Alibaba’s Qwen-Image Gets Virtual Camera Control for Instant Angle Generation
Alibaba’s research team has unveiled a breakthrough feature for Qwen-Image-Edit — Camera Angle Control, an AI system that generates entirely new object perspectives from a single static image.
From still images to cinematic motion
The new system allows Qwen-Image to simulate 180-degree rotations, close-ups, and wide-angle shots using only textual commands or adjustable sliders. Each render preserves the original lighting, textures, and object geometry while recomposing the scene from a different viewpoint — as if moving an invisible virtual camera around the subject.
The result: high-fidelity, photorealistic angles without the need for 3D data. Artists can now recreate studio-level shots or generate full visual storyboards using a single image reference.
Speed through Lightning LoRA
The technology is powered by a custom LoRA adapter called dx8152, optimized for Lightning mode inference. It completes generation in just eight diffusion steps instead of the usual forty to fifty — reducing rendering time to under one second on modern GPUs. This speed makes interactive workflows finally viable for AI-driven image editing and simulation.
Applications across design and simulation
Beyond creative arts, Qwen-Image’s virtual camera has immediate potential in product design, cinematography, and 3D modeling. With only one reference photo, users can produce complete view sets for prototypes, marketing assets, or training datasets. The feature transforms how developers and designers approach multi-angle visualization — compressing hours of manual work into moments of computation.
Editorial Team — CoinBotLab