Adaptive Photon Mapping Based on Gradient

Recently, deep learning-based approaches have led to dramatic improvements for Monte Carlo rendering at the low sampling rate. Most of these approaches are aimed at path tracing. However, they are not suitable for photon mapping. In this paper, we develop a novel accelerate stochastic progressive photon mapping approaches with neural network. First, our framework utilizes the particle-based rendering and focuses on photon density estimation. We train a neural network to predict a kernel function to aggregate photon contributions at shading point. Then we construct a estimation images with the prediction network. During experiments, we could find that there are spike pixels and noises in estimation images sometimes. So we present the improved denoising network to post-process the estimation images. Finally, we can obtain the high-quality reconstructions of complex global illumination effects like caustics with an order of magnitude fewer photons compared with previous photon mapping methods. Besides, our denoising network can reduce most multi-scale noises on both low-frequency and high-frequency areas while preserving more illumination details, especially caustics, compared with other state-of-the-art learning-based denoising methods.

show abstract

Novel accelerated Stochastic Progressive Photon Mapping rendering with neural network

Xing

Chen

2021

J. Phys.: Conf. Ser.

View full text Add to dashboard Cite

show abstract

Deep Kernel Density Estimation for Photon Mapping

Zhu

Jensen

et al. 2020

Computer Graphics Forum

View full text Add to dashboard Cite

Figure 1: We present a novel learning-based photon mapping (PM) method that can be used to synthesize photorealistic images (f) with detailed caustics (shown and compared in the insets) from very sparse photons for scenes with complex diffuse-specular interactions. In particular, we use our method with only 15k photons (∼0.06 photons per pixel) to compute accurate global illumination for light-specular paths. We use path tracing (PT) with a moderate number (300) of samples per pixel (spp) to compute the other paths and apply the Optix learning-based denoiser (based on [CKS * 17]) to remove the Monte Carlo (MC) noise. In contrast, pure PT leads to noisy results lacking focused caustics (a) even with 1000 spp that is significantly more than our photon and path samples. While this noise can be mitigated using a learning-based denoiser, this introduces artifacts and cannot recover the caustics (b). Combining PT and standard PM [Jen96] with 15k photons, and then denoising (c), avoids these artifacts but still does not reconstruct caustics accurately from such low photon counts. While providing 1.5M photons (this is 100 times the number of photons our method uses) and applying the advanced stochastic progressive PM (SPPM) [HJJ10] enables a more accurate result (d), it is still slightly worse than ours. In contrast, our result (f) accurately reproduces the caustic effects in the global illumination, as compared to the ground truth (g), with significantly fewer samples. Ours is comparable with (if not better than) the result from adaptive progressive PM (APPM) [KD13] with 100 times the number of photons (e).

show abstract

scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.

Contact Info

customersupport@researchsolutions.com

10624 S. Eastern Ave., Ste. A-614

Henderson, NV 89052, USA

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.

Blog Terms and Conditions API Terms Privacy Policy Contact Cookie Preferences Do Not Sell or Share My Personal Information

Made with 💙 for researchers

Part of the Research Solutions Family.

Adaptive Photon Mapping Based on Gradient

Cited by 2 publications

References 13 publications

Novel accelerated Stochastic Progressive Photon Mapping rendering with neural network

Novel accelerated Stochastic Progressive Photon Mapping rendering with neural network

Deep Kernel Density Estimation for Photon Mapping

Contact Info

Product

Resources

About