Similar Posts

Subscribe
Notify of
8 Answers
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
alaskamusher
1 year ago

To the info!

Last month, a YouTube user named demonflyingfox invited a video titled “Harry Potter von Balenciaga”.It showedCharacters from the Harry Potter films – Hagrid, Ron, Hermine, Snape, McGonagall, Dobby – as tough models with aggressive cheekbones (lightlyinsane), dressed in Gothic shoulders and leather jackets. To a runway-worthy Electronica Beat blink, nod and speak the actors lines from the books that were remixed with fashionable references. “You’re Balenciaga, Harry,” says Hagrid, instead of sending the message that Harry is a magician. The video is strange and hilarious. Within three weeks it has received almost five million calls; A continuation published less than a week ago has recorded more than one and a half million. Popcultural mashups of a famous thing with another are an archetype of internet meme education. The unusual thing about “Harry Potter by Balenciaga” is that it was created with tools of artificial intelligence. As the creator of the video, the Berlin-based photographer Alexander Niklass, who created the channel demonflyingfox, said, the video demonstrates the newly discovered ability of the AI to create “film-like moments”.

AI tools were involved in each step of the Niklass process and in each element of the video. He created the basic static images with Midjourney and recalled the Harry Potter actor and outfits by text requests such as “male model, grotesque, Balenciaga advertising”. Then he used ElevenLabs – a “Voice-Cloning” tool – to create models of actors’ voices based on previously recorded audio recordings. Finally, he fed the images into a service called D-ID, which is used for creating “avatar videos” – subtle animated portraits that are not so far away from those appearing in the Potter world newspapers. D-ID added the characteristic lip synchronisations and head nods, which according to Niklass were an appeal to models that tended their chin for the cameras. The combination of child-friendly film and luxury fashion for adults had neither a special symbolism nor expressed an artistic intention. It’s “sustainability,” Niklass said. But the most convincing aspect of the video is perhaps its void, a meaningless collision of cultural symbols. The nonsense is the point.

K.I. tools were perhaps able to replicate the faces of actors and to generate fashionable outfits, but only Niklass could imagine the concept that required a sharp observation of both the skin couture and the magic world – and also a very specific, extremely online sense of humor. With tools like Midjourney, which are open to everyone online, “can everybody create something visually appealing,” he said. “But AI can’t make any taste yet,” he continued. With “taste” he meant “a good aesthetic judgment” – background knowledge about what you create, and a sense of what looks good without falling too much into the creepy valley. In other words: The execution may have been democratized by generative AI, but not ideas. Man is still the author, editor and curator of the impact of AI. A proof of Niklass’s taste is the many imitators of his videos that are now on YouTube: Everyone can access the same technology and try to create a replica according to the formula he set. (OneVideo tutorial, created by an AI training channel called PromptJungle, shows the exact process.) There is “Matrix by Gucci“Star Wars by Balenciaga” and “The Office by Balenciaga“videos, but none of them is as appealing as the original.

EinTyppie
1 year ago
Reply to  alaskamusher

? Boy what’s going on? XD Have you let your answer scale up AI or something. The type just wanted to know how to do it, and you still write every person, every character, etc, wtf

alaskamusher
1 year ago
Reply to  EinTyppie

Not really, is from the New Yorker…!

EinTyppie
1 year ago

Take absolutely garnix lol

mompf03231
1 year ago

Midjourney. For animation, for example, pikalabs, stable video diffusion or runway gene 2 . The lipsyncing is another, maybe first order motion model or something newer

mompf03231
1 year ago
Reply to  Betelgeusi

Ne