Reproduce “FL4N: Feed-forward Latent Model for 4D Novel View Synthesis” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Eunbeen Hong, Hyeonseo Yu, Honggyu An, Jooho Son, Hyunah Ko, Soowon Son, Chaehyun Kim, Jahyeok Koo and 3 more
FL4N renders dynamic scenes from requested viewpoints and times by freezing a pretrained static encoder and renderer, then adding a lightweight time-conditioned residual. On ADT novel-time synthesis, it reports 19.693 dB PSNR, 0.5948 SSIM, and 0.4490 LPIPS, compared with lower C4G scores. C4G’s poses were aligned using target images, so this comparison applies only to the stated evaluation protocol.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.