bytev-charts-beta/lib/plugin/three.js/examples/js/shaders/ConvolutionShader.js

基于echarts和JavaScript及ES6封装的一个可以直接调用的图表组件库，内置主题设计，简单快捷，且支持用户自定义配置； npm 安装方式: npm install bytev-charts 若启动提示还需额外install插件,则运行 npm install @babel/runtime-corejs2 即可;

import "core-js/modules/es.array.join.js";
console.warn("THREE.ConvolutionShader: As part of the transition to ES6 Modules, the files in 'examples/js' were deprecated in May 2020 (r117) and will be deleted in December 2020 (r124). You can find more information about developing using ES6 Modules in https://threejs.org/docs/#manual/en/introduction/Installation.");
/**
 