Stability AI has released a new generative AI model called Stable Virtual Camera, which can transform two-dimensional images into videos with realistic depth and perspective. According to the company’s announcement, this model allows users to create new scene views from one or several images (up to thirty-two) and select camera angles. Stable Virtual Camera can generate videos that move along dynamic camera trajectories or use presets such as “Spiral,” “Dolly Zoom,” “Move,” and “Panorama.”
Currently, the model is available as a research version, enabling video creation in square (1:1), portrait (9:16), and landscape (16:9) formats, with a length of up to one thousand frames. However, Stability warns that in certain situations, the output quality may be lower, especially if the images contain people, animals, or “dynamic textures” such as water. The company notes that complex camera trajectories or irregular objects may cause flickering artifacts.
Stable Virtual Camera is available for research use under a non-commercial license and can be downloaded from the AI development platform Hugging Face. This step is part of Stability AI’s efforts to regain its market position after financial difficulties that led to a change in management and new investments. In recent months, Stability AI has hired a new CEO, added renowned director James Cameron to its board of directors, and released several new image generation models.
