Divide and Conquer: Rethinking the Training Paradigm of Neural Radiance Fields
This work addresses a specific bottleneck in NeRF training for 3D scene synthesis, offering an incremental improvement over existing methods.
The paper tackles the problem of suboptimal rendering quality in Neural Radiance Fields (NeRFs) due to equal importance given to all training images, and results in enhanced rendering quality on datasets like NeRF synthetic and Tanks&Temples, with convergence to a superior minimum.
Neural radiance fields (NeRFs) have exhibited potential in synthesizing high-fidelity views of 3D scenes but the standard training paradigm of NeRF presupposes an equal importance for each image in the training set. This assumption poses a significant challenge for rendering specific views presenting intricate geometries, thereby resulting in suboptimal performance. In this paper, we take a closer look at the implications of the current training paradigm and redesign this for more superior rendering quality by NeRFs. Dividing input views into multiple groups based on their visual similarities and training individual models on each of these groups enables each model to specialize on specific regions without sacrificing speed or efficiency. Subsequently, the knowledge of these specialized models is aggregated into a single entity via a teacher-student distillation paradigm, enabling spatial efficiency for online render-ing. Empirically, we evaluate our novel training framework on two publicly available datasets, namely NeRF synthetic and Tanks&Temples. Our evaluation demonstrates that our DaC training pipeline enhances the rendering quality of a state-of-the-art baseline model while exhibiting convergence to a superior minimum.