Rendering program online optimization method

Abstract

The invention discloses a rendering program online optimization method comprising the following steps: building reduced programs of an original rendering program, selecting a plurality of reduced rendering programs as representative rendering programs according to the rendering cost and drawing error, and building a reduced dependency graph based on the dependence relationship; and in the rendering process, monitoring the parameters of a scene to be drawn, using a new round of online optimization when there is severe change, and completing one optimization through multiple cycles. The following operations are performed cyclically in each optimization: selecting K reduced programs from all the reduced programs as candidate reduced programs according to the reduced dependency graph, determining the current cycling result according to the drawing error and rendering cost, and deciding whether the current online optimization ends according to a plurality of cycling results. Optimal reduced rendering programs are selected dynamically, decoupling of rendering programs from a scene is realized, and the enumeration parameter space problem of off-line optimization is avoided. Moreover, rendering program error calculation and time measurement and scene drawing are carried out at the same time, operation is quick, and the real-time performance is strong.

Description

technical field [0001] The invention relates to the field of real-time rendering, in particular to an online optimization method for a rendering program, which is used for optimizing the rendering program online during the rendering process. Background technique [0002] In the field of real-time rendering, there is always a need to speed up the rendering speed, and the complexity of the rendering program is an important factor in determining the rendering efficiency. Rendering programs are composed of various shader codes, including vertex, hull, subdivision, geometry, pixel, calculation and other types. Each of these shaders corresponds to a stage in the programmable hardware rendering pipeline. During the drawing process, each shader in the rendering program will be called and executed in different rendering stages, and the number of times different shaders are executed varies with the corresponding stages. In the application process, the simpler the rendering program c...

AI Technical Summary

Problems solved by technology

But a large number of simplification procedures need to consume a long precomputation time

[0006] 2. Offline optimization fixes the rendered model. When calculating time and error, only a limited number of viewing angles can be selected for calculation, and it cannot cover all possible spaces during operation.

Images