Sukie recently had eye surgery which left her without the use of an eye for a few weeks. She remarked that it was interesting that she still could judge depth - don't you need two eyes for stereo vision?
For years people have been building 3D displays. The idea usually amounts to recording an image with two slightly offset lenses and then presenting the images to the eyes in such a way that you can detect the parallax. As far back as the 1950s many of these were intended to revolutionize movies and then television. Many are clever and the battle to put 3D everywhere continues today, but there are some serious technical problems that include the need to match camera lenses, screen brightness, lowered resolution and the fact that some cues required for vision are broken resulting in a less than quality experience for anywhere from 10 to 30 percent of viewers. Combine this with the need to create and polish a different directing technique for live action like sports or post production with movies and it is easy to see why ESPN is dropping support for 3D sports.1
It is important to think a bit more deeply about how the senses work. To date the approach has been to deliver what is believed to be the right proxy for realistic 3D to a pair of retinas. After all - this should fool the brain which should just assemble the right image. The problem is that realistic vision doesn't work this way. In fact the parallax information is only one of several cues we use to create images.
Several years ago I was given a talk at the Disney Animation Studio in Burbank, California. A fabulous place complete with a Mickey Mouse Wizard's hat, the place is filled with history. It was my first visit to the place and we got a great tour of the place, its people and some rather curious historical kit. The piece that really interested me was the multipane camera.
I had too many questions, so the guide scurried off and brought back one of the older animators to explain. You've probably noticed animations move a background relative to the foreground when motion is involved. The technique gives one of the depth cues our brain uses to create images. What you may not have noticed is the old Disney films use up to seven layers of depth and the motion is very smooth.
The technique was mostly invented by Ub Iwerks - a legendary animator and director at Disney, although he was not with the studio at the time. Several groups refined the technique during the 30s and the one I saw was invented by William Garity. Its first movie won an academy award for an animated film. Disney was an interesting place at the time. It did serious perceptual work and invented many of the tools of animation to use in-house - very much like Pixar does today. If it is core to the business, there is a need to invent, own and perfect it.
What I saw was one of the old production machines that was used in Bambi, Peter Pan,Fantasia and Pinocchio, but the film it was invented for was called Snow White and the Seven Dwarfs:-) The technique is still used in computer animation and has been highly perfected at studios like Pixar and Disney Animation - I saw one of the tools in action one some preliminary work for Tangled. It is also used dramatically by Hayao Myazaki at Studio Ghibli.2
Here Walt talks about it
One of many things that impresses me about the film makers I know at Disney Animation and Pixar is the rich model they have for human vision. This and subsequent discussions noted they see depth as coming from several cues - things like closer objects being larger than farther ones, the blurring of far away objects, relative motion, lighting cues, parallax (sometimes called stereopsis) and the focus of the eye (often called accommodation). Regular movies and video can have all of the cues except parallax and accommodation. 3D movies only add parallax and in many cases it is broken as the camera optics are not well enough matched to the eye and some crosstalk is created.
Some studios have achieved a fair amount of success with current 3D technology, but they tend to use it sparingly - mostly on the foreground and using great care with the viewing box. This is somewhat easier in animations and mixed mode films (like Avatar) where great control is possible, skilled artists are present, and there is a time and resource budget. But it is not a real 3D effect and perhaps we are entering a form of uncanny valley. The last bit - accommodation is really hard.
It is interesting watching the spec wars among 3D vendors. They have metrics which they are optimizing and some fine work has been done, but they are only attacking part of the problem. They are ignoring the full experience and, for many people, dropping an imperfect parallax leads to lower production costs and a better viewing experience.
You really need to consider the full experience.
In the late 80s the Japanese were taking over the tech world at the time and a huge effort was underway for the US to create a digital standard high definition standard that would be better than the analog version the Japanese were working on. Bell Labs was part of this Hail Mary effort and some of the physical scientists and applied math folks were asked to make sure the tests made sense. I was invited to be part of that critique.
They were doing all of the right engineering things and it was a lot of fun getting to see a bit of the future, but my experience was while the video signal was better than anything I had ever seen, what really blew me away was the audio. Since then I've talked with a dozen other people who saw the demos and tests and everyone said the same thing - really good video, but mind numbingly wonderful audio!
