To be done well, a 3D film really needs to be 3D from the outset, as the tools and techniques that work in three dimensions are very different to those most cinematographers have grown used to. In fact, some even require a step back in time to a more artistic age of cinema, as Buzz Hays explained.

Depth of field

Take depth of field, for example. It’s a staple of modern films, commonly used to direct the viewer’s eyes to the key point of the screen, guiding us through the scene as the director intended. Yet it’s an effect that simply doesn’t work in 3D. To demonstrate, Buzz showed us a 3D still from the movie Monster House, in which a character stands with his arm outstretched towards the camera. Only his face is in focus.

“Note where your eyes take you when this shot hits the screen,” began Buzz. “Cinematographers are trained in the art of this particular technique, so they know exactly where to look, but I’m going to hazard a guess that you looked at his hand first?” He was right. Despite the lack of focus on anything but the face, the hand drew the attention.

“Typically, when you look at 3D, whether it’s the real world or a movie, you tend to look at whatever’s closest to you. In 2D depth of field works well to direct your eye to the main character’s face, but in 3D it creates a disconnect, because there’s no logical reason why the world is out of focus. It creates visual confusion, and this could be one of those moments where the viewer is pulled out of the story because he’s not quite sure where to look.”

He then replaced the image with the same shot, all in focus. “Now your eyes will still go to his hand first, but as you wander the frame [and it’s in focus] it takes the curse off.” And it did.

Lighting

With depth of field less effective, cinematographers need to use other methods of directing the viewer’s eye. For inspiration they should look back through the decades, to a time when lighting was more prominent tool in a cinematographer’s box.

“When you think about some of the world’s greatest cinematographers of yore, like Greg Toland who shot Citizen Kane for Orson Welles, they used very deep-focus photography. Every aspect of the image was in focus, so instead they used lighting to direct the eye, and shadow to sculpt objects. [There's a fascinating piece by Toland here.] We’re starting to find with filmmakers working in 3D, that it is like theatre. In theatre we can’t simulate out of focus, so we use lighting to direct the eye – and we’re starting to lean back towards that method.”

Games designers have long used lighting to guide the player through increasingly complex levels, partly because depth of field effects have only recently become feasible using consumer graphics cards. Filmmakers will also have to relearn these more traditional techniques, which is something Buzz is all for. The problem is that many in the industry today have become tied to depth of field as a tool to the point where they don’t even consider other methods.

“A lot of the younger filmmakers I’ve talked to say, ‘I can’t lose my use of depth of field because that’s how I direct the eye in the frame’. It’s because it’s the only way they know how to direct the eye. A lot of them have never tried anything else because it [depth of field] has been around since they started in cinema.”

As an interesting aside, Alfred Hitchcock was no stranger to 3D, having filmed the 1954 classic Dial M For Murder using a stereoscopic camera rig. For one scene he wanted a close-up of a finger dialling the titular M on a rotary dial telephone, but the gigantic rig just couldn’t get close enough to capture the effect at such a small scale. Hitchcock’s solution? To build a huge wooden telephone (right) and a wooden finger to dial it with. By the time the film was released it played mostly in 2D, as the short 1950s 3D craze had already fizzled out.

Editing

A move towards the methods of theatre should have another side benefit for viewers tiring of the ever-faster editing that seems to be plaguing modern cinema.

“You tend to involve an audience in a much deeper way if the performance just unfolds. If you have a really powerful, great actor giving a performance, when you watch that in 3D, especially from a given camera position, you feel that performance coming through as if you’re sitting right there in a theatre. As we start to edit around that, we start to lose the sense of the power of the performance.”

These days many directors seem to assume the viewer has a short attention span and requires fast editing to retain the sense of involvement in a scene, but that shouldn’t be the case with 3D.

“A number of visual scientists that have looked at 3D over the past ten to fifteen years have commented on this idea that when you look at a still 3D image, your brain tends to do the editing for you.”

“Working in filmmaking we use editing as a principal technique for driving the story forward, to ellipse time and all sorts of things. But presented with a 3D image your eye tends to wander. A lot of filmmakers will find that as they introduce new environments and characters, they might actually open up the pacing of the edit just a little bit to let the eye wander before they cut to the action. Just like in comedy you don’t want to step on the laugh, here you want to give the viewer the chance to acclimatise to the environment before you jump in with the action.”

As an example of filmmakers experimenting with this idea, we can look back to two 3D movies which used different approaches.

