Whether or not the “thread” is dead, the concept certainly is not. However, I don’t have too much hope for it, because there are actually so many “human beings” involved in the creation of a movie – not just the actors. The only place where we now see any sort of “AI techniques” (that is: "nondeterministic techniques") being attempted is: “video games.” With some amount of interesting success.
So, alas, I think that we’re still quite a long way from actuallly being able to tell robots: “We are here now, en - ter - tain us …”But, by gawd, let us never stop pushing that technological envelope!
[quote=“sundialsvc4, post:21, topic:690117, full:true”]
…there are actually so many “human beings” involved in the creation of a movie – not just the actors[/quote]
Yes, many forget this, and it can be disturbing the influence seemingly minor people have. My guess is that technology will make key or easy work more efficient until all but a fraction of a certain job can be automated, then the tools will put more tasks under fewer people. Ironically, a very organic way for computers to take over large chunks of the creative process.
This is where I am mostly on board with the possibilitie right now, the idea of using AI (not real AI, but what people call AI) as an inspirational or adjusting tool in a project. Until some personal tragedy, I was working hard on this, with some interesting results. I basically tried the Seventh Sanctum approach to filmmaking, and it was… interesting! It also happens to be how the infamous “Harry Potter and the portrait of what looked like a large pile of ash” was made. To link it back to the previous quote, I think there could be some fun to be had from chaining together smaller bits of procedural generation, then having some kind of user friendly curator interface. Once I get through these current troubled times, I may look into it in practice…
I am absolutely in favor of continuing to find new ways for "the digital slave" to make things easier for creative humans. One way in which we very-successfully use “AI techniques” is with “crowd generators.” The computer can easily produce a crowd of thousands of digital actors which actually react to their surroundings. It would be impossibly tedious – and a colossal waste of time – to do the same thing by hand. "So, that’s what computers arefor!"
What computers so-far do not know about is “human context.” For example, in the early days of computing there was an effort to produce a “Russian to English automatic translator.” When an English sentence was fed into the translator and then back out, “The spirit is willing but the flesh is weak” became: “The wine is acceptable but the meat has spoiled.”
As we strive to improve “artificial intelligence,” of course what we are really studying is … ourselves. And may we never stop doing that.
Since 1999, Seventh Sanctum has been your source for random generators of ideas, from story concepts to science fiction weapons to extremely silly spells. If you need ideas, inspiration, full stories, or a good laugh, then go on and enjoy!
Plus, it has home-page links to about a dozen more such sites. Yep, “sometimes randomness is good for a laugh.”
It’s nice to see this thread still active. As I believe I’ve said in the original post, my idea is based on using a mixture between gaming and 3D rendering. It’s technology that has been there for years, just that I haven’t seen it used in this way so far.
Take a game like Grand Theft Auto for instance: During a typical GTA mission, the player will follow an AI that spawns in a certain spot, walks over to a car, enters it, drives to a location, exits it, walks to another character sitting there, they have a conversation, they then walk away together, etc. Using such a powerful system, you could obtain procedurally generated movies using two simple steps:
Randomize the names and visual traits of characters, from a list of names and using a set of handles (like how you can customize a character in a MMORPG). Now give them roles in a random story, by spawning points where various actions can be preformed which can be randomly picked in a logical order (eg: Go to the restaurant sit on chairs and eat).
Have the camera follow the action, jumping between angles and characters. It can even be interactive to allow the player to look around while the action is unfolding!
It would be simplistic if not done well, that’s one problem. It depends on how many locations and possible actions can be generated. The idea would be to get something original each time, not a thing that looks and feels the same way just in a different order.
In most superhero movies, you have 11th hour heroics saving the day when the villain is just minutes from succeeding in his mission. The first half of the movie meanwhile will almost always have 1 or 2 action scenes (often at the beginning or at the midpoint) with the plot not coming into focus until the start of act 3.
If a formula gets tickets sold, Hollywood will keep using it until the first few box office bombs.
Formula stories is nothing new. The irony is that AI are the epitome of formula work, but do it ‘right’ and the AI will create something completely insane. Not usually coherent, but at least interesting!
