How to do 3-D
3-D Film Theory
3-D Film List
3-D Film Review
3-D Camera Build
21st Century 3-D
The merits of mocapped 3-D
It is no surprise that the path to the current 3-D boom is being beaten by animated and motion-captured feature releases. Chicken Little, The Polar Express, The Nightmare Before Christmas, Monster House, Meet the Robinsons, Beowulf and the upcoming Tin Tin and Avatar are all animated and mocap CGI titles. Animation provides ultimate, pixel-perfect camera alignment - and control, and in 3-D Stereoscopic film that is a very good thing. 3-D has always been - and still is - a technically challenging medium on the filming side because of the camera alignment factor. Slight misalignment means a disturbance of the 3-D and will result in eyestrain, unless the misalignment problems are corrected as much as possible in post. One bad live-action title now and 3-D runs a risk of dying a premature death (again) as 3-D has proven in the past that it can acquire a bad name very easily.
Mocap and Beowulf
This review is not intended to discuss the issues with mocap, but it is important to mention its shortcomings and merits in relation to the 3-D. From what I hear and read about the experience of Beowulf by movie goers – from professionals to journalists and regular consumers – the wax face and wooden movement typical to mocap is a big put-off and it makes people think twice about going to see another 3-D feature in the future. People don’t want to go and see in-game style animation in a cinema, they want to see real humans act out human stories. Mocap is all too often described as ‘technology for its own sake’. “Why use a CGI version of an actor that looks like the actor is you can show the real actor, without the use of CGI?”, is a line I have heard and read every time in relation to mocap in general, and Beowulf in specific. I will leave that issue with the decision makers in Hollywood.
The 3-D does help the mocap to be accepted by the audience, as it distracts the viewer from analyzing the computer generated image. This is an unintentional side effect, as the mocap was used to make the 3-D easier to produce. A vital element to any movie is the ever so important suspension of disbelief and by using mocap, it is a struggle to keep doing this. The game-like visuals and movement of characters put the audience back in the cinema rather than out there in Medieval Denmark, with the viewer analyzing the 3D modelling, lighting, shading and rendering technique rather than following the ‘human’ interaction.
A case in point is the first appearance of Grendel in the movie, which heavily damages the suspension of disbelief. Because as an audience you have been trying very hard to accept the human characters as real, the sudden appearance of a fantasy monster makes that creature look more real than anything else. That’s because we don’t know what such a monster is supposed to look like and thus our brains can accept the appearance as it is. But in doing so it creates a point of reference for what looks real and suddenly the human characters standing next to Grendel look very fake and computer generated. The brain is ruthless that way.
The same thing happens to a lesser extend with the play-acting dwarf: because we don’t look at dwarfs every day, we don’t know how they are supposed to look, exactly, and how they should be moving. He therefore looks more convincing than any of the normal size people around him.
It feels like Hollywood is downgrading its quality standards to computer game levels just to get the 3-D right. But what is more important in the end? The image quality should not a slave to the 3-D: it should be the other way around! The way of releasing CGI 3-D films as is happening right now, the movie-going public will identify 3-D with CGI, and mainly with mocap. And judging from people’s reactions, that is not necessarily a good thing.
The cinematic use of 3-D in Beowulf
Positive and negative parallax: Depth
This is a good film to compare with as both releases were directed by Robert Zemeckis and produced by the same production company, both using motion capture. Overall, Beowulf is not as engaging as ‘The Polar Express’ (TPE). TPE is more enveloping because of the bountiful and better use of negative parallax: audience space and out of screen effects. TPE envelops the viewer and draws the audience into the picture. With Beowulf, we remain an observer outside the window of this movie – the stereo window, to be precise. But TPE also uses more and better positive parallax: it has deeper scenes with more eye-guiding planes, characters and objects.
The location of the principal characters just on the screen plane is very nice and always works well in 3-D, as long as those characters don’t touch the sides of the frame. Such slight negative parallax enhances the volume of the character and makes him or more engaging and more pleasant to look at.
The deep scenes that do appear in Beowulf lack depth reference and negative parallax, the big open space in front of the screen, is not explored. Z-axis travelling is minimal, with most actors staying on their depth plane, which creates a very static performance in terms of composition and layout. This is not a play performed on a narrow stage, so allow the actors to move diagonally and sit in a more interesting composition in depth once in a while.
