Stability AI has unveiled Stable Virtual Camera, a new AI-powered model designed to transform 2D images into immersive videos with realistic depth and perspective.
The company claims this advancement will provide greater control and customization in digital filmmaking and animation.
Virtual cameras are commonly used in 3D animation and digital filmmaking to navigate and capture scenes dynamically. Stability AI’s latest innovation integrates generative AI into the process, allowing users to create novel views of a scene from one or more images—up to 32 in total—at specified camera angles.
The tool supports dynamic camera paths, including preset movements like “Spiral,” “Dolly Zoom,” “Move,” and “Pan.”
The research preview of Stable Virtual Camera offers videos in square (1:1), portrait (9:16), and landscape (16:9) aspect ratios, with a maximum length of 1,000 frames. However, Stability AI warns that certain conditions, such as human and animal images or complex textures like water, may result in lower-quality output.
The company also notes that ambiguous scenes, irregularly shaped objects, and complex camera paths intersecting surfaces can cause flickering artifacts, particularly when the target viewpoints significantly differ from the input images.
Stable Virtual Camera is currently available for research purposes under a noncommercial license and can be downloaded from the AI development platform Hugging Face.
The launch comes as Stability AI attempts to stabilize its business following recent financial struggles. Once a leader in AI-generated imagery with its flagship model Stable Diffusion, the company faced turmoil under its former CEO and co-founder Emad Mostaque. Reports of mismanagement led to staff resignations, a failed partnership with Canva, and investor concerns.
In an effort to regain footing, Stability AI has brought in a new CEO, appointed renowned filmmaker James Cameron to its board of directors, and introduced several new AI models.
Earlier this month, the company collaborated with chipmaker Arm to develop an AI model capable of generating audio, including sound effects, for mobile devices powered by Arm chips.
