Abstract:
Not knowing where to put your hands when taking pictures and having a stiff posture when facing the camera are common pain points for many ordinary users. In response to this daily social trouble, Superpose Labs, a start-up founded by former TikTok core executives and computer vision experts, recently officially launched its first AI photography application, Superpose. The application is positioned as a "personal AI photography coach in your pocket" and aims to use end-to-end vision and real-time AI interaction to help users and photographers find the most natural body movements and composition postures before pressing the shutter.
Superpose’s core R&D team has a deep background in technology and social products. Its founding team includes the former product leader who led the launch of TikTok Creator Marketplace, and the former TikTok technical director who co-founded Bellus3D (later acquired by Meta) and holds dozens of computer vision patents. The team has previously received early investment support from Khosla Ventures, a well-known venture capital institution in Silicon Valley. The founder revealed that the inspiration for this project came from high-frequency embarrassing scenes in life: when traveling, dating or in daily life, ordinary people often find it difficult to clearly express the desired angle to their companions, and the subjects often take stiff and failed photos because they do not know how to pose.
Unlike most photography tools currently on the market that rely on post-shooting retouching, filter beautification, or AI-generated redrawing, Superpose places the intervention node in front of the moment the shooting occurs. When the user composes the scene, Superpose uses the camera to scan the surrounding environment, light and character outlines in real time, and uses AI algorithms to instantly recommend four best posture references that fit the atmosphere of the current scene. The shooting interface will directly project a lightweight skeleton or posture guide outline (overlay) in the viewfinder. The person being photographed only needs to adjust the body position and body angle according to the on-screen instructions to easily complete a natural and aesthetically pleasing shot.
This real-time guidance mode not only reduces the subject's lens anxiety, but also greatly liberates the "tool man" shooting companions. Through the visual positioning guidance of what you see is what you get, photographers do not need to have professional photography composition skills. They only need to align the characters in the shot with the dotted line to press the shutter, effectively reducing the cost of repeated reshoots and communication. In addition, the application also combines scene recognition algorithms to dynamically adjust the action recommendation style according to different backgrounds such as indoors, streets or natural scenery where the user is located.
At present, the application has been officially launched on iOS, and will continue to expand more scene-based gesture models and real-time interaction features in the future.
Comments