Perhaps it is more a matter of balance, though, where TPE manages to go from deep shots to calm narrow shots, whereas Beowulf rarely uses deep scenery and keeps the narrow depth scenes not very engaging. The use of shallow depth in 3-D is a good thing, especially when followed by deep, heavy 3-D. The eyes need rest and with a 2 hour long movie you’ll need all the shallow scenes you can put in. But they do still have to be interesting narrow shots. This is 3-D cinema we are talking about, so the 3rd dimension has to be used just as much as any other cinematic compositional element.
There are very few shots using extreme negative parallax, poking at the viewer. These shots are often the most memorable ones of a 3-D movie, as they threaten the viewer directly, but they are, by no means, the benchmark of a successful 3-D film. It is difficult to use functional eye-poking, almost as difficult as finding ways for sex scenes to be ‘functional’ in a story and in the overall narrative, and more often than not the audience perception is one of ‘gimmickry’. There is one much talked-about shot of a spear being pointed at the audience; a POV shot through Beowulf’s eyes. It is much talked about because of its peculiar scaling of the spear and the fixing of the camera to the tip of the spear. By doing this, the shot is out of place in the sequence of shots and sticks out like a sore thumb.
Here is something Beowulf is doing very, very right. The volumes of the characters, especially those of Beowulf himself and the monster Grendel are a treat to look at. The design of Grendel is perfect for an appreciation of his form in 3-D. Looking at volumetric shape like this is something that 3-D makes the viewer do and an elements that has to be thought of seriously in any 3-D film production. In Beowulf, it has been done very properly, with round, organic shapes that have plenty of texture and imperfections to make for great volume and depth recognition. Grendel becomes very touchable, very real, and because of this, very scary. The 3-D really helps to enhance his scariness and 3-D is the best way to show Grendel on-screen. The movie could actually have been flat for the most part, with the scenes involving Grendel and his mother going to 3-D. This would have worked with the story as the appearances of these characters are within the realm of the fantastic and the magical, while the human world is grim, dark and rather flat anyway.
The flat background planes help in enhancing the volume of the foreground characters. Again, the effect of the 3-D is a very dependent on reference.
Something that allows the viewer to really focus on 3-D space is when characters and objects rotate, spin and flip over, while the camera keeps them in the centre of the frame, not touching the frame sides. This is done quite a few times in Beowulf when people in the great hall are thrown about by Grendel. It gives the viewer a gratifying volumetric treat to see an object or person from all sides in 3-D.
A very nice volumetric experience occurs with a slow camera move past the king’s crown, allowing the viewer to truly examine the detail on the crown. So that’s more of a geometric and metal working appreciation than anything else. But it works well in 3-D.
Overall, what happens is that with the CGI imagery the viewer ends up scanning the image for volume and volume detail all the time because of the crisp and physically flawless quality of the rendered image. This distracts from an emotional experience of the movie and puts the viewer’s brain into technical analysis mode. And that is very damaging to the very emotional experience that make one enjoy a movie and get lost in the way it tells its story.
The only 3-D value that is varied and controlled in Beowulf is the interaxial value. Convergence has not been used because there is no need to use convergence on a CGI production, other than for creative reasons. It could be used to compress space in a particular way, but the resulting geometric distortions of the image will have to be corrected and so there will have to be a real creative imperative to use convergence with CGI. As there was clearly no desire to manipulate the image in this way, convergence was not used.
That leaves the interaxial. Beowulf uses very conservative interaxial values, close to the regular human sight of 65mm. On wider shots, a bigger interaxial is used, but nothing too spectacular is done in this respect. It is one of the most boring uses of the interocular possible and Beowulf never surprises or does anything unpredictable or even creative in terms of interaxial values. In terms of pushing the envelope, it’s certainly not done with the 3-D camera work.
There is one particular shot which uses a very nice interocular to create the most fitting 3-D image. It happens during the battle with the sea monster, when Beowulf climbs up to the monster’s head. A large interaxial here makes Beowulf look smaller, almost like a miniature person. The choice of a wide interaxial was clearly made because this is a wide shot, but its effect makes Beowulf look like a toy figurine. That works very well with the situation he is in: fighting a mystical, massive sea monster and as this is a story being told by Beowulf, it is fantasy anyway. The perceived smaller size of Beowulf on this monster therefore works very well.
It’s just too bad this kind of shot is not used in a creatively conscious way, supporting the scene, the composition, the framing and the story, rather than slavishly setting the interaxial according to focal length and camera distance to subject.
