensibly about the industrial production of pinup photography it serves much better as a tragic parable of the perils of long lenses for stereo imaging.

Long focal-length lenses are the bread and butter of the fashion photography world. Their narrow field of view is perfect for creating graphic backgrounds which nicely isolate subjects, and the perspective compression has a flattering effect on faces.

Unfortunately that same perspective compression is the death of volume. You can see this readily enough in a monoscopic image and especially well in sports photography where focal lengths can get extremely long. Look at that quarterback sixty yards away about to fire off a pass, and then look over his shoulder at the people in the seats another fifty yards behind him and notice how they're almost the same size as he is. Our loss of perspective information, one of our depth cues which uses relative size to judge distance, compresses the world into flat, diorama-like planes of indeterminate spatial separation.

In stereo imaging longer focal length lenses (we're talking, like, over 40mm for a super-35mm image plane) start to squeeze all of the "roundness" out of objects. This is a poor aesthetic choice, in my opinion, for shooting compelling stereo images of voluptuous bodies.

SIS2011:T3DE must have been an extremely frustrating experience for supervising stereographer Matthew Blute. I know he's a very talented guy because I've seen stereo footage from the Russian movie Stalingrad, which is currently in production, and it's spectacular. I have to assume the DP insisted on using long lenses to better match the look of what Sports Illustrated photographer Bjorn Iooss was using on his DSLR for the stills while Blute wrung his hands over how all the models, volcanic rocks, waving foliage and crashing waves were being squashed into cardboard cutouts. There is a painful irony during a filler segment where everyone on screen is talking about the glorious, lush, tactile tropical location and the audience is essentially shown bent postcards at arm's length.

The production apparently used Cameron|Pace rigs loaded with Sony HDCP1 bodies, and here's an instance where maybe the 2/3" format acted as a small bandage on the long focal length problem by contributing greater depth of field. At super-35mm sensor sizes and larger the models probably would have looked like Gilliamesque paper dolls cavorting in front of impressionist paintings.

Prometheus

Yes, transparent space helmets are an excellent subject for stereo cinema.

Prometheus is, as I'd hoped, an exemplar of the Second Wave of the recent 3D revival. During the first wave, let's say "mistakes were made". Few people hired to make stereo features had much of an idea of how to do it well and the learning curve is steep. Eyestrain, amateurish stereo problems and weak creative decision-making were all too common hallmarks of the 3D movies which rolled out in the Avatar era (not to mention the ongoing horrors of exhibition) and I've been crossing my fingers that the 3D movies of summer 2012 would start showing the public how good 3D can really be.

The stereo work in Prometheus is beautiful and more restrained than I expected. There's gorgeous composition and camera movement throughout and none of it feels like it's "working around" the 3D. All of the CG elements are, naturally, beautifully done and if there were many geometry problems with the photographed material as shot they were well fixed by skillful application of Mistika in post.

Stereographer James Goldman appears to have done really well at getting good depth with judicious amounts of deviation. Obviously this is assisted by being able in many cases to multi-rig the CG backgrounds but it's no small thing to capture live, native stereo and have it blend so beautifully with composited backgrounds. But regardless, the entire image consistently had a nice roundness and separation to it and there were only a handful of times when I wished for more dimensionality in a closeup.

Praise is also due to the other members of the stereo team. Adam Sculthorp's aerial stereo work is lovely and Damien Fagnou clearly held up some exacting standards for the 3D while supervising the CG work at MPC.

I saw the movie at a full-sized IMAX screening with the image occupying an area wider than my fixed field of view and I'm curious how much reconvergence was done for the IMAX digital master. There was very little (none, really) "gag" use of negative screen space, though negative space was nicely employed on facial closeups and in other subtle ways which maximize the depth window.

The story certainly suffers a bit from writer Damon Lindelof retreading some of the perhaps overearnest explorations of faith which brought LOST to such a clunky close, and there's some artless hammering home of the theme involving a cross-shaped necklace. But aside from some momentary chagrin, none of these things (or some other weak story points) seriously impacted my enjoyment of the movie. If you're looking for an entertaining space monster flick with beautiful 3D I recommend Prometheus.