The audio was there because it could be. Digitally encoding the signal allowed for a lot of compression and there was space for a compressed multichannel audio signal. The test videos were carefully made with a big budget and the audio guys were world class. The demo room was set up to take advantage of this. It dawned on me that the experience of watching a movie or television is much greater than the moving image ... the audio is important.
When you talk with some of the serious movie guys they'll sigh and note 3D is there to fill movie houses and first and second to save television makers. Some good work is being done, but there is a lot of garbage and it is increasingly being viewed as a checklist item that only adds a tiny amount of value to the movie artistically at best. The monetary value is rapidly dropping and there is historical precedence for this. All of them said what they really want is better control of sound. The problem of sound field reconstruction is a thorny one and doubly difficult if the listening stage is much larger than a single seat. When it is done at a single seat level you get a dramatic response from the viewer that is much greater than any 3D effect that has been tried. This, if and when it arrives, will probably require a new direction and post production language, but may well be worth the effort.
Multipane made animation dramatically advance. 3D hasn't had a similar effect on film. You have to understand the limitations of an art deeply and have a good enough model of perception if it is to become art. Techniques or may not easily jump from one art form to another. And you have to ask yourself what is the full experience? Is your little corner of the world inclusive enough?
Apple appears to have asked that question with iOS 7. While the reception has been a bit polarizing, one thing they've done is multipane to add the illusion of depth.3 When I saw a video of their keynote demonstration I was immediately reminded of my tour of that building with a wizard's hat and William Garity's amazing machine. I'm assuming they have probably learned at the feet of their friends at Pixar as a considerable amount of experience is needed to make this feel right. I don't know how well it will work, but Apple is asking the right set of questions.
The right set of questions applies in so many disciplines. I recently read a couple of great urban planning books and recommend them as examples of people who have thought deeply about core problems - Cities for People by Jan Gehl and Walkable City by Jeff Speck. Sometimes a few questions outside the popular frame give a much deeper understanding of the real issues.
__________
1 Sports is particularly difficult as scene transitions can be jarring. ESPN is forced to use about 80% fewer transitions than best practice television uses. This makes the experience worse for 2D viewers. Fatigue is an issue for a significant percentage of their viewers and that gives advertisers something else to worry about with placement position. Some viewers become more immersed, but many don't. Both might have a negative impact on advertising. Additionally it is very expensive and difficult to maintain precisely matched zoom lens pairs on multiple cameras.
2 Very convincing flying scenes can be created with it:-) Animation is all about good enough immersion.
3 Apple's video showing the early beta of the interface. You begin to see multipane a bit after a minute into the video and then around 2:15. (there are other depth cues)
__________
Recipe corner
I had a birthday recently and that means cherry pie. Mine isn't too fancy, but demands high quality tart pie cherries. If you don't have a source it is best to forget about baking one. I won't go into crust building as that is religion. I use a combination of crisco and butter and never lard (I'm a vegetarian). A good flour is important as is chilling it. I don't use the popular vodka trick as Sukie is allergic to potatoes.
The recipe is for some good canned cherries I get from Michigan for off-season baking. This is good for one 9" pie. I like to lattice the top:-)
Tart Cherry Pie (assuming you have a the crust prepared)
Ingredients
° three 425g cans of sour pie cherries (about 5 yo 6 cups of pitted cherries if you have fresh)
° 150 g white cane sugar
° 3/4 tsp cinnamon
° 45g quick cooking tapioca or 100g of King Arthur Pie Filling Enhancer
° 1 tsp almond extract
° 1/2 tsp salt
° 30g butter
Technique
° line a 9" pan with pastry dough
° Drain the cans and reserve 2/3 cup of cherry water from one. Put the cherries and water into a large bowl
° combine sugar, tapioca or pie enhancer and cinnamon and stir in the cherry bowl
° stir in the extract and salt. Let it sit about 20 to 30 minutes if you're using tapioca.
° pour the filling into the pie pan and dot with bits of the butter
° work a lattice on top or just roll out a second crust. If it is a second crust you need to cut some vents. Get fancy with the design
° put it on a baking sheet (I put some parchment on the sheet to keep spills from burning in) and pop into a preheated 425°F oven for about 40 minutes until the crust is golden brown.
° remove and cool it on a rack. You may have to defend it.
Comments