Move over, Sora – Higgsfield introduces a cutting-edge AI video generator with advanced camera controls

by

in

1. Higgsfield is a new AI video generation platform funded by Menlo Ventures, focused on camera and motion control, particularly in human movement.
2. Higgsfield offers personalized and realistic human characters and motion, using a combined transformer and diffusion model for smooth and coherent video output.
3. Higgsfield is slowly rolling out access, starting with content creators and offering an app called Diffuse for iPhone users in the U.S. to create and share AI-edited videos.

Higgsfield is a new AI video generation platform funded by Menlo Ventures, focusing on providing advanced camera and motion control, particularly in human movement. It can generate clips up to 10 seconds, with a focus on realistic human characters and motion. The combination of transformer and diffusion models allows Higgsfield to output smooth and realistic videos, similar to OpenAI’s Sora engine. The startup is slowly rolling out access to a small group of content creators before wider availability.

The key differentiator of Higgsfield AI is its unparalleled personalization and control, offering realistic human motion that sets it apart from other AI video models. While existing models like Stable Video Diffusion use diffusion technology, Higgsfield combines transformer and diffusion models for smoother and more coherent sequences. This combined architecture allows for the creation of “world models” that simulate the physical world, producing longer, smoother, and more professional-quality content.

The startup aims to democratize social video creation with its foundational model that creates personalized characters with lifelike motion from just a single selfie, all on mobile devices. Higgsfield is expected to offer a studio-grade video marketing platform for creators and enterprises in the future, in addition to consumer products like the Diffuse app for AI-edited videos. Though access is currently limited, interested users can download the Diffuse app in the U.S. to create and share short AI-edited videos using the Higgsfield model.

Source link