Search papers, labs, and topics across Lattice.
This paper introduces ParaScale, a novel module for transferring camera motion from a reference video to a target scene while addressing scale discrepancies that can lead to exaggerated or imperceptible motion. By defining the Parallax Number (Pi), a gauge-invariant descriptor, the authors establish that preserving this quantity is crucial for effective scale-faithful motion transfer. The results demonstrate that ParaScale significantly reduces Parallax Consistency Error (PCE) by over three times compared to uncalibrated methods, while maintaining visual fidelity across diverse scale regimes and backbone architectures.
Transferring camera motion across vastly different scales can now be done with over 3x improvement in accuracy without sacrificing visual quality.
Transferring the camera motion of a reference video to a freshly generated one lets creators reuse cinematic moves. Yet reference and target often live at incompatible scales -- a sweep across a galaxy versus a nudge across a desk -- and naively reusing the recovered trajectory yields either imperceptible or violently exaggerated motion. We trace this to a geometric fact: translation-induced image motion scales as ||T||/Z, so a monocular trajectory is meaningful only up to a depth-scale gauge. We distill this into the Parallax Number Pi = ||Delta T|| / Zbar, a dimensionless, gauge-invariant descriptor of how strongly a camera move is felt, and prove that it -- not the raw trajectory -- is the quantity that scale-faithful transfer must preserve. ParaScale is a plug-and-play module that reads Pi off any reference video and re-realizes it against the target scene's own depth, per frame, leaving rotation untouched. Sitting between pose extraction and pose injection, it requires no retraining and drops into any pose-conditioned generator. We further introduce the Parallax Consistency Error (PCE), a scale-symmetric metric that -- unlike the similarity-aligned TransErr -- exposes scene-scale mismatch. Across scale regimes spanning four orders of magnitude and multiple backbones, ParaScale keeps the realized parallax on the identity line and cuts PCE by more than 3x over uncalibrated transfer with no loss of visual fidelity.
