Achieving global illumination is crucial for delivering realistic lighting in real-time rendering applications. Despite recent advancements in hardware raytracing, the computational demands of full path tracing remai...
详细信息
Achieving global illumination is crucial for delivering realistic lighting in real-time rendering applications. Despite recent advancements in hardware raytracing, the computational demands of full path tracing remain largely impractical for real-world production scenarios. We introduce a novel two-level radiance caching system that exclusively utilizes on-surface caches, diverging from conventional approaches that combine screen-space and world-space caches. Unlike previous texture space techniques, which mostly prioritize closely matching the resolution to screen space to minimize artifacts, our focus is on achieving optimal visual quality with minimal texture space resolutions. By caching directional radiance information on both primary and secondary hits, our approach delivers high-quality renderings of global illumination while being computationally efficient. Overall, this leads to an up to 5% to 10% improvement in both speed and quality compared to other state-of-the-art approaches. Our proposed method is versatile, handling not only diffuse global illumination but also addressing (glossy) reflections. Furthermore, our approach is well-suited for multi-viewer rendering, as the utilization of on-surface caches enables information sharing among different viewers, making it applicable to cloud-native rendering environments.
We present a kernel-predicting neural denoising method for path-traced deep-Z images that facilitates their usage in animation and visual effects production. Deep-Z images provide enhanced flexibility during compositi...
详细信息
We present a kernel-predicting neural denoising method for path-traced deep-Z images that facilitates their usage in animation and visual effects production. Deep-Z images provide enhanced flexibility during compositing as they contain color, opacity, and other rendered data at multiple depth-resolved bins within each pixel. However, they are subject to noise, and rendering until convergence is prohibitively expensive. The current state of the art in deep-Z denoising yields objectionable artifacts, and current neural denoising methods are incapable of handling the variable number of depth bins in deep-Z images. Our method extends kernel-predicting convolutional neural networks to address the challenges stemming from denoising deep-Z images. We propose a hybrid reconstruction architecture that combines the depth-resolved reconstruction at each bin with the flattened reconstruction at the pixel level. Moreover, we propose depth-aware neighbor indexing of the depth-resolved inputs to the convolution and denoising kernel application operators, which reduces artifacts caused by depth misalignment present in deep-Z images. We evaluate our method on a production-quality deep-Z dataset, demonstrating significant improvements in denoising quality and performance compared to the current state-of-the-art deep-Z denoiser. By addressing the significant challenge of the cost associated with rendering path-traced deep-Z images, we believe that our approach will pave the way for broader adoption of deep-Z workflows in future productions.
暂无评论