“G-Force was about 84 minutes, I believe, and we had 1,800 shots in the movie. Yet Beowulf, which was over two hours long, had only 840 shots. [Director] Bob Zemeckis realised that using camera motion helped tell the story rather than lots of cutting. It was more conducive to telling the story in 3D, and letting the action play out on camera versus creating a performance in edit.”

3D to enhance story

Zemeckis also used other ways to experiment with 3D, not all of which you may have noticed while watching the film.

“We had a situation in Beowulf, where we had a discussion with Bob about using 3D from a storytelling perspective. Because the film involves characters in constant shifting power, Bob made it so as people were losing their power, they’d actually start to lose some of their dimension. Conversely those who were gaining power started to become more realistic in dimension. It was a very subtle use of 3D in terms of telling the story.”

Too subtle for most to notice, but an interesting approach nonetheless, and similar to one which Hitchcock has been known to use. Of Dial M For Murder, the director once told an interviewer, “We did an interesting colour experiment with Grace Kelly’s clothing. I dressed her in very gay and bright colors at the beginning of the picture, and as the plot thickened, her clothes became gradually more somber.”

The old classics had directors looking for ways to add to the story of a film, and hopefully 3D will have them experimenting all over again.

“To us 3D technology should support the storytelling, it shouldn’t be the story,” said Buzz. As another studio slaps a headline-grabbing 3D conversion on another perfectly good 2D film, few could argue with those words.

Storm Trooper image courtesy of pasukaru76.

As senior producer of 3D Stereoscopic Feature Films for Sony Pictures Imageworks for more than five years, Buzz Hays has been there for the peaks and troughs involved in getting 3D cinema into the mainstream. Over the course of an afternoon he led us through many aspects of 3D, but for this blog I’ve collated his experiences of working on various films into one timeline. It begins a mere six years ago on a train, and ends with the revelation that Avatar was, in some respects, a disappointment…

2004 – The Polar Express

“The Polar Express was our very first project [at Imageworks] in 3D, with Robert Zemeckis directing. Only six years ago there were very few public venues that could show 3D – in this case I believe the number was about 84 theatres in the entire world. That’s a fairly limited release for a picture like that, but it proved to be quite successful for a number of reasons, not least of which was that Jeffrey Katzenburg credits Polar Express as the film where he really came to the conclusion that 3D was a logical extension of cinematography and wanted to take Dreamworks Animation into that.”

2006 – Monster House, Open Season

“The first digital cinema release in 3D was a Disney film called Chicken Little, and they had a whopping 86 screens, something like that. By the time we got to Monster House digital cinema had taken hold, and we had 172 theatres that could show 3D, still not a vast number but the studio felt it was enough to actually make the effort worthwhile.

“In the same year we actually released Open Season as a 3D film – interestingly, the film was actually in the can ready to be released, at which time they decided it should be a 3D movie. So we had to go back into the archives, get all the assets out and then recreate the entire film in 3D in a three-month schedule. It was built in a computer, it’s a totally CG movie, so it was a relatively straightforward exercise – it’s what we call a native version, it’s not a conversion. And it was overlapping Monster House, so we were actually working on two 3D films at the same time. Monster House proved to be quite successful. It was made for IMAX, which had about 90 screens worldwide.”

2007 – Beowulf

“Up to this point, it was mostly family films using 3D, and mostly 88 minutes or under. Beowulf was the first picture for an older audience that was nearly two hours long. We had concerns because conventional wisdom at the time was that children had an easier time looking at 3D that might perhaps be uncomfortable to look at, and adults had a harder time with it. That’s when we really started looking into the physiology side of it for the first time. How do we see 3D at all? Why does it work sometimes and sometimes it doesn’t?

“We had a director in this case, Bob Zemeckis, who wanted a really deep film, but he wanted a story told in three dimensions. Yet the two-hour movie was a big concern for us. We had a ten-month schedule to fine-tune things, and it was only after Beowulf that we realised that over the course of those last three years, we’d worked with hundreds of visual effects artists who’d worked with 3D all day every day for a year or two at a time, and we didn’t have a single health problem.

“A mere year after Open Season, when Beowulf released we had 1,100 screens, so suddenly it was a realistic number to get a lot of people to see the picture. That’s when the floodgates really started to open in terms of 3D production. We had a condundrum, though: in 2007 there were very few 3D releases, so we had no opportunity to show a 3D trailer for the film; all we could do was tell people it was 3D.”

