PrismMirror is the first monocular human frontal view synthesis model to achieve real-time inference (24 FPS) without external geometric models.
March 17, 2026
Original Paper
Real-Time Human Frontal View Synthesis from a Single Image
arXiv · 2603.15433
The Takeaway
It replaces heavy auxiliary models with a lightweight linear attention distillation framework and coarse-to-fine learning. This makes high-fidelity 3D telepresence feasible on standard hardware from a single camera feed.
From the abstract
Photorealistic human novel view synthesis from a single image is crucial for democratizing immersive 3D telepresence, eliminating the need for complex multi-camera setups. However, current rendering-centric methods prioritize visual fidelity over explicit geometric understanding and struggle with intricate regions like faces and hands, leading to temporal instability. Meanwhile, human-centric frameworks suffer from memory bottlenecks since they typically rely on an auxiliary model to provide inf