tttLRM: Test-Time Training for Long Context and Autoregressive 3D Reconstruction

TL;DR

tttLRM introduces a test-time training approach for large-scale, autoregressive 3D reconstruction that efficiently integrates multiple observations into a compact implicit representation, enabling high-quality, scalable 3D modeling.

Contribution

The paper presents tttLRM, a novel model combining test-time training with autoregressive 3D reconstruction, scalable to long contexts with linear complexity and improved performance.

Findings

Outperforms state-of-the-art in 3D Gaussian reconstruction

Supports progressive reconstruction from streaming data

Achieves faster convergence and higher quality results

Abstract

We propose tttLRM, a novel large 3D reconstruction model that leverages a Test-Time Training (TTT) layer to enable long-context, autoregressive 3D reconstruction with linear computational complexity, further scaling the model's capability. Our framework efficiently compresses multiple image observations into the fast weights of the TTT layer, forming an implicit 3D representation in the latent space that can be decoded into various explicit formats, such as Gaussian Splats (GS) for downstream applications. The online learning variant of our model supports progressive 3D reconstruction and refinement from streaming observations. We demonstrate that pretraining on novel view synthesis tasks effectively transfers to explicit 3D modeling, resulting in improved reconstruction quality and faster convergence. Extensive experiments show that our method achieves superior performance in feedforward 3D Gaussian reconstruction compared to state-of-the-art approaches on both objects and scenes.