2009 – Jonas Brothers: The 3D Concert Experience, G-Force

“Jonas Brothers was our first live-action experience, and proved to be a very interesting one – not least because my ears were ringing for about six weeks afterwards because the shoot was just girls screaming for ten hours. It was unbelievable. Working with the live footage – and this might surprise you – we thought, wouldn’t it be great if the cameras and images actually lined up? Turns out they didn’t very well at all.

“Technology’s getting better, but these are hard situations to shoot. Basically when we’re shooting with the left and right-eye camera the differences between the two were so vast in some cases it looked like the two eyes were shot on two different days, so the amount of work we had to do in post was frightening. There’s a current filming philosophy to just get it close and fix it in post – which they don’t have to do, we do. We realised then how important it is to start working with camera rig manufacturers to make the camera systems better, so that the post-production process is easier.

“With G-Force, they’d experimented with the idea of doing a 3D movie, then decided not to for a variety of reasons, and mid-way through the shoot – on 35mm film in 2D – they decided, yes indeed, we are making it a 3D movie. So this was our first foray into this notion of converting 2D into 3D. Basically no one had done it – there had been The Nightmare Before Christmas which was stylised with no people in it, and a few IMAX movies – but this was the first time we’d really been confronted with how to convert material.

“We had the luxury of only having to worry about the background photographic plates they’d shot of people and buildings because the main characters in the movie are generated by computer. So, we came up with the tools and techniques to derive depth out of the photographic elements and put them into a 3D environment, in this case Maya, then added the CG characters and rendered the whole thing in 3D. So it was really a hybrid approach.”

2010 – Alice in Wonderland, Avatar

“Alice in Wonderland was similar, although a 180-degree opposite approach: the actors had been photographed against a green screen, and the environments were created in 3D. Despite the various press you’ve been reading about the pictures that have recently been converted, it actually can be done very well, it’s just time-consuming and very expensive.

“As for theatres, things today are both better and worse. Think back to 2007, around Beowulf we had the luxury of a release window for as long as the film needed – six or eight weeks was a reasonable amount of time for a 3D picture to stay in cinemas. Flash forward to now and Alice in Wonderland, the release window had collapsed to just shy of two weeks because of the number of 3D releases and the fact that the theatres hadn’t kept up with the number of installations they had committed to. Avatar was supposed to be the big pivotal film where 3D would explode and we’d have a zillion theatres, but as it turned out it barely had 4,500 or so, maybe 5,000. We were expecting more than 6,000.

“We’re getting past the point where mass audiences are seeing this as a novelty, and it’s hopefully just becoming another way people watch films. So we’re starting to move into a territory that’s not just the blockbusters, not just the animated films, but getting back to traditional storytelling. Over the next year or two we’re going to see some very interesting motion pictures and television programmes in 3D.”

3D is an ever-evolving process, which is why the effect can be such a hit-and-miss affair. But those who insist 3D glasses give them headaches are a little wide of the mark, according to the man who trains the filmmaking pros.

“It’s not the technology’s fault, it’s really the content that can cause these problems,” explains Buzz Hays. “The more care taken when making the content, the better off everyone’s going to be. My mantra is that it’s easy to make 3D but it’s hard to make it good – and by ‘good’ I mean taking care to make sure that this isn’t going to cause eyestrain.”

There are several common mistakes that can cause discomfort, and easy ways for that to be reduced, yet they’re only just being learned and put into regular use.

Interaxial distance

The interaxial, or the distance between the two cameras, controls the overall depth of the 3D effect. Objects will appear closer or further away but they won’t change in size, so it’s important not to increase the interaxial distance too much. Filmmakers are gradually gaining experience with what types of scene work with different depths of 3D, and Buzz was keen to point out that framing a scene for 3D has similarities to composition for still photographers.

“When it comes to composing in 3D… by using the heads of the audience [in a U2 concert clip] or the ground plane, or some continuous sense of depth in the shot, it holds the shot together. One of the complaints people sometimes have about 3D is that it feels like a cardboard cutout: that there’s a cardboard cutout, then some space and then another cardboard cutout. By using a careful choice of interaxial spacing, and also by having something in the frame like the ground plane, or smoke or atmosphere or something, then you can start to hold the shot together.”

Convergence

Our eyes converge inward as we look at an object moving towards us. In 3D it’s essentially the same thing: we converge (or “toe-in”) the angle of the left and right cameras, and this alters the particular 3D plane to which our focus is drawn. Objects in front of the convergence point appear to be coming out at us, while objects behind do the opposite. Care needs to be taken, however, particularly when fast cutting is used.

