Reproduce “TTB: Test-time Baking for Efficient Rendering of Decoder-only View Synthesis Models” 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

Chin-Yang Lin, Ryo Hachiuma, Jaesung Choe, Min-Hung Chen, Yu-Chiang Frank Wang, Yen-Yu Lin, Wei-Chen Chiu, Yu-Lun Liu and 1 more

Test-Time Baking (TTB) renders novel views of photographed scenes by adapting an MLP to reproduce cross-view attention outputs during reconstruction. Context pairs align image-bearing and pose-only tokens; the baked MLP supplies features for new camera queries. On 140 DL3DV test scenes, with 64 input views, it renders at 23 FPS versus DVSM’s 1, scoring 30.20 versus 31.11 dB PSNR; reconstruction takes longer.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.