# SAC-NeRF: Adaptive Ray Sampling for Neural Radiance Fields via Soft Actor-Critic Reinforcement Learning

arXiv: https://arxiv.org/abs/2603.15622 · HTML: https://mlsanity.com/paper/2603.15622/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in RL with diffusion and flow models, one of 15 directions in Deep reinforcement learning methods.
- Builds on 16 of its 23 references that are in the map.
- First posted on arXiv 31 Dec 2025.

## Placed in

- Deep reinforcement learning methods / [RL with diffusion and flow models](https://mlsanity.com/deeprl/rl-with-diffusion-and-flow-models/)

## Builds on (most cited first, 16 of 16)

- [Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actor](https://mlsanity.com/paper/1801.01290/) (2018)
- [ScanNet: Richly-annotated 3D Reconstructions of Indoor Scenes](https://mlsanity.com/paper/1702.04405/) (2017)
- [DONeRF: Towards Real-Time Rendering of Compact Neural Radiance Fields using Depth Oracle Networks](https://mlsanity.com/paper/2103.03231/) (2021)
- [NerfAcc: Efficient Sampling Accelerates NeRFs](https://mlsanity.com/paper/2305.04966/) (2023)
- [FlashGS: Efficient 3D Gaussian Splatting for Large-scale and High-resolution Rendering](https://mlsanity.com/paper/2408.07967/) (2024)
- [AdaNeRF: Adaptive Sampling for Real-time Rendering of Neural Radiance Fields](https://mlsanity.com/paper/2207.10312/) (2022)
- [3D-HGS: 3D Half-Gaussian Splatting](https://mlsanity.com/paper/2406.02720/) (2024)
- [DashGaussian: Optimizing 3D Gaussian Splatting in 200 Seconds](https://mlsanity.com/paper/2503.18402/) (2025)
- [How Far Can We Compress Instant-NGP-Based NeRF?](https://mlsanity.com/paper/2406.04101/) (2024)
- [FrugalNeRF: Fast Convergence for Extreme Few-shot Novel View Synthesis without Learned Priors](https://mlsanity.com/paper/2410.16271/) (2024)
- [MF-NeRF: Memory Efficient NeRF with Mixed-Feature Hash Table](https://mlsanity.com/paper/2304.12587/) (2023)
- [NGP-RT: Fusing Multi-Level Hash Features with Lightweight Attention for Real-Time Novel View Synthesis](https://mlsanity.com/paper/2407.10482/) (2024)
- [A Probability-guided Sampler for Neural Implicit Surface Rendering](https://mlsanity.com/paper/2506.08619/) (2025)
- [Bayesian Soft Actor-Critic: A Directed Acyclic Strategy Graph Based Deep Reinforcement Learning](https://mlsanity.com/paper/2208.06033/) (2022)
- [Broad Critic Deep Actor Reinforcement Learning for Continuous Control](https://mlsanity.com/paper/2411.15806/) (2024)
- [Bidirectional Soft Actor-Critic: Leveraging Forward and Reverse KL Divergence for Efficient Reinforcement Learning](https://mlsanity.com/paper/2506.01639/) (2025)

## Abstract (arXiv)

Neural Radiance Fields (NeRF) have achieved photorealistic novel view synthesis but suffer from computational inefficiency due to dense ray sampling during volume rendering. We propose SAC-NeRF, a reinforcement learning framework that learns adaptive sampling policies using Soft Actor-Critic (SAC). Our method formulates sampling as a Markov Decision Process where an RL agent learns to allocate samples based on scene characteristics. We introduce three technical components: (1) a Gaussian mixture distribution color model providing uncertainty estimates, (2) a multi-component reward function balancing quality, efficiency, and consistency, and (3) a two-stage training strategy addressing environment non-stationarity. Experiments on Synthetic-NeRF and LLFF datasets show that SAC-NeRF reduces sampling points by 35-48% while maintaining rendering quality within 0.3-0.8 dB PSNR of dense sampling baselines. While the learned policy is scene-specific and the RL framework adds complexity compared to simpler heuristics, our work demonstrates that data-driven sampling strategies can discover effective patterns that would be difficult to hand-design.