In the video, a roaring crowd fills a bustling summer music festival. As the beat drops over the speakers, the performer finally steps onto the stage: it’s the Joker. Donning his iconic red suit, green hair, and trademark face paint, the Joker raises his fist and dances energetically across the stage, even hopping down a runway to get closer to his sea of fans. When it’s time to start rapping, the Joker bends his knees and springs off the ground, bouncing up and down before executing a flawless 360-degree spin on one foot. The move looks effortless, yet attempting it yourself would likely result in a faceplant. The Joker has never been cooler.
In another video, NBA All-Star Joel Embiid emerges from backstage, greeting the crowd before flawlessly mimicking those same dance moves. Then, “Curb Your Enthusiasm” star Larry David takes the stage. However, in each instance, whether it’s the Joker, Joel Embiid, or Larry David, something appears slightly off – the performer’s body movements seem jittery, while their facial expressions remain static.
All of this is courtesy of AI technology developed by a company called Viggle.
The original video features rapper Lil Yachty taking the stage at the Summer Smash Festival in 2021, an entrance that a YouTube video with over 6.5 million views has labeled as “the HARDEST walk out EVER.” This iconic moment morphed into a trending meme format in April, as people began inserting their favorite celebrities – or notorious figures like Sam Bankman-Fried – into the footage of Lil Yachty’s grand entrance.
Text-to-video AI technology has seen significant advancements, yet it’s still not sophisticated enough to interpret highly specific prompts accurately. For example, typing “Sam Bankman-Fried as Lil Yachty at the 2021 Summer Smash” won’t yield precise results on platforms like Sora. This is where Viggle distinguishes itself.
Viggle operates uniquely on its Discord server, where users submit a video depicting some kind of movement—often a TikTok dance—and a photo of an individual. Viggle then generates a video of that person emulating the movements from the initial video. Although these videos are evidently artificial, they are quite entertaining. Viggle’s popularity surged following the viral success of the Lil Yachty meme, and this momentum continues.
“We are aiming to build what we call a controllable video generation model,” Hang Chu, Viggle’s founder, explained to TechCrunch. “When we create content, we aim to meticulously control character movements and scene aesthetics. Current tools focus predominantly on text-to-video, which lacks the capacity to capture all visual nuances.”
Chu elaborated that Viggle serves two primary user groups: meme creators and professionals in game design and visual effects production. “For instance, a team of animation engineers can transform concept designs into rough animation assets quickly. This helps in visualizing how they will look and feel in the final plan. Traditionally, this process can take days or even weeks to set up manually, but with Viggle, it can be done almost instantly, saving a lot of repetitive modeling work.”
From a few thousand members in March, Viggle’s Discord has expanded to over 3 million members by late May, surpassing the servers of popular games like Valorant and Genshin Impact combined.
Viggle’s exponential growth has led to longer wait times for video generation due to high demand. Given its strong reliance on Discord, Discord’s developer team has provided direct support to help Viggle manage its rapid expansion.
Discord’s experience with MidJourney—a community-driven AI with 20.3 million members—has prepared it to support startups like Viggle. Discord hosts around 200 million monthly users, many engaged with generative AI services like Viggle and MidJourney.
Ben Shanken, Discord’s VP of Product, noted, “No one is fully prepared for viral growth, so we collaborate with these startups during this stage. A significant portion of Discord’s current message traffic involves generative AI, and they demand robust support.”
For companies like Viggle and MidJourney, leveraging Discord allows them to focus on refining their AI models without needing to build a separate platform from scratch. Discord’s tech-savvy audience and built-in content moderation are particularly beneficial for a lean team like Viggle’s, which has just fifteen employees.
Chu highlighted the synergy between the two: “We can focus on the backend model development, while Discord’s infrastructure supports our frontend needs, enabling faster iteration.”
Before founding Viggle, Chu was an AI researcher at Autodesk and contributed to research for Facebook, Nvidia, and Google.
While hosting these AI-driven services is beneficial for growing user metrics, it does present challenges for Discord. The platform’s infrastructure must handle the heavy load of video generation, live gaming streams, video chats, and voice calls. Simultaneously, these startups benefit from the scalability and support that Discord provides.
Shanken emphasized Discord’s readiness, “Scaling any company is challenging, but Discord is structured for this kind of scale, allowing us to facilitate their absorption of growth effectively.”
Despite the advantages, managing content moderation for a large community remains challenging. For instance, Lil Yachty memes violate Viggle’s guidelines, which discourage generating images of real people without consent.
Currently, Viggle’s saving grace might be that its output isn’t entirely realistic. While the technology is impressive, the resulting animations, such as quirky Joker renditions, remain clearly distinguishable from reality, maintaining an element of humor and entertainment.
