Read the original at HF Daily Papers
Researchers introduce NAMVIS, a diffusion-free framework that generates multi-view images using autoregression and projective pose encoding, running over three times faster than diffusion baselines.
Carried by: HF Daily Papers. First seen: .