Qvest has introduced V-Cropper, an NVIDIA-powered blueprint for turning horizontal sports and live video into vertical and other mobile formats. The system tracks subjects such as players, athletes and balls, then moves the crop to keep relevant action inside the narrower frame instead of applying a fixed centre cut.
The application uses NVIDIA NIM microservices for inference and NVIDIA Brev for reusable deployment configurations. Its containerised architecture can run locally or in cloud environments, allowing media organisations to keep content inside their chosen infrastructure. A labelling and tuning interface is intended to adapt the tracking behaviour to different sports and shot types rather than relying on one generic model.
Automatic reframing is technically useful when live moments must reach social platforms before their value fades. The hard problem is not producing a 9:16 canvas but deciding which moving subject carries the story as play develops, especially when the ball is small, briefly obscured or passes between players. Qvest plans to publish V-Cropper through NVIDIA Build as an open-source foundation. The licence, model-performance benchmarks, supported input formats and behaviour on difficult multicamera edits have not yet been detailed publicly.