Quest For The Best 3D in NYC

I'm accepting nominations for the best 3D screening rooms in NYC, and intend to test the matter thoroughly now that the big native 3D movies of the year, like Prometheus, are starting to roll out to theaters.

Any thoughts? Respond on here, or you can get @ me on Twitter @QXZ.

Christopher Nolan Critiques 3D

Christopher Nolan is a filmmaker who I respect and greatly admire, and it's nice to see that despite choosing not to make his movies in stereo he has a considered and thoughtful opinion on the subject. The following is excerpted from an interview he gave with DGA Quarterly, the Directors Guild of America's magazine:

I find stereoscopic imaging too small scale and intimate in its effect. 3-D is a misnomer. Films are 3-D. The whole point of photography is that it’s three-dimensional. The thing with stereoscopic imaging is it gives each audience member an individual perspective. It’s well suited to video games and other immersive technologies, but if you're looking for an audience experience, stereoscopic is hard to embrace. I prefer the big canvas, looking up at an enormous screen and at an image that feels larger than life. When you treat that stereoscopically, and we've tried a lot of tests, you shrink the size so the image becomes a much smaller window in front of you. So the effect of it, and the relationship of the image to the audience, has to be very carefully considered. And I feel that in the initial wave to embrace it, that wasn’t considered in the slightest.

I'm interested by his comment about S3D providing each audience member with an individual perspective as opposed to the mass perspective provided by monoscopic films. I suppose he's technically right about this, as different viewing positions will certainly provide different experiences of the stereo image. It's also intriguing that Nolan feels S3D can be too intimate, while most proponents of S3D tout its increased intimacy as a positive.

However, his attitude about S3D being inappropriate for creating images which are "larger than life" strikes me a slightly odd and I wonder about these tests which made him feel that S3D shrinks his canvas. Did he discover that appropriate deviation parameters for very large screens limited his ability to capture a feeling of depth appropriate to his framing? It seems unlikely; deviation does not equal depth, as we know, though perhaps Nolan is unwilling to embrace using wider lenses.

Nolan has more than earned his right to fulfill his vision in any way he chooses. While it's disappointing that he hasn't found S3D to his liking I'm gratified that he clearly gave it thoughtful attention rather than handwaving it away by spouting anti-S3D talking points in the manner of many of his peers.

The guys at Stereoscopic Filmmaker have recently produced a 3D workout video starring Brittney Palmer, UFC "fight girl" and Playboy model, which was shot on the Panasonic HDC-Z10000 stereo camcorder.

(red/cyan anaglyph glasses required)

While "Z10000" is probably the most absurd product name going (where do you go next? The Z eleventyzillion?), in my messing-about with it I found it relatively easy to use and decently versatile. Its small sensor can be helpful if you're one of those who likes a lot of depth of field in their 3D, but on the downside it makes everything look a little homespun. For example, the Palmer workout video looks at times unnecessarily "porny" due both to the small-chip look and the off-kilter camerawork, among, uh, other things.

That said, I was somewhat impressed with the shooter's ability to move quickly in from a wide shot to a closeup without the overall deviation going too totally nuts. There are some questionable depth placement choices, but if nothing else this video exemplifies what the Z10000 will do best: open up the world of quick and easy 3D.

Game, Set, Match!: The Art And Science Of Lenses for 3D Capture

pg. 57

For rig technician and additional stereographer on Step Up 4 Keith Putnam, the choice came down to RED Epic on 3ality TS-5 rigs using ARRI Ultra Primes plus a few Angenieux Optimo DP zooms (16-42 and 30-80). Putnam says his first consideration was finding lenses that had front diameters the right size to fit into a TS-5 mirror box. "We found that the Ultras are small and lightweight, and, advantageously, they also are all the same barrel length from 16 through 100 millimeters, which makes changing lenses much faster because you can set a camera-to-mirror distance once and leave a body stopper at that distance," Putnam explains.

"It's not enough to simply choose Ultras with close serial numbers and matching production letters," Putnam continues. "We had to put every given focal length pair up on the rig [16 possible pairings for each focal length with the primes from four source sets] and visually assess them with the 3ality SIP."