As a general rule, fast camera moves don’t work with 3-D. The eyes can’t focus well and lose 3-D fusion, which creates discomfort. The camera pans and swoops in Beowulf are too fast and do indeed create this discomfort.
On camera trucks and crane shots – and Zemeckis does love a good flying camera – too often the frame is filled with passing objects like actors, supporting beams, chairs, tables, trees, rocks and so on. This is 2-D cinema framing and it just doesn’t work with 3-D. The reason it is very unpleasant to look at, with audience members yanking off their 3-D glasses, is that objects at the edge of the screen – left and right – cause retinal rivalry. Retinal rivalry is a big enemy of the 3-D filmmaker: image parts that appear in the left eye image but not in the right – something that is almost unavoidable for any background scenery but an issue that is very avoidable for foreground objects.
One of the elements that make this movie look like in-game cinematics is the use of full depth of field in too many shots. Having a full depth of field is 3-D fool’s gold. The eyes may be able to scan the image this way, but unless a shot presents a scene that is worthy of being scanned completely by the eyes, there is no point to this. Getting rid of the cinematic visual technique of limited depth of field is a big mistake. In cinematic terms, it’s like cutting of an arm. It’s not just that the film now suddenly doesn’t use a cinematic tool that has been used since the beginning of film, over 100 years ago; having full DOF makes focussing on foreground objects more difficult. This is, because the eyes cannot fuse the foreground and the background planes at the same time, and that makes for visual distress. A blurred background allows the eyes to fully focus on the foreground object, making for calmer 3-D and overall, a more cinematic look.
3-D professionals are of different opinions on this subject matter, but it is my strong conviction that it is a big mistake to use full depth of field just because a stereographer tells you to do so. The audience is not looking at still frames of the film, which may well work better with full DOF; they are looking at a cinematic image that is moving and changing and is demanding on the eyes in focussing on the subject alone.
The framing of Beowulf is really very classic and is clearly done form a 2-D perspective. True framing with 3-D in mind would not involve over-shoulder shots and frame-filling foreground objects with camera travels. Somebody who frames objects like this is not keeping the 3-D in mind because such placement by the frame edges creates retinal rivalry. The frequent use of such shots in Beowulf means quite unpleasant viewing and although I can understand that veteran film makers are used to framing shots that way, again it has to be stressed that this is not 2-D film making and different rules apply. One can’t simply ignore the facts that come with a different medium. If you want to film to work in 2-D as well, edit a different version. Since this is a mocap feature, moving an actor or a tree over to get out of the shot for that 2-D version should not be a massive undertaking. With 3-D you just can’t have your 3-D cake and eat it in 2-D as well.
One tempting use of object movement that works in 2-D but not in 3-D is the movement of objects into frame from the top or the bottom of the frame. Objects flying into the frame or out of the frame never work with 3-D, as they create heavy negative parallax and eyestrain at their furthest forward point. When such an object suddenly disappears off-screen, the eyes have to reconverge on the scenery and this can be an unpleasant experience.
Those are physical limitations, but psychologically there are issues as well. When objects come forward through depth and then suddenly disappear off-screen, the mind gets annoyed with that object first being right in front of the viewer and then suddenly not being there any more. To say: “Oh, but people are used to that: objects fly out of the frame in regular film all the time” doesn’t quite cut it because this is a different experience of the image. It is not 2-D cinema, where the viewer is not having a physical experience of the imagery: this is 3-D, where the viewer feels the objects in front of him are actually there.
In Beowulf, such situations occur with the tail of Grendel’s mother entering the frame slowly and characters walking in and out of the frame, besides the appearance of spears, arrows, chains, trees, and so on.
Colour & lighting use
The world of Beowulf is a boring, dull winter land with greys and blacks rather than the lush colours of the Polar Express. Of course that fits better with the story, but it provides no guidance for the eye as characters blend into the background in terms of colour use.
The lighting design of the scenes is very dark. Yes, American DOP’s love dark scenes, but dark is not good for 3-D. First and foremost, it is difficult to place objects in depth when there is little background reference as a result of darkness. Secondly, darkness creates high contrast, which results in ghosting. This especially happens when bright objects sit on a dark background. In Beowulf, there are oh so many candles sitting on black backgrounds, which means a lot of ghosting. Real-D’s ‘ghosting reduction’ can’t help with these shots, because one can only make a black so grey before it damages the colour integrity of the shot.
