Deep Compositional Denoising on Frame Sequences

Path tracing is the prevalent rendering algorithm in the animated movies and visual effects industry, thanks to its simplicity and ability to render physically plausible lighting effects. However, we must simulate millions of light paths before producing one final image, and error manifests as noise during rendering. Therefore we propose an 8-bit quantization scheme of the intermediate results, which halves the peak memory consumption. Our proposed quantization is carefully crafted to avoid introducing any quantization artifacts.

Path tracing is the prevalent rendering algorithm in the animated movies and visual effects industry, thanks to its simplicity and ability to render physically plausible lighting effects. However, we must simulate millions of light paths before producing one final image, and error manifests as noise during rendering. In fact, it can take tens or even hundreds of CPU hours on a modern computer to render a plausible frame in a recent animated movie. Movie production and the VFX industry rely on image-based denoising algorithms to ameliorate the rendering cost, which suppresses the noise due to rendering by reusing information in the neighborhood of the pixels both spatially and temporally. We propose an 8-bit quantization scheme of the intermediate results, which halves the peak memory consumption. Our proposed quantization is carefully crafted to avoid introducing any quantization artifacts. In summary, we propose a novel extension for processing frame sequences that provides context from neighboring frames while avoiding redundant computation and a novel quantization approach that reduces the memory overhead without introducing artifacts.