[links] Bruce Willis sold his face [updated], Darth Vader sold his voice, and the Time Bandits are back
the endless void of same is a wonderful drug
We'll never go out of style
Getty is banning AI-generated images, not because of the flood of synthetic imagery like the online art communities, but because of unclear legal implications regarding copyright. Bot questions are interesting. Art communities are banning AI art to preserve a sense of exclusivity around human made art and to serve their users who are, at least a huge chunk of them, hostile to the technology. I spoke to some of the harsher critics of this tech and we’ll see where these dialogues are going.
I also wrote about the legal pitfalls of AI image synthesis, here and here. I suspect that OpenAI and Stable Diffusion and Midjourney are sitting on a copyright-timebomb, because the used an academic dataset (which would be legal under fair use) for a commercial product. However, one thing is clear: Even when some rights holders sue the shit out of these new image makers, in two years latest we will have a black market of specifically trained stable diffusion versions (see below) trained on highrez scans of all Calvin & Hobbes comics that can reproduce Bill Watterson if he likes it or not. You can download them via piratebay and calvinandhobbes anything you want in all image editors. This is Napster, but with artistic styles.
Greg Rutkowski, arguably the most prominent artistic style in the Stable Diffusion dataset, has an Interview with Technology Review and says that A.I. Should Exclude Living Artists From Its Database. I get that sentiment. I think living artists should be able to sell their artistic style in the form of specially trained Stable Diffusion Plugins in an Image Generator from the future, which has a baseline trained on art history on works in the public domain, and in which you can add modules "in the style of Greg Rutkowski". This will open the door to a black market for art styles, but this already happens anyways. I don't want to sound too cynical, i worked as an illustrator myself, but instead of complaining, artists at least can compensate somewhat this way, because this technology will not go away (ask Metallica, Greg) and progress is rapid these days.
Progress is rapid and Furries, the subculture that lives as fluffy anthropomorphic animal thingies and also has a very very active subsubculture engaging in illustrated porn, developed their own Stable Diffusion model to create, you know, images of fluffy beasts with genitalia in all shapes and forms. This will be interesting to watch, i can promise.
Two weeks ago a company behind AI-avatars presented AI-generated videos from a single image. These were just talking heads and are similar to Deepfakes. Yesterday, Meta presented Make-A-Video, a Dall-E like technology to generate Videos on prompt (here’s The Verge article). These videos are not usable or great and they are short. But: A few hours later researchers published Phenaki, a model to generate, well, Videos with "prompts that can change over time, and videos that can be as long as multiple minutes." These videos, too, are not usable or great, but they are long. You know what was not usable or great, one year ago? Dall-E 1.
In a related move, researcher presented DreamFusion, a Text-to-3D using 2D Diffusion, a model to generate 3D-models from a prompt. Let me give you a scenario where this is leading: In one year you will be able to generate any CGI-character as a 3D-model, import it into a video-model and make it dance to "Thriller". Maybe this pipeline will be automated, maybe it will be as easy as one single textprompt. I expected all of this next year, maybe late next year. Surely not yesterday, and surely not within one day. The speed of AI development was insane, 2 weeks ago, and it keeps speeding up. I'm pretty confident that no copyright case will change anything about that.
Here's two artists who are embracing this new tech and you know both of them: Bruce Willis sold the rights to the look of his face to a deepfake company [update: looks like this is completely made up by “journalists” at The Telegraph]. It's the same startup he did that russian commercial last year and he say's, "a great opportunity to go back in time". Its also a great way to freeze time, forever. Who needs a new Bruce Willis when you can have the face of the original Bruce in Die Hard 27? If this technology was available in 1960, we'd today have no Jim Carrey because nobody would have needed a new Jerry Lewis. And Darth Vader sold the rights to his voice to an AI company too. In any case, you will meet Bruce Willis as an artificial human in a cinema near you and Darth Vader will never be voiced by a young upcoming actor, that much is safe to say. It’s also safe to say that in 10 years you may very well be able to license the voice of Darth Vader to put into the mouth of a synthetic Bruce Willis who is acting in a Matrix Movie that you created with a text prompt. I’m not sure if thats a horror movie or not.
The Internet already made us into atemporal creatures clinging to nostalgia, and as much as I am fascinated by AI and its possibilities, the outlook of eternal artificial ABBAs and Bruce Willisses and Darth Vaders is frightening.
We’ll never go out of style, frozen in time.
Time Bandits and the art of the archive
Speaking of Nostalgia and Time: 5 years ago Taika Waititi spoke about plans to reboot the impeccable Time Bandits and I haven't heard since. Now there's a casting and a first plot outline ("a comedic journey through time and space with a ragtag group of thieves and their newest recruit: an eleven-year-old history nerd") and a plan for 10 episodes two of which will be directed by Waititi himself, all of which sounds wonderful and I have nothing to add, except that I myself too can't escape the beautiful cozy lull of nostalgia.
This paragraph in this short article about AI art captures this moment perfectly: “Creative AI is the art of big data, perhaps the most direct form of artistic reproduction. What could be more suitable to our moment? As Jean-François Lyotard wrote in The Postmodern Condition, ‘Data banks are the Encyclopedia of tomorrow. They transcend the capacity of each of their users. They are ‘nature’ for postmodern man.’ Creative artificial intelligence is the art of the archives; it is the art derived from the massive cultural archives we already inhabit.”
This widely shared article about God and Machines, the magic and unexplainableness of AI, is a bit too esoteric for my tastes, but I like it’s teaser: “The future of artificial intelligence is neither utopian nor dystopian—it’s something much more interesting.” I’m not sure if it helps to compare stochastic libraries to magic, as these are just very big statistical models. But maybe thats, too, what magic actually is: Datasets too large for us to understand. After all, that’s what a magician actually does: Overloading your senses with smoke and mirrors and suddenly a rabbit appears. In this case, it’s a white robot rabbit luring artificial Alice into a new Wonderland. (356 bunnies and counting).
The Rebirth Of Magic - AI Short Film (Terence McKenna): “AI short film based on a prophetic talk by Terence McKenna from 1998 about the new advances in virtual reality and the rebirth of magical speech in the community of digital artists.”
You can play Lunar Lander in the file copy dialogue on Windows now
More gaming: The Gartner AI hype cycle 2022. Robot cars are entering the plateau of creativity, generative AI is one step before climaxing, and AGI is still more than 10 years away, regarding Gartner that is.
Brain Image generation with Stable Diffusion. Here’s some tweets with images of artificial brains. You can use fixed seeds to mutate these brains. This is just great.
The WaPo recreated Ferris Buellers Day Off to see if you could do all what he did in one day in Chicago. Spoiler: They pulled it off, but cheated somewhat at the end. Which is, frankly, perfect Ferris Bueller style. <3
The lab meat industry could use some munchies. Maybe people should look at the Omakase Beef Morsels instead of McDonalds, “richly marbled bitesize cuts of 100% cultured beef using a unique 3D-printing process”.
Butcher Billy illustrates Atari 2600 Classics and all of these are great. Look at this Yars Revenge beast though.
Das digiTal ist endlich bei Colagen und sprechenden Geburtstagskarten angekommen :)
Insgesamt löst es leider keine Probleme der Jetztzeit und Zukunft. Am Ende stehen wie eh und jeh nur monetäre Interessen. Das ist schade aber gewohnt menschlich.