The Optimos were much more difficult to match he continues. Being zoom lenses, the barrel distortions change throughout the range. "They have what I would describe as a 'deviation drift' from the center of the image to the edges." To further complicate matters, Putnam had to assign Optimo pairs to specific rigs since some lenses would match better on some pairs of Epic bodies than others. 

Question: Flares?

Joe asks:

Saw John Carter in 3D last night.. there were several shots with lens flare. Some was deliberate and possibly added in post - shooting directly into the "sun", etc. But a few others seemed real, incidental and true... Are flares possible in 3D and how does that work?

In general, real lens flares are a huge problem in shooting 3D. Before any direct light hits a lens it either reflects off of or passes through a beamsplitter (assuming you're not shooting side-by-side). Direct light hitting a beamsplitter tends to bounce around in the mirror box all kind of ways and produce ugly smears, so you rarely get a nice aesthetic lens flare and if you do it'll usually be accompanied by a bunch of crap. Plus you wind up with bad rivalries where there are size, brightness, color and polarization differences between the flares as recorded by the two cameras. Flares can contribute to an uncomfortable or distracting viewing experience. People try really hard to keep any direct light off of the beamsplitter whenever they can. Lens flares in side-by-side rigs can be a problem too, unless the flares in each eye just happen to play well with one another. John Carter has the advantage in the flare arena of having been shot in 2D and post-converted into 3D. In that case "all" you have to do is get your roto artist to exactly duplicate the lens flare from the real image, and then you can position it at will to make it watchable. Or, as you mention, just add straight up VFX flares into shots where there was no real flare.

SOPA/PIPA

While this is a slight swerve off topic, the Stop Online Piracy Act and the Protect Intellectual Property Act are huge issues in our industry. Unfortunately the union I belong to, IATSE (Local 600, the Cinematographers Guild), has taken a public and vocal stance in favor of SOPA/PIPA under the guise of "protecting jobs". While protecting our jobs is, indeed, the nominal purpose of our union, IATSE has in this instance allowed itself to be drafted into the service of protecting the massive profitability of giant corporate conglomerates for which entertainment is but one small piece of their empire. There are far more achievable and relevant ways to promote craft jobs within the industry than championing a futile and draconian legislative assault on "piracy".

SOPA/PIPA as written are absurdly overreaching and prone to serious abuse. Reducing piracy to zero is impossible and enacting dangerous, overly broad and potentially unconstitutional laws in a misguided attempt to eradicate piracy will only result in failure and putting a powerful weapon in the hands of people I don't trust to wield it justly.

As a professional in the media industry I believe that if we can't inspire audiences to want to pay for entertainment, if we can't create experiences which people find valuable, then either we've failed or we're irrelevant. Show business is about evoking wonder and joy and excitement, not further eviscerating the 4th Amendment to prop up corporations which can't survive now that the digital box they built for Pandora has been opened. Oppose SOPA/PIPA and focus on creating irresistible art and entertainment.

Hugo

I am way overdue in talking about the use of stereo in Hugo, I suppose, but here's the thing: it was fine.

Mostly fine. There's excessively extreme negative parallax stuff happening all over the place, as a joke, but otherwise it's competent (that dog's face diverges, it just does, and that was easily avoidable [ditto: Sacha Baron Cohen's nose]). Also, the rapid lateral camera moves aren't helping anything. Try out the Tessive Time Filter next time you guys want to do fast lateral camera moves in 3D.

What amazes me is all this press about how incredible the 3D is in Hugo, like it's some kind of groundbreaking festival of creativity and wonder when in reality what people probably think is "Martin Scorsese and Robert Richardson are both awesome filmmakers so clearly what I am seeing is also awesome." And yes, Richardson shot a very beautiful movie and Scorsese worked his usual... okay, who am I kidding, no one would have any idea Scorsese made this movie if they weren't told because it bears essentially zero resemblance to anything he's ever made before in any way. That isn't meant as a slam, because there's nothing wrong with doing something different, but I feel like Scorsese's only real mark on this movie was making sure there was a story-halting, overly long digression into how wonderful he feels George Méliès' films were.

By the way, respect to the stereo conversion of the archival silent film footage. I was actually quite impressed. Good job, Legend3D.

