So much cinematic AI to talk about. It deserves its own post.
This AI-generated music video by London-based music producer Pavel Zhigarev, who mainly composes music for commercials through his small studio Noizlab. He started to play with generative AI, and was inspired to create a faux music video for his favorite band, Limp Bizkit, “What if Limp Bizkit Were a Country Hit.” The video went viral. The band posted it to their socials, racking up a million views. Fred Durst messaged Pavel and asked him to make an official music video for their track “Turn It Up Bitch.”
“The whole process of making “Turn It Up Bitch” took around 2-3 weeks.” Zhigarev said. “It mostly involved learning how to input the correct prompts and then choosing the right shots from an enormous number of randomly generated images.” Luma.ai launched just as they were finishing the first cut. Zhigarev felt if he didn’t implement it into his video, the video would look outdated. The AIs he used included: ChatGPT – for creation of multilayered prompts; Midjourney and Leonardo AI – for image generation; Photoshop AI – replacement/addition of elements (removing six-fingered hands and three legs); Krea.ai – enhancement of surrounding elements;FaceSwap – replacement of generated faces with the artist’s faces; Topaz Gigapixel – upscale to 8k; Runway ML – create video from photos; and Luma.ai – to create video with the artist’s face (less face distortion).
AI does a great job with genre mash-ups. Of course, it helps when it’s in the hands of a real storyteller. “To craft the film,” explained Philadelphia-based filmmaker Tim Simmons, known on YouTube as Theoretically Media, where he is just 700 shy of 100,000 subscribers. “I assembled a motley crew of the best AI Tools on the high seas. With this story, I wanted to showcase how we can bring them all together into one narrative. I’ll have a full roundup on the channel later, but for now, I invite you aboard… but with a warning: There’s a storm on the horizon.” Simmons used Midjourney (images), Kling (video gen), Runaway Gen-3 (video), Luma AI Dream Machine (video), Eleven Labs (voices), Suno (music), and Magnific (upscaling).
“The new models are a real breakthrough; here it’s 95% Gen-3 text to video, and the atmosphere and flow are quite close to what I had in mind.” said creator of “Le Voyager,” @A_B_E_L_A_R_T, a French filmmaker and cinematographer who prefers to remain anonymous. He treated much of the content of the film as found footage (e.g. “Blair Witch”). He felt this kind of faux found footage was an excellent way to use AI, as you don’t have any more control over found footage than you have over AI generated images. “I can quite imagine two forms of generative AI in the future: one where we have total control, and the AI is 100% at our service, and a second where the AI acts independently, transmitting chaos and opening the door to “the other side,” like a mysterious search from an unexplained and independent reality.
I discovered Tokyo-based digital creator Arata Fukoe on X. He used tools we see mentioned over and over in this column to make this short. Fukoe said on X that he “used the latest generation AI to make something like a music video (it’s a mess),” he posted. “In particular, when you look at people on a large screen, the imperfections are noticeable, but depending on how you use it, you can hide them somehow.”
If you’re interested in reading more about AI and filmmaking, you might be interested in two other wo stories I dropped on Forbes earlier this week.
DreamFlare AI Launches Interactive AI-Generated Streaming Service The new service merges technology and creativity to deliver interactive storytelling experiences with Cinematic AI.
Toonstar Uses AI To Power Hit YouTube Series Parker James’ YouTube show uses AI to achieve the speed, scale and efficiency that eludes the broader entertainment industry.
