Stability AI has released a new AI model, a stable virtual camera that the company claims they can transform 2D footage into “immersive” videos with realistic depth and perspective.
Virtual cameras are tools often used in digital filming and 3D animation to capture and navigate scenes in real time. With a stable virtual camera, stability intended to add generative AI to the mix to deliver greater control and customizability, the company said in a Blog post.
A stable virtual camera generates “new views” of a scene of one or more images (up to 32 overall) at camera angles that a user specifies. The model can generate videos that travel along “dynamic” room paths or presets including “spiral”, “Dolly Zoom”, “Move” and “Pan.”
The current version of a stable virtual camera, research forecast, can generate videos in a square (1: 1), a portrait (9:16), and a landscape (16: 9) appearance proportions up to 1,000 frames in length. Stability warns that the model can produce less high quality results in some scenes, yet especially with images with humans, animals or “dynamic textures” such as water.
“Very ambiguous scenes, complex room pathways that intersect objects or surfaces, and irregularly shaped objects can cause lightning artifacts,” stability notes on its blog, “especially when target views differ significantly from the input images.”
A stable virtual camera is available for research use under non -commercial license. It can be downloaded from the Ai Dev Hugging Face platform.
Stability, the troubled company behind the popular image -generation model Stable dissemination,, raised new money last year As investors including Eric Schmidt and Napster founder Sean Parker sought to turn the business. Emad Mostaque, the co -founder and former CEO of Stability, reportedly misused stability into financial ruin, leading personnel to resign, a partnership with Canva to fall, and investors will grow worried about the company’s prospects.
In recent months, stability has hired a new CEO, appointed Titanic director James Cameron to his board, and released several new image -generation models. Earlier in March, the company joined with a ragged arm To bring AI model, which can generate sound including sound effects to mobile devices running arms.
