OpenSuperSampling

Public training dashboard

OpenSuperSampling is training a unified game reconstruction pipeline for super-resolution and frame extrapolation from one temporal model.

GitHub repo updated just now loading
training status loading

v6.1 Pico

waiting for metrics.json

Latest measured result
25.703 dB

v5-pixel-temporal, LPIPS 0.1666, temporal ratio 0.337x

GPU usage
loading GPU status

Run history

Active training first, then measured and superseded references.

Architecture memo

The canonical v6 design is the source of truth for the persistent Gaussian canvas, covariance-resampled rasterizer, cross-attention fusion path, and OSS-FX frame extrapolation plan.

Open canonical memo