Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

A real-time illumination rendering algorithm based on a deep learning network

A deep learning network and network technology, applied in the field of real-time lighting rendering algorithms, can solve problems such as inability to approximate global information, inability to accurately obtain detailed information, etc., and achieve the effect of improving general results, obvious dynamic effects, and increasing quantity and diversity.

Inactive Publication Date: 2019-03-01
安徽虚空位面信息科技有限公司
View PDF0 Cites 8 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

The advantage of data-driven methods is that global information can be generated in the short term, but detailed information cannot be accurately obtained
Traditional rasterized rendering methods can be used to produce accurate results, but cannot approximate global information

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • A real-time illumination rendering algorithm based on a deep learning network
  • A real-time illumination rendering algorithm based on a deep learning network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0022] Example: refer to Figure 1-2 , a real-time lighting rendering algorithm based on a deep learning network, including a training phase and a running phase; the training phase includes the following steps:

[0023] S1. Determination of the generator network structure: The generator network adopts the CNN model structure based on the U-Net structure. The input samples generate output results through 6-layer downsampling units and 6-layer upsampling units. Each downsampling unit contains a convolution layer, an activation layer and a downsampling layer, each upsampling unit contains a convolutional layer, an activation layer and an upsampling layer;

[0024] S2. Determination of the discriminator network structure: the discriminator network adopts a CNN model structure based on image classification. The format of the input data is the same as the output of the generator, which is sequentially connected to three fully connected layers through two convolutional layers and do...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a real-time rendering algorithm under a global illumination condition based on a deep learning network. The real-time rendering algorithm comprises a training stage and an operation stage. The training stage comprises the following steps: S1, determining a generator network structure; S2, determining a discriminator network structure; S3, collecting sample data; S4, neuralnetwork training; the process of the operation stage is as follows: the rendering cache is used as an input sample to be input into a generator network to obtain an output result, and then pixel coloring is performed on the output result to generate a rendering result under the global illumination condition, so that the traditional complex rendering calculation is replaced, and real-time renderingis realized. The confrontation neural network training is carried out on the multiple rendering caches and the photon mapping result under the same visual angle, the network output result is used asthe illumination prompt in pixel coloring under the global illumination condition including direct illumination and indirect illumination, and the final result of the confrontation network training ismore accurate and effective.

Description

technical field [0001] The invention relates to the technical field of real-time lighting rendering algorithms, in particular to a real-time lighting rendering algorithm based on a deep learning network. Background technique [0002] Some existing methods that can realize real-time global illumination rendering mainly include ray-tracing-based rendering methods, physically-based rendering methods, and data-driven rendering algorithms. The basic idea of ​​​​realizing global illumination is to calculate the bounce of light on the surface of the scene based on physical and geometric structures. Ray tracing is implemented by casting light to each pixel in the line of sight direction, and each pixel sampling can achieve indirect lighting according to the collision of the light with the surface and the bounce result of the light after the collision. The later photon mapping method is the inverse process of ray tracing, by tracing the light reflected by the light source to collect...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06T15/50G06N3/04G06N3/08
CPCG06N3/08G06T15/506G06N3/045
Inventor 黄佳维
Owner 安徽虚空位面信息科技有限公司
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products