“There’s a situation where every time we cut to a new shot, the subject of interest is at a slightly different distance from us,” explained Buzz, demonstrating a rapidly cut clip of two people at different convergence points. “What’s happening is on every single cut, your eyes are making an adjustment to depth – you’re trying to find that object. It’s a very subtle distance, it’s not a great distance, but that’s what you’re feeling in your eye muscles as you’re trying to work to catch up with the shot. That’s called the vergence-accommodation conflict.”

“The way we make it much easier to look at is by using convergence in post-production. In that same sequence I adjust the convergence in post [production] to massage the depth. Now your eyes are making the adjustment once in the very first shot, and from that point on they don’t have to adjust again. It’s very subtle but if you don’t do it, it’s the difference between a comfortable experience and a splitting headache after 90 minutes.”

What filmmakers are now learning is that trying to control the convergence during filmmaking is, as Buzz bluntly puts it, “a waste of time”. As cuts are made and scenes are shifted around, it’s difficult to know exactly what shot will follow another, so trying to predict it all is futile.

“It’s far better to find the comfortable place to put the convergence level during shooting, then adjust it in post-production once the edit is finished – that ultimately makes the difference between good and bad 3D.”

Divergence

The opposite of convergence is divergence, and just as our eyes can only converge to a certain point before we go cross-eyed, so they can only diverge to parallel. Overuse of divergence can cause big problems.

“Typically, when we look at an object in the world our eyes are either parallel if it’s at distance, or they’re converged inwards for objects that are closer,” continued Buzz. “There’s a condition that can be created unintentionally where your eyes are forced to rotate outward in order to fuse this image – which frankly only works if you’re a horse or a goldfish, and they don’t buy movie tickets.”

At this point Buzz put a scene on the screen in front of us and had us don our specs. A figure at the back of the image was simply impossible to bring into focus, and even trying was as uncomfortable as you’d expect. Removing the glasses showed why: the left and right views of the figure were several feet apart on the big screen.

“Divergence occurs based on the size of screen you’re using. You might make a neat adjustment [during filming] so it looks great on a monitor, but when you scale it up to 40ft it hurts like heck. Experienced stereographers will be able to avoid it, but some low-budget 3D films have been filled with divergence, as they’ve made the cardinal mistake of falling in love with the image on a video monitor when it was really intended for a cinema display. They’re dialling the depth to within an inch of its life and getting everything they wanted on the small monitor, so their camera settings are out of whack. It can’t be fixed in post – unless you just abandon [the image for] one eye and convert 3D from the other.”

These were just a few of the common faults covered in our brief time with Buzz, and it was clear from his honesty about current 3D’s shortcomings that there really isn’t a true 3D expert in existence. The people teaching it are still learning while they go, and doing their best to pass that knowledge on. The hope is that viewers will benefit from gradually better 3D – and, hopefully, fewer headaches.

Goldfish image courtesy of bensonkua. Convergence diagram courtesy of Panasonic.

Stargate Atlantis is rubbish. This is a fact. This one single photo should tell you just how rubbish it is, but if it doesn’t, well, let me tell you: It’s rubbish. Really.

But the real eye-opener is not that people like it, it’s that in 2008 more people downloaded each episode from torrent sites worldwide than watched it on TV in America. While the big shows mentioned above still roped in vastly more TV viewers than downloaders, Stargate Atlantis tipped over the edge.

Why this show? Well, by the look of the various fan sites and forums out there it’s globally popular, but that popularity isn’t necessarily respected by the publishers and networks. Complaints come from New Zealand, Europe, even as close to the US as Canada; there’s a long wait between the show airing in the US and making its way abroad, so it’s inevitable that these fans will look elsewhere for their fix.

The primary reason most TV downloaders do so is not to steal or to fight back at the big corporations, it’s simply that they want to watch a TV show right now and downloading it is usually the easiest way. And it’s an argument I’m increasingly coming round to.

You’ve followed a show for years, complete with intrusive adverts, yet you’re still expected to wait months for it to reach these shores after it’s finished in the US. It’s not only a little insulting, it’s also incredibly frustrating, particularly as spoilers instantly fill the internet for you to avoid as you count down the months till you can watch it legally.

It’s the same with region-locked DVDs. The studios try to control the release across multiple territories to coincide with charm offensives in each region in turn, but the moment the first territory gets it and starts raving about it (usually the US) the rest of the world gets annoyed and just downloads it. And why shouldn’t they?

A few shows have seen the light and changed schedules – 24, Prison Break and several other big names now air over here in close approximation to the US – but as long as others don’t, the industry can have few complaints when people veer towards the path of least resistance: the good old internet, destroyer of all things traditional.