Month: July 2018

One of the stranger things about existence is that everything seems normal. You are born, you grow up surrounded by fellow creatures much like yourself, you become socialized.

So a lot of things just fall under that category of “normal”, because that’s the way it is and that’s the way it has always been. As far as you can tell, that’s the way it will always continue to be.

But every once in a while you catch yourself looking at something “normal”, and find yourself thinking “wait, this is weird.” Recently I have been thinking that way about sleep.

Imagine visitors from another planet who have no concept of sleep. Unlike us, their biology does not require it.

Such beings would not need to spend roughly one third of every day in an unconscious and essentially helpless state. They might be amazed to see that we, fellow intelligent beings with whom they have just held a perfectly nice conversation, are suddenly collapsing all around.

Our visitors would never dream. They wouldn’t even have the concept of dreaming until we have described it to them.

I am trying to imagine how sleep-prone creatures like us would seem to such observers. I’m pretty sure they would find us to be intriguingly non-normal.

Some people worry that as techniques of procedural animation develop, the result will be less work for animators. I beg to disagree.

The reason any of this is an issue is a consequence of the forthcoming wearables, which will soon replace SmartPhones. Wearables lead inevitably to ubiquitous augmented reality, which will pretty much demand the presence of interactive procedurally animated characters.

I’ll explain. In order for responsive augmented reality characters to work properly, they will need to be driven not by traditional animation techniques, but rather by procedural animation. But that doesn’t mean animators will be out of a job.

In fact, when animated characters go completely procedural, they will need to be trained by great animators. So the animators will not be creating animations individually, but rather will be “training by example”, functioning essentially as acting coaches for a new breed of virtual actor.

This means that the value created by the animator will be monetized in the form of licensable I.P., rather than by payment for animation services on a specific production. This will be great for the animator.

That’s because it’s always better to be paid for use of one’s property, rather than relying on an hourly wage. A good animator will have the opportunity to make a lot more money, because her creative output will be able to be used on many productions — and she will not even need to be involved in those productions.

This evening I got a dance lesson from a friend who really knows how to dance. It was wonderful and fun and enlightening, but also humbling.

The experience made me realize how much I generally live within my head, and not within my body. I wonder how many experiences I am missing by not having my full body involved in the processes of my mind.

Something my friendly dance instructor said really stuck with me. She told me “I need to teach you not to think.”

Yes, I can see that. There is ancient wisdom in the body. Alas, it is a particular form of wisdom that the mind can all too easily forget.

I am reminded of Zen and the Art of Archery: You can’t think your way to learning how not to think.

Late breaking news in today’s NY Times (lightly edited for style):

White House Doesn’t Abstain from Denying Trump Didn’t Avoid Saying Russia Isn’t Refraining from Not Targeting U.S.

Asked whether Russia hadn’t stopped “still targeting” the United States, President Trump did not deny declining to eschew saying “No,” and so evaded desisting from appearing to not fail to contradict assertions from none other than his own intelligence chief.

Hours after the comments, the latest in a dizzying series of conflicting statements, the White House said Mr. Trump was answering a different question.

Right from the start I knew today was going to be a long day. For some reason I woke up around 5:30am, so I got into the lab long long before anybody else.

By the time anyone else arrived I had gotten all of those things done that I had been telling myself I really aught to do (but hadn’t). And then other people started showing up.

At which point the day turned into an hours long hack-fest. For most of the day, the members of our little team were huddled together working intensely on the project we will be showing at SIGGRAPH four weeks from now.

And then I headed up to Lincoln Center to see Visconti’s The Leopard. I had seen it before, but never on a big screen with a gorgeous print (which is really the only way to see a Visconti film).

My friend had never seen anything directed by Visconti — or any movie starring Burt Lancaster for that matter — which really added to the fun and magic.

Looking back, I would say it was a good day.

David Alan Smith wisely pointed out yesterday that the mammoths I create are imaginary, so all the rules are different.

His insightful comment has inspired me to write a poem, which I’ve posted to our Future Reality Lab blog. Click on the image below to read it.

Today I accompanied a friend to a store that sells kitchen furnishings. She and her boyfriend recently moved into a new house, and they are doing a massive renovation — hence the search for kitchen furnishings.

The salesman at the store asked my friend about the layout of her kitchen, and she said it might vary. The reason is that they are considering moving the position of one of the windows.

I found myself thinking about the immense amount of effort it takes to move a window. Essentially, you need to cover over one window, both inside and out.

