neat video optimised for cleaning cgi renders?

suggest a way to improve Neat Video
Post Reply
matt_ins
Posts: 1
Joined: Sat Aug 31, 2013 12:45 pm

neat video optimised for cleaning cgi renders?

Post by matt_ins »

CGI these days is moving towards more accurate simulation of light (mainly through variations of monte-carlo pathtracing), which as a by-product results in noisy renders.

We've been using Neat Video to help us clean up renders, and its been working very well, however we've been seeing some interesting research papers on denoising renders using information provided 'for free' from the render engine.

Is this something Neat Video would be interested in persuing?

Specifically, Neat Video is only aware of the input video stream. Because in 3D rendering we can provide MUCH more information about each pixel, and break that info out into separate image sequences, it seems possible that this could be used to do more accurate noise removal, and require less computation. For example, we can easily provide information about where each pixel is in space, how fast its moving, its texture independant from light and shadow, its angle relative to camera, its shadow component, masks for where there's important edge detail etc.

Here's some videos and papers that go into more detail.

http://www.youtube.com/watch?v=Ee51bkOlbMw
http://graphics.tu-bs.de/publications/Bauszat2011GIF/

Noise in pathtracers is a big problem in CG, Neat Video could become an even more popular tool if it came up with a good solution!
NVTeam
Posts: 2745
Joined: Thu Sep 01, 2005 4:12 pm
Contact:

Post by NVTeam »

Matt,

Thank you for the suggestion and links. That is indeed an interesting idea, which may be worth exploring in more details. I will add it our wish list and we will consider it as we work on newer versions of Neat Video.

Thank you once again.

Vlad
Zom-B
Posts: 1
Joined: Wed Mar 12, 2014 12:14 am

More Papers

Post by Zom-B »

Since I'm also using Neat Video (only) for denoising CG rendered output I'm also interested in enhancements in that direction.

I would like to share some more Papers regarding the denoising of CG output, maybe there is something usefull in there.

This is a mix of (speed) Enhancements methods, new denoising approaches and also rendering based stuff:

Dual-Domain Image Denoising
Adaptive Manifolds for Real-Time High-Dimensional Filtering
Domain Transform for Edge-Aware Image and Video Processing
Removing the Noise in Monte Carlo Rendering with General Image Denoising Algorithms
Practical Noise Reduction for Progressive Stochastic Ray Tracing with Perceptual Control
On Filtering the Noise from the Random Parameters in Monte Carlo Rendering
Implementation of Random Parameter Filtering
Probabilistic illumination-aware filtering for Monte Carlo rendering
Radiance Filtering for Interactive Path Tracing
Generalized sampling in computer graphics
Smoothed Local Histogram Filters Pixar Technical Memo 10-02
Adaptive Rendering with Non-Local Means Filtering
Edge-Optimized À-Trous Wavelets for Local Contrast Enhancement with Robust Denoising
Edge-Avoiding A-Trous Wavelet Transform for fast Global Illumination Filtering
Noise Reduction for Progressive Photon Mapping
Local Laplacian Filters: Edge-aware Image Processing with a Laplacian Pyramid
Accelerating Spatially Varying Gaussian Filters
Gaussian KD-Trees for Fast High-Dimensional Filtering
GPU-efficient recursive filtering and summed-area tables
Fast High-Dimensional Filtering Using the Permutohedral Lattice
Adaptive Rendering with Non-Local Means Filtering
Robust Image Denoising using a Virtual Flash Image for Monte Carlo Ray Tracing
Post Reply