Particularly on shots involving negative parallax, high contrast creates ghosting, and this happens quite a bit with well lit characters on black backgrounds.
There is one very dramatic and very effective use of colour contrasts: the golden dragon. Especially when framed against the dark grey, blue sea, the bright golden dragon stands out beautifully. Colour definitely places objects in depth as well, as the brain does interpret such colour contrasts as indicators of Z-axis position and relevant placement in depth.
Water in 3-D
Water in 3-D can be a dangerous thing. The detail of waves and moving water is often too much to take in and will easily make the viewer lose fusion of 3-D, resulting in discomfort over the area of the water. The swimming match scene in Beowulf does just that, but the saving grace of the offending shots is the focus on the swimming men, drawing the eye away from the waves.
The very large, towering waves then produce fantastic planes of reference and nice volumetric, well textured shapes. Because of the bigger size of such massive waves, the 3-D works well. It’s smaller detail that can be difficult to focus on. A good example of such difficulty is the pebbles on the beach in a later shot, with the camera sitting low on the ground, showing a deep scene of pebbled beach. That is simply asking for visual discomfort in 3-D, and that is exactly what happens in the shot.
There is one particularly beautiful scene with the mother of Grendel appearing to Beowulf in his dreams as the queen, floating in the air with her hair waving like she is underwater. This kind of slow motion movement works very well with 3-D and it is a pity it is not done much more in the rest of the film. The climatic shot of the film with Beowulf grabbing into the dragon’s chest, reaching for his heart, is another fantastic example of the power of slow-motion and slow movement in 3-D producing beautiful results. That shot stands as one of the best shots in the movie, and it being the climax of the movie, that is fantastic timing.
Much more of a classic 3-D shot, and one that works brilliantly, is that of the dragon about to eat the POV camera, even sticking out his tongue to gobble us, the audience, up. That is not just a beautiful shot because of its framing, use of colour, volume of the dragon, and use of slow-motion, it also works really well because it taps into the audience’s primal fear of being eaten by a big animal. And because this is 3-D imagery, the audience will subconsciously feel very threatened, resulting in the perfect response to the imagery: real fear and admiration at the same time.
Losing the focus
A funny thing happens after prolonged viewing of 3-D imagery: the brain gets comfortable with the 3-D and doesn’t just stop being excited about the imagery being 3-D, it starts processing the information as if it were 2-D again. After about 10 minutes of intense 3-D, the brain will get tired and zone out. In Beowulf, the 3-D is very mild and plenty of narrow scenes are presented so that the point of losing focus on the 3-Dness of the imagery happens around 30 minutes. This is a strong case for using strong and even extreme 3-D shots to wake up the viewers’ brains from time to time. That doesn’t mean the 3-D should be constantly stimulating like this, because then the zoning out happens much sooner. It just means a 3-D movie needs well placed and well spaced jolts of excitement. In Beowulf, such stimulation is few and far between.
There are many reasons why prolonged viewing of 3-D is tiring to the eyes and brain. It’s not just bad 3-D or bad projection, it is the very fact that the brain thinks it should be focussing the eyes before and behind the screen, while the actual imagery is on the screen plane. In Beowulf the tiredness hits during the long and plentiful talking scenes.
3-D film is a ticky medium and it requires much more attention during shooting and in postproduction. But then you are choosing to produce a 3-D movie so accept the technical and creative consequences! This is 3-D film making, so don’t try to avoid having to deal with the core issues of the medium. I feel that Beowulf is an exercise in 3-D film production without having to deal with too many 3-D aspects, both technical and creative cinematic ones.
The 3-D in Beowulf is done in a safe way, aiming to avoid any discomfort with the viewer where it could enhance the 3-D experience and neglecting to frame shots differently when they cause discomfort in 3-D. The discomfort is minimal, though, and overall the experience of the movie is alright. It could have been super with just a bit more attention to the 3-D and a bit less attention to making Beowulf’s beard hair absolutely perfect. That is not something the viewer remembers of a film.
Beowulf is not a masterpiece, but it’s not bad either. A bit more guts from the stereoscopic direction, as shown by the character Beowulf himself, would have been very nice. But we’ll leave that to non-risk-averse 3-D filmmakers and the ones who really understand what Stereoscopic 3-D is capable of.
© 2009 3-D Revolution Productions