TL;DR
This paper introduces a multiresolution hash encoding for neural graphics primitives that significantly accelerates training and rendering, enabling real-time applications with high-quality results.
Contribution
It proposes a novel hash-based input encoding that reduces neural network size and computational cost while maintaining high quality, facilitating fast training and rendering.
Findings
Achieves several orders of magnitude speedup in training and rendering.
Enables training of high-quality neural graphics primitives in seconds.
Supports real-time rendering at 1080p resolution.
Abstract
Neural graphics primitives, parameterized by fully connected neural networks, can be costly to train and evaluate. We reduce this cost with a versatile new input encoding that permits the use of a smaller network without sacrificing quality, thus significantly reducing the number of floating point and memory access operations: a small neural network is augmented by a multiresolution hash table of trainable feature vectors whose values are optimized through stochastic gradient descent. The multiresolution structure allows the network to disambiguate hash collisions, making for a simple architecture that is trivial to parallelize on modern GPUs. We leverage this parallelism by implementing the whole system using fully-fused CUDA kernels with a focus on minimizing wasted bandwidth and compute operations. We achieve a combined speedup of several orders of magnitude, enabling training of…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Code & Models
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
Methods*Communicated@Fast*How Do I Communicate to Expedia? · Exponential Decay · Dense Connections · Feedforward Network · Absolute Position Encodings · Adam · Robinhood Customer Care Number +1-833-534-1729
