flowiz for papers

flowiz is built to make optical-flow figures reproducible and consistent with prior work.

Match the literature

The color wheel is bit-compatible with flow_vis, the de-facto standard used by RAFT and successors. Your visualizations line up with published figures without tuning.

One-call comparison figures

import flowiz as fz

pred = fz.from_tensor(model(images))
gt   = fz.read("sintel/flow/frame_0001.flo")
fz.compare_grid(pred, gt, save="fig3_sintel.png")

Consistent qualitative videos

flowiz video 'results/*.flo' -o supp/qualitative.mp4 -r 24

Sequence-wide normalization keeps colors stable across frames so reviewers see motion, not flicker.

Citing flowiz

A citation is appreciated (requested, not license-required):

@software{gach_flowiz,
  author  = {Giorgi Gachechiladze},
  title   = {flowiz: the optical flow visualization toolkit},
  url      = {https://github.com/georgegach/flowiz},
  version = {3.0.0},
  year    = {2026}
}

The machine-readable CITATION.cff lets GitHub render a "Cite this repository" button. A Zenodo DOI is minted per release.