Higgsfield is touting the latest update as a fix for one of the most persistent problems with generated video: consistency.
On Thursday, the startup released Cinema Studio 2.0, an update to its AI video platform aimed at creators and advertising teams looking to turn quick prompts into something closer to commercial-grade storytelling.
The new version focuses on controlling areas where many generative video tools still fall short, such as camera movement, scene composition, and narrative flow. Among the added features is a feature called “What’s next” where AI can suggest how a scene should progress, allowing creators to iterate on visuals and narrative at the same time.
Higgsfield CEO Alex Mashrabov likens Higgsfield to a creative co-pilot. “[AI] It helps complete the story and tell what happens next,” he said.
The update also adds more granular controls, allowing creators to make subtle camera movements and adjustments while preserving scene integrity, and is intended to mimic the way cinematographers work on set, rather than the way AI models typically produce output.
Mashrabov said the update aims to address the persistent challenge of keeping characters and environments consistent even when angles, lighting, and framing change. “The core problem with any model on the market is that [is] If someone tries to show the same scene or the same character from a different angle, the identity becomes distorted,” he said.
Higgsfield has been beta testing Cinema Studio 2.0 since October with about 100 external creators. The platform is available through a freemium model, with paid plans ranging from $9 to $250 per month for teams managing multiple projects.
Founded in 2023 by Mashrabov, who previously led generative AI efforts at Snap, Higgsfield quickly scaled to meet demand. The startup, valued at $1.3 billion, counts approximately 15 million creators and agencies on its platform and generates more than $250 million in annual recurring revenue, ADWEEK previously reported.
But as creators and marketers increasingly rely on AI tools for production, some say they’re reaching a limit. Complaints range from slow generation to inconsistent output to limitations of paid plans, highlighting the gap between experimentation and reliable commercial use.