Then you need make a new hole in the wall somewhere else, avoiding structural support elements of course. Finally you need to finish the second window, both inside and out.

To the naive observer, it will look as though the window has simply moved its location along the wall. Yet in fact it is an entirely different window.

So I asked myself the following question: Will there ever come a day when real and virtual realities become so thoroughly mixed that such massive reconstruction is no longer necessary? Will we ever get to the point where to move a window in our house we simply slide it along the wall to another location?

I realize that there are a lot of different conversations packed into that one deceptively simple question. But that’s rather the point, isn’t it? We might as well start those conversations now.

One thing you learn when you are developing procedural animation software is that when things go wrong, they can go very wrong. Unlike creatures in real life, a virtual creature can fail in very spectacular ways.

The head can fly off the body, knees can bend the wrong way, or parts of the body can simply disappear entirely. When this sort of thing happens, all you can do is start tracking down the bug, and hope that such symptoms won’t pop up in the next public demo.

Recently I managed to track down the cause of one insidious bug. Every once in a while, when my interactive mammoth moved its legs in a particular way, it’s entire upper body would vanish.

It turned out that the problem was a negative number under a square root. For those of you who don’t know, there is no real answer to the question “what is the square root of a negative number.” So when you try to take the square root of a negative number, you get back a result that is, quite literally, not a number.

Once those first “not a number” values show up, everything starts to break. Creatures move in impossible ways, body parts disappear, and general mayhem ensues.

When I next saw my colleagues at the lab, I felt somewhat triumphant. After many hours of searching, I had indeed traced the problem to a negative number under a square root sign.

Of course by this time I was somewhat tired and burnt out and maybe a little sleep deprived. So the way I explained it to my two colleagues was this: “I know why the mammoths went extinct.”

“Why?” asked one colleague.

“Because,” I explained, “there was a negative number under the square root sign.”

To my relief, they both understood completely.

The Jurassic Park franchise is now more than a quarter of a century old. I recently watched the latest installment not really expecting to see a movie, but rather a collection of very impressive special effects.

And I was not disappointed. What I saw didn’t feel like a movie in the traditional sense. I happily watched giant dinosaurs chasing people, and I didn’t feel a bit emotionally involved. And that was ok, because the dinosaurs looked completely awesome, as expected.

I found myself wondering why I felt so detached from all the chasing and chomping happening on the big screen. At some point it occurred to me that it was because of what I now think of as the “Jurassic Park superpower”.

Basically, any character who you care about in a Jurassic Park movie is completely invulnerable. The biggest baddest carnivorous dinosaur cannot harm such a character, no matter how hard it may try.

This includes any character who is a hero, or a heroine, or a cute child, or simply has a good self-deprecating sense of humor. Which of course leaves the bad guys.

Bad guys in these movies have a sort of anti-superpower. The iron clad rule in a Jurassic Park movie is that if you are the sort of low minded fool who willfully unleashes the mighty beasts of death and destruction (usually because you want to get rich), then said beasts will inevitably devour your sorry self before the end credits roll.

It would ruin everything if anyone on-screen realized they were actually in a Jurassic Park movie. None of the good guys would ever run away in terror, the bad guys would know better than to mess with Mother Nature in the first place, and the whole enterprise would fall apart.

Good thing they don’t know.

When I was a child I learned that in some traditional folklore, it was believed that inside of every human there is a miniature fully formed human, a homunculus, which is the true source of human intelligence. Sort of like having an operator inside a robot.

Of course this leads to a reductio ad absurdum. If you follow this concept to its logical conclusion, then it’s homunculi all the way down.

Yet we do something vaguely analogous in software. We build software that is based on other software, and below that is still other software. Eventually you get down to the actual processing hardware, but these days that is indeed many levels down.

The fascinating responses to yesterday’s post revived an old idea that I had pondered quite a few years ago. Suppose you made a simulated world, and populated it with virtual people. Then suppose each of these people was actually a colony of virtual beings, collectively charged with keeping their uber-individual functioning.

Now suppose you made this recursive. It would be fascinating to implement such a system, and try to understand its properties.

Precedents abound in literature, ranging from The Thirteenth Floor to the scene with Burt Reynolds and Tony Randall in Woody Allen’s extremely loose adaptation of Everything You Always Wanted to Know About Sex * But Were Afraid to Ask.

But suppose we were to try this sort of thing for real. I suspect we would quickly learn that there are non-obvious rules to working versions of such systems.