PTD augments a student video model with a training-only curvature head.
Polynomial student paths improve four-step video distillation quality
Parametric Trajectory Distillation trains video generators on adaptable curved paths, then removes the added curvature head for inference.
Big Tech
Lan Feng · Peter Karkus · Maximilian Igl · Julius Berner · Yuxiao Chen · Shuhan Tan · +3 more
EPFL · NVIDIA · Stanford University
Research Digest··3 min read
Feng and colleagues introduce Parametric Trajectory Distillation (PTD), a method for compressing iterative video generation into four model evaluations.
Why this paper
From NVIDIA and 2 others
In one line
Parametric Trajectory Distillation fits teacher trajectory segments to polynomials and supervises along the student's own predicted path, preserving motion and detail in four-step video generation.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ·No stated limitations found
- ✓Reports numbers on named benchmarks (2 benchmarks)
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§