Indie pop artist Washed Out's new song “The Hardest Part” is about lost love, among the most human themes.
But ironically, to explain the song's sense of longing, the musicians turned to something far less flesh and blood: artificial intelligence.
With Thursday's release of “The Hardest Part,” Macon, Ga.-based Washed Out (real name Ernest Green), a music video using OpenAI's Sora text-to-video technology, has joined forces with major music artists and movies. This was our first collaboration with a producer. According to the singer-songwriter's record label, Sub Pop.
The nearly four-minute video, directed by Paul Trillo, gives viewers a quick zoom into key elements of the couple's lives. The audience sees the characters – a red-haired woman and a dark-haired man – go from flirting and smoking in high school in the 1980s to getting married and having children. “Don't cry, it's okay now,” Green exclaims. “The most difficult thing is that there is no going back.”
The couple isn't played by real actors. These are completely digitally created by Sora's AI.
This video could be the start of a potentially groundbreaking trend in using AI in video production.
“I think where we are right now is just on the verge of exploding. So I'm looking forward to being able to incorporate some of this new technology and seeing what it comes up with for me,” Green said. he said in an interview. So if it's something pioneering, I'd love to be a part of it. ”
“The Hardest Part” is the lead single from Green's new self-produced album “Notes From a Quest Life,” scheduled for release on June 28th, and is the longest music video ever made with Sora technology. . This program creates short clips based on written text prompts. This allowed Trillo to construct scenes in ways that would be many times more expensive using real actors, sets, and locations.
“We didn't have any budget constraints or the need to travel to different places, so we were able to explore different endings in this couple's lives,” Trillo said.
read more:AI companies are courting Hollywood. Will they come in peace?
Trillo is one of the creators who has early access to Sora, which is not yet available to the public. OpenAI announced Sora in February and has been testing the system with directors and meeting with Hollywood executives and producers. Trying to solve problems and address intellectual property issues.
AI innovations have been highly controversial in a variety of sectors, including the music industry, which has been plagued by the use of video and audio “deepfakes,” which falsely use an artist's image or voice. Musicians and others are pushing for legislation to combat such misleading creations, and talent agencies are working with tech startups to crack down on unauthorized digital copying.
The introduction of Sora, from the same company that created the text-based AI model ChatGPT, sparked concerns in Hollywood and elsewhere that it could have a devastating impact on jobs and production. . Still, it sparked excitement among some creators about how they could bring their visions to life on screen without being constrained by special effects budgets or travel restrictions.
Both Green and Trillo said you can do more with Sora on a budget than with a physical set. Sub Pop does not disclose how much it costs to produce the video. This music artist does not pay OpenAI any fees to use its technology in his music videos.
The pair had explored other ideas, such as hiring dancers and filming in locations similar to the green hills featured in the artwork for Green's new album, but time and financial constraints made this difficult. There was found. Trillo then proposed using Sora for an experiment.
read more:Dead Beatles, fake Drakes, and robot songwriters: Inside the panic over AI music
Green, who music TV viewers may recognize from the theme song for the satirical sketch comedy show “Portlandia,” was hesitant at first.
“I feel like most of my music and the videos I've made over the years always start from a really emotional and honest place,” Green said, adding that the AI videos he's seen so far He noted that there are many examples of the frightening “uncanny valley” in which humans appear human but are eerily artificial.
read more:After the launch of OpenAI's Sora, can Hollywood adapt to the “matrix” of artificial intelligence?
Nevertheless, Green was willing to experiment. So Trillo tried out different concepts to see what would work for video. Using this technology, we can now explore all the different outcomes of a couple's life across multiple locations by creating elaborate text-based prompts. He compiled about 55 clips of the roughly 700 videos he generated using Sora and completed the video in about six weeks.
“This eliminates the need for me to edit it myself,” Trillo said. “Because we were able to really try things out, we were able to throw a lot of things against the wall and see what stuck, and that organically gave rise to a different kind of story. is completed.”
To produce a usable clip, Greene had to create a prompt that included enough specific details about the image itself, as well as the angle of the shot and the movement of the characters. “We zoom in on the bubble popping, zoom on the bubblegum, and enter an open soccer field,” Trillo writes as part of the video's short snippet prompt. “The scene is moving rapidly, showing a frontal perspective, and you can see the students getting bigger and faster.”
The final music video for “The Hardest Part” shows several locations, including a high school, a grocery store, a hilly area, a hallway with billowing white sheets, and a wall of flames.
There were some limitations. Sometimes Trillo would come up with an idea and Sora would make it happen. It can also create something chaotic and unusable. The video contained inconsistencies, which Trillo sometimes ignored. The characters vary slightly from clip to clip, but so do the couple's children.
Part of the video's artistic appeal is its dream-like state, a recollection of the couple's life that explains the fuzziness of human memory.
Trillo said of Sora, “You need to know where to fight.” “To tackle this, you have to surrender a little bit of your free will and you have to accept the nature of how chaotic it is.”
“I was certainly amazed at how far he could go in putting together a story,” Green said.
read more:AI is here and making movies. Are you ready for Hollywood?
Greene and Trillo said they believe AI has the potential to give people more possibilities to advance the music video art form. Music videos are a logical medium to experiment with AI because they are typically short and cost much less to produce than feature films or TV episodes.
But for Trillo, it's important that this isn't used as a new creative method, but as another tool in the toolbelt.
“A lot of music videos don't have the budget to dream big,” Trillo says. “I think AI can help the music industry in terms of creating things that even Ernest would dream of, and perhaps didn't have the courage to dream of before.”
Sign up for the Wide Shot newsletter to get the latest entertainment business news, analysis and insight.
This story originally appeared in the Los Angeles Times.