Anyway, what the press seems to be reacting to in re the stereo is that the stereo is, again, fine. As in competent. Which until recently has been a sad rarity. I see unfortunate parallels to the critical reception of Avatar, though in that case what the press overlooked because of the stature of the filmmaker was that lots of the live action has painful and dumb stereo problems and also that the movie is just outrageously boring on every level. Hugo is not boring (mostly) and the stereo is, again, generally very competent, but the critical reception just smacks loudly of the cult of personality rather than honest reactions to what's on the screen.

What I can praise Hugo for is being a pretty okay movie that a lot of people wanted to see which is also unlikely to be an uncomfortable 3D experience and, therefore, might do good things for the public's feelings about stereo cinema.

Meet The CSCSwPCL

A continent-spanning group of researchers from MIT, Disney Research Zurich, UNC and other places is developing a programmable stereo camera rig with some pretty impressive potential. It goes by the totally marketable name of Computational Stereo Camera System with Programmable Control Loop.

The two features which caught my attention were the slick touchscreen interface (of course) and the sophisticated real time tracking of subjects in motion within the frame. The touchscreen allows an operator to tap an area of the image to instruct the rig to converge at that point, focus at that point, or do both. The rig can also track a moving feature, holding convergence and focus on that feature while adjusting IO to maintain preset disparity parameters.

That latter point is really key: it doesn't do much good to let a convergence tracker run wild if it's just going to create insane, unwatchable disparities. Human depth pullers do the same thing, adjusting IO while reconverging to maintain a chosen deviation, but humans have to do it with their hands like it's some kind of baby's toy.

In the demo video you can see that the convergence and IO changes feel a little, well, mechanical, but the group has and is working on various data filtration schemes to keep the adjustments feeling a little more human.

I can and do take issue with the aesthetics of locking convergence onto a subject moving from the background to the foreground (or vice versa), because it's weird to me to see the whole stereo window sliding back and forth in service of keeping the subject at screen plane, but as we all know: guns don't kill people, people kill people (with bad stereo). The potential, in good hands, is obvious.

Given that Disney is involved I imagine the people really champing at the bit to implement this system and, ahem, Reduce Labor Costs would be ESPN.

Cool stuff, folks. Use these powers for good.

A Very Harold & Kumar 3D Christmas

First off, this movie is exactly what you'd expect if you like Harold & Kumar movies so if that's not your thing (drugs, boobs, broad humor, Neil Patrick Harris) I'd say you could skip it.

AVH&K3DC is chock full of silly stereo gags and most of them work fairly decently. And while the negative parallaxes on the Stuff Comin' Atcha were within range to be and stay fuseable themselves (no 'splitting'), it was fairly common in those shots for the total amount of onscreen parallax to exceed what my eyes could comfortably fuse at cinema screen sizes. Perhaps choosing to converge even deeper behind the gag to close up the background deviation could have made the shots easier to watch overall AND exaggerated the Comin' Atcha effects even more. As semi-extreme as the gags were there was still a minor failure to get a feeling of objects really extending out into the audience.

Related to that, a fair amount of the movie felt pretty flat. The depth planes were there, and it stopped short of "diorama effect", but many shots lacked roundness. I'd like to blame focal length choice here but I'll need to see the movie again to get a better handle on what, exactly, seems off. Much of the movie seemed placed unnecessarily far behind the screen, too, without foreground objects, and that contributed to the flatness. Maybe this is an artifact of doing reconvergences in post and being too conservative with the depth window? So, not the best use of the stereo space in AVH&K3DC but certainly within the bounds of acceptability. Director Todd Strauss-Schulson (Private High Musical, Sorority Pillow Fight) didn't stray too far from standard coverage found in monoscopic movies, so perhaps what I found to be excessive behind-the-screenness was insurance against aggravating edge problems when doing over the shoulder shot/reverse shot setups.

Stereographer Paul Taylor's work was generally competent and I didn't feel too many straight-up stereo errors. Edge violations were handled decently, though mostly without floating windows (good use of darkness). I noticed focus mismatches a couple of times, primarily via differently sized bokeh in out-of-focus point sources of light (like Christmas lights in the background), a few instances of luminance mismatch in hot highlights and/or reflection polarization problems, and chroma problems between eyes in deep shadows.

