Generative AI video is no longer just about typing a scene into a prompt box and waiting for a clip. The rise of Viggle shows a different path: users bring a motion reference, add a character image, and let the tool rebuild the movement with a new performer.
That workflow has turned Discord into an unexpected operating base for AI video. Viggle’s server grew from a few thousand members in March to 1.8 million by mid-May, then to over 3 million members with June just days away.
A meme made controllable AI video easy to understand
The clearest example began with Lil Yachty’s entrance at the Summer Smash Festival in 2021. A YouTube video of the moment has more than 6.5 million views, and its title calls the entrance “the HARDEST walk out EVER.”
In April, that clip became a meme template. People used AI to place other figures into the same walk-out, including Joaquin Phoenix’s Joker, NBA All-Star Joel Embiid, “Curb Your Enthusiasm” star Larry David and Sam Bankman-Fried.
The results were recognizable as artificial. The characters could move through the performance, but their bodies appeared unstable and their faces did not really react. Still, the clips were entertaining enough to spread widely, and they made Viggle’s core idea visible in a way a product demo might not have.
Instead of relying only on a text prompt, Viggle lets users upload two things on Discord: a video showing movement and a photo of a person. The system then generates a new video in which that person follows the motion from the reference clip.
Why Viggle is different from text-to-video tools
Text-to-video systems can create impressive scenes, but the source article highlights a practical limit: a written prompt may not carry enough detail to reproduce a specific viral performance. A phrase like “sam bankman-fried as lil yachty at the 2021 summer smash” does not necessarily tell a model exactly how the body should move.
Viggle founder Hang Chu described the company’s focus as “the controllable video generation model.” In his explanation, the goal is to give creators more control over how a character moves and how a scene appears, rather than asking text alone to specify every visual detail.
That control matters for more than memes. Chu said Viggle has two main groups of users:
- People creating internet memes and viral clips.
- Teams using the product during production work for game design and VFX.
For animation teams, the tool can turn concept designs into rough animation assets quickly. Chu said work that usually takes days, or even weeks, can be done instantly and automatically with Viggle, reducing tedious and repetitive modeling work.
Discord gives AI startups a ready-made platform
Viggle’s growth is tied closely to Discord. The startup is two years old and has just 15 employees, so it benefits from using an existing platform rather than building every community, interface and moderation layer itself.
That is also why Discord has become important to other generative AI products. Midjourney also runs on Discord, and its server has 20.3 million members, making it the largest single community on the platform. Discord overall has about 200 million monthly users.
For startups, the appeal is direct. Discord already has users who are comfortable with technical tools, shared servers and fast-moving online communities. It also provides infrastructure and content moderation tools that young companies would otherwise need to build or manage themselves.
Discord’s VP of Product Ben Shanken told TechCrunch that the company works with startups during major viral growth moments because they are not prepared for that kind of scale. He also said a major share of messages and usage on Discord now comes from Viggle and Midjourney.
For Viggle, that arrangement means the company can focus on its model as the back-end service while Discord handles much of the front-end infrastructure. Chu said this lets the team iterate faster.
Growth brings infrastructure and moderation pressure
Discord’s role in the generative AI boom is useful, but it is not cost-free. Hosting video is technically demanding, especially on a platform where people are also streaming live video games, video chatting and using voice calls.
Demand can also affect the user experience. The source article notes that Viggle’s growth had not shown signs of slowing, except that heavy demand for video generation had made wait times longer for impatient users.
Moderation is another challenge. Viggle can rely on Discord’s content guidelines and moderation apps, but a server with over 3 million members is hard to manage. The Lil Yachty walk-out memes also show the policy tension around synthetic media: Viggle’s own rules encourage users to avoid generating images of real people, including celebrities, without consent.
For now, the videos are not fully realistic. Their visible flaws make them easier to identify as AI-generated, which reduces some immediate confusion. But the same examples show why controllable generative AI video is drawing attention: even imperfect clips can move quickly across culture when the format is funny, familiar and easy to remix.
Viggle’s rise suggests that the next phase of generative AI video may be shaped as much by distribution as by model quality. Discord is not only a place where users gather; for companies like Viggle and Midjourney, it has become the foundation that lets an AI product find an audience before the startup has built a full platform of its own.