← all work 3d-computer-vision

Mesh or novel view

A hash-grid NeRF, Instant-NGP's encoding reimplemented in plain PyTorch, trained from

Result

The finding

Nearly all of the model is the lookup table, and it memorises

The MLP has 9,555 parameters. The hash table has 16.78 million — 1,756× more. So this is not a network that learned a scene so much as a lookup structure that stored one, with a tiny decoder on top.

The train/test gap shows what that costs: **33.592 dB on training views against 28.235 on held-out ones, a gap of 5.357 dB**. Rendering from a pose it was fitted on is markedly easier than rendering from one it was not, which is exactly what a memorising representation predicts and what a single training-view render would have concealed.

The per-view spread matters as much as the mean. The worst held-out view scores 18.933 against a best of 34.242 — a range of 15.3 dB on the same trained model. A single averaged PSNR hides that some viewpoints are reconstructed far worse than others.

Elsewhere

Code and full results on GitHub

Every number is read from the file the run produced. github.com/hammasbuilds · kaggle.com/muhammadhammas13