AVH&K3DC was shot with 3T Quasar rigs loaded, oddly enough, with Panavision Genesis bodies. I'm not sure if DP Michael Barrett (Zookeeper, Takers) was uncomfortable with more, uh, current digital systems or if he just likes the look, but it seems strange to me to shoot a stereo feature film on such a low-resolution camera system.

Man, I'm hungry.

Cameron|Pace Throws Stereographers Under The Bus

James Cameron on the business of 3D from Lars Steenhoff on Vimeo.

James Cameron and Vincent Pace, appearing in their combined form as the Cameron|Pace Group, addressed the recent IBC about the business side of Three Dee. Cameron described his vision of C|PG's future and that of stereo content (which are one and the same in his mind):

The next step is to industrialize it, is to take it to scale, is to bring the cost down through volume, and to put it in the hands of global content providers, to fill in that content gap.

Industrial strength content fill for your gap, coming to a screen near you.

Naturally one can't fault a healthy profit-motivated business venture from attempting to both grow its market and seize the greatest portion thereof.  C|PG evangelizing about 3D is good for 3D, respecting both the gap and the content needed to fill it. Much of their work is, laudably, about developing better tools to make better stereo.

The unfortunate part of C|PG's pitch, specifically Cameron's pitch, is the choice to demonize The Stereographer. The Stereographer is repeatedly cast as an embodiment of limitation, or pejoratively styled a 'guru', to illustrate how C|PG's glorious cost-cutting technology can free content providers from additions to the payroll:

I think it's been a problem with a lot of 3D practicioners... they say to newcomers to 3D 'oh, well, you can't do that and you can't do this and 3D is very special, it's very exotic, it's very fragile. If you don't do it right it'll look bad, it'll hurt people's eyes. You've gotta listen to what we say.' Well, what is that person doing? What is that <scare quotes>stereographer or I call them the gurus, y'know, and I always imagine them in kinda white lab coats, standing there, you konw, pulling their chins, telling you what you can't do. Hitting you, sort of, with a stick if you have an edge violation or something like that. I think we've all experienced this... you know your jobs, you've been doing them for years, and who are these 3D upstarts to come along and tell you how to do it?.. We don't want to elevate 3D trained people into those positions. We want people who are doing those jobs in 2D to just learn what they need to know to do good 3D just as part of their portfolio.

Cameron gently accuses himself and Pace of being 'probably guilty in the early days' of playing Baghwan Sri Stereo, but the implication is that C|PG's technology has obsoleted this oppressive specialist in creative stereo imaging. It's unsurprising that Cameron has a gut-level emotional aversion to anyone telling him what to do or how to do it, but it seems a little too disingenuous to push forward the stereographer as the fall guy for why good stereo has yet to become a turnkey operation.

Pace gave an interview to HD Magazine around the same time as C|PG's address at IBC in which he expressed his confidence that his technology could eliminate the need for several crew positions.

Stereographers and convergence pullers are going to be part of the history of 3D, the status right now is that they’re thinning out, the technology’s getting smarter, it’s performing better and that’s thinning out the number of people that we need and it’s going to go to zero! It’s just inevitable that the technology evolves to handle that part of the business.

Ah yes, I recall when the advent of autofocus killed all the focus pulling jobs out there. Local 600 eliminated 1st Assistant Camera from their rolls entirely.

Cameron continues:

The second you have a 3D Expert, a stereographer, standing there telling them what they can and can't do you're going to set up barriers... This is a guru trying to make themselves important in the mix. Trying to tell you how everything that you've done before you can't do, now you have to have different horses for courses, different 3D for different sizes. Well it's all bull! You don't. Avatar plays fine on an iPad, it plays fine on an IMAX screen.

Avatar doesn't really play "fine" on anything, given all the egregious stereo errors throughout that movie which have caused so many people I know to find watching it a painful experience. But Cameron has sales figures to point to which tell him Avatar is Good Enough.

It's hard to escape C|PG's implication that to industrialize stereo content creation, in order to fill that gap, 3D merely has to be 'good enough' or not technically unwatchable garbage. This is likely, and sadly, true. But it's one thing to have a computer comparing total onscreen parallax with the distance to the nearest object and finding an IO which keeps everything watchable and another thing for a human mind to play the stereo like an instrument, serving the story from moment to moment, creatively and consciously.

While technology is good at coloring inside the lines it tends to fail to pick the best crayon.

Ocula 3.0

The Foundry has released the third version of their stereo finishing Nuke plugins, Ocula, to the masses, and it contains some pretty powerful tools as demonstrated in this video clip:

OCULA 3.0: New Features from Hieronymus Foundry on Vimeo.

I'm personally most impressed by Ocula 3's ability to mitigate focus mismatches between eyes and its retiming (speeding up or slowing down) tool which takes data from both eyes simultaneously in order to avoid the usual horrors introduced by retiming the eyes separately. The color matching seems a bit less powerful than that found in Mistika, but Ocula's occlusion detection system (and its use of occlusion data, especially with color matching) is pretty sweet. And, of course, it's totally integrated into Nuke which makes for happy stereo compositing.

Also, it brings a huge smile to my face that footage from Avatar is featured heavily as source material for Ocula to correct. Avatar is hilariously full of terrible stereo mistakes, despite its place in history as "the movie that revived 3D". One hopes they'll nail it for Avatar 2 and 3.

LG D2342 23" 3D 1080p Display

There is a reason inexpensive things are inexpensive: they're cheaply made. And sometimes it shows.

The LG D2342 is marketed as a computer monitor rather than a TV and can be picked up from Best Buy right now for the pittance of $265 (plus tax and shipping). I bought it for the same reason you might: it's an easy way in to a 3D monitoring solution. And my criticisms below are not to suggest the display isn't useful. We had two of them mounted on the director's 3D video village cart on a recent stereo feature and they were serviceable, though visibly inferior to their 47" brethren over at the stereographer's tent.

The primary and most serious problem with the smallest member of LG's Cinema 3D line is that it's effectively impossible to find a single viewing position which will allow the image to fuse completely once any deviation approaches 1%. This is a big shame, as 1% is a fairly common target for maximum background deviation (and should be more common) while negative parallax deviations are frequently seen in the 2-3% range and beyond. What this all means is that in essentially every piece of 3D content you can display on this monitor you'll be seeing ghosted, unfused images all the time. And the problem gets massively worse if there's a lot of contrast between depth planes, like a bright middle-ground object against a dark background. I haven't measured the contrast ratio which triggers ghosting (through LG's supplied RealD-spec passive lenses), but it's not terribly high. You'll spend most of your time trying to hold as still as possible with your eyes centered to the screen, but you'll still need to move higher or lower at times to make images fuse better, especially ones close to the top or bottom of the frame.

I have to assume this is the result of slapdash manufacturing and loose QC specifically with respect to the positioning of the alternating polarizing filter LG has employed to achieve their passive stereo ends. The bands of alternating polarization are probably just slightly misaligned from the rows of pixels they're meant to polarize. This is doubly clear because it's possible to see a nicely fused image at the top or bottom of the screen while the same amount of deviation on the opposite edge of the frame won't fuse well at all. I never encountered this issue on the 47" version of the monitor, so one imagines the QC goes up along with the price.

3D being as dependent as it is upon field of view, making a small monitor is a challenge from the standpoint of optimal viewing distance. LG recommends a minimum distance of 90cm (a little less than 3 feet). This does seem to be a decent semi-sweet spot. When viewed closer than that the stereo loses a lot of roundness. The small size of the monitor makes the stereo effect become extremely exaggerated starting at about seven or eight feet. This display really is better suited to desktop viewing rather than couch viewing, hence its marketing as a computer monitor. Unfortunately, there is no distance which solves the poor fusing problem.

On the bright side, the color rendition is very nice and the, uh, brightness is, uh, bright. A monoscopic 1080p image displays beautifully. The monitor supports Side-By-Side, Top/Bottom and Line Interleaved stereo content, but not Frame Packing; the latter is left to LG's higher end displays. You can flip the eyes, if the incoming content is pseudostereo. It's light, small and inexpensive and thus good for portability.

If your wallet is light, or if you need an "expendable" display for field use, by all means try out the LG D2342. Just understand that it's inexpensive because it's also cheap, in all senses of the word.

I tested this monitor mostly by watching Megamind, which features some really lovely stereo work. Thumbs up to Phil 'Captain 3D' McNally and his team, again.