AI video is moving beyond clip slop

These are Lowpass and Janko RoettgersA newsletter about the ever-changing technology and entertainment industry, created for the Seaside subscribers once a week.

Hollywood is cooked – or so the masses on social media want you to believe. Their evidence: AI-generated videos a Daniel Craig rides a Vespa through the city of Italy, Godzilla fights King Kongor The Avengers are marching through Manhattan.

In fact, cheap fare like this isn’t going to replace Hollywood blockbusters anytime soon. However, the new generation of AI video solutions can improve the way studios work. That’s because, until recently, AI companies tried to sell Hollywood on the same idea as those Twitter guys, with a little twist. Call, in short: AI video will allow anyone to make videos faster, cheaper, better – faster at the same time.

“The idea was: Put your camera in our videos,” said Luma AI CEO Amit Jain, whose company also developed studios. But when it started dealing with the entertainment industry, it received a crash course in how Hollywood works.

“It’s not enough to just make a movie,” says Jain now. “Why?” Video clips are usually 10 to 16 seconds long, says Jain: “It’s not just a recording. “Putting out short videos is not enough.”

Now, AI companies like Luma believe they have found the perfect way to sell Hollywood on AI. The bottom line? Don’t just use AI for video – use it for everything.

Luma has been working on AI assistants that can help with the entire production process. Jain compares this shift to how software development and AI have changed, with companies like Anthropic moving from simple vibe coding to agency workflows.

“It’s not enough to just create a little code,” says Jain. “We want these systems to work long-term, end-to-end. That’s what solves problems for people.” AI agents can do the same in Hollywood, he believes.

Luma is not alone in that approach. This week only, Google has been revealed a new version of the AI ​​media authoring platform Flow that also emphasizes end-to-end work on simple generation of graphics. “There’s this huge evolution going on in creative tools,” said Google Labs VP Elias Roman. “Going forward, they’re going to be like assistants.”

In the new Flow model, the assistant guides the user through a series of steps, from ideation to planning to creating characters to establish the desired look. And when it’s time to make a video, the assistant uses what he’s learned along the way to achieve the desired results without being told anything.

One issue that needs to be addressed is consistency. Generative AI has long struggled to make characters look the same from clip to clip. In the new version of Flow, users can quickly add a person who created a project just by pinning it, just like adding a friend to a Slack conversation.

The new generation of cinematographers also has a better understanding of the physics, time-frames, and languages ​​of cinema. Google’s Flow is powered by the company’s new Gemini Omni global model, while Luma has developed the Uni-1 as a collaborative model that doesn’t need too much complexity to understand the imagined world.

Luma recently partnered with Amazon to produce Ancient Stories: Moses, exclusive partner of MGM’s The house of David show. While shooting Mosestheater it will do in front of the LED walls showing the background created by Luma animations, while their clothes were also created by AI.

If the shot didn’t look right, it just took a new moment to create something new. Jain said: “This production can take six to eight weeks for each hour of television. Now, it takes them a week.”

Some studios are very receptive to this change. Netflix acquired Ben Affleck’s AI company InterPositive in March, is started his own AI studio the same month. Major Hollywood studios are already using Luma’s AI assistants, Jain says. He declined to name names, but the company has been publicly celebrating small wins: Luma recently announced the launch of a partnership. work and indie studio Wonder Project, which developed it Moses.

This will undoubtedly lead to job losses, although the extent is not yet known. If studios can make a TV show in one month instead of 10, they won’t be sending out nine months worth of checks. The benefits of AI that tend to improve is that this will lead to more. This could be a silver lining in Los Angeles in particular, what has been seen production days have declined significantly in recent years.

Now, we’ll just have to see if Hollywood uses this technology for things people want to watch.