The rapid pace of innovation over the last year in the AI space has been staggering, and video-to-video generative AI is one of the areas which has come into the spotlight as the next frontier. The latest version of the Videoleap app, developed by Lightricks, showcases some jaw-dropping capabilities, pushing the mobile app industry to new standards.
Video-to-video generative AI technology allows users to transform the style of a video they record or upload, so it becomes something completely new. Similar to other forms of visual generative AI, this is often done through a prompt such as “80s punk” or “Japanese manga.”
Using Videoleap, videos can also be transformed based on an uploaded image or style presets such as “Abandoned” (applies a post-apocalyptic look, with decayed walls and vine overgrowth) or “Fire” (adds flickering flames in a hand-drawn anime style).
Algorithms That Move
A year ago, text-to-image generators were frequently the source of ridicule, as they were plagued with “hallucination” issues, but their stability has increased to the point where they are now in common use.
According to a recent study published by Lightricks in partnership with YouGov, about 92% of creators are already strongly familiar with generative AI, and 62% are already using it.
The same study found that among creators who are concerned about widespread adoption of AI, the most common worry is the potential for the proliferation of manipulative deep fakes. While video-to-video AI can’t yet convincingly turn everyday users into Tom Cruise doing magic tricks, for example, the future definitely has the potential for nefarious use cases.
“The genie is out of the bottle,” Lightricks co-founder and CEO Zeev Farbman told a reporter last week. “A year ago, state agents were able to create any kind of deep fakes they wanted. What’s changed now is that we’re finally getting public awareness of the fact that you should be extremely critical of every visual tab that you’re seeing. Making the technology more widely available is the only way to tackle this issue.”
Video-to-video generators, after all, are in this emergent phase still, as the underlying algorithms are being refined and improved.
Today, the crucial bottleneck is the amount of server-side resources needed to generate each frame of a video, which means users are often left waiting for the finished product much longer than they’d be used to from other apps.
A Dynamic, Competitive Space
Signs of progress can already be seen amongst the offerings currently on the market. There are several players who have healthy competition between them, and all have their pros and cons. Lightricks was the latest to enter with the “AI Effects” functionality on their Videoleap app, whereas Picsart already offers a limited video-to-video tool.
Both are intuitive to use but imposes a limit on video length. Picart allows just three seconds, Lightricks allows ten seconds. This restricts the potential for current use case viability but does let users get a taste for the technology. With many players in the visual mobile tool space working hard to keep up with each other, it’s unlikely to be a long wait for the cap to extend beyond the typical length of short-form content.
In Videoleap, Lightricks has pushed the boundary slightly further than the others, with the ability to edit the AI generated image directly on-device. This takes only 30 seconds to render, which is 20 times faster than other industry attempts so far, according to assessments shared by Lightricks’s publicity team. They’ve also included a mobile “AI Infinite Zoom” feature, which allows users to use “outpainting” for expanding the boundaries of their videos based on input.
It will be fascinating to see who takes the lead in the market as the demand increases. Each company has a slightly different focus, which could play into which users they attract. Runway, another player in the space, has its own AI studio, and they may push toward the premium end of the market for filmmakers for instance. Picsart is primarily focused on creatives, with a relatively budget suite of tools.
Lightricks’s apps are widely adopted by a large swath of the creator economy, with 30 million users using the company’s various products in any given month. This means that Lightricks is likely to capture a huge chunk of the demand for video-to-video AI tools, as Videoleap users are already plugged into a vast creator ecosystem.
What’s Next for AI Video
The video-to-video generative AI space is just starting to heat up. At the moment, it’s a time to experiment with the different features, but soon they could be widely used by both businesses and consumers alike.
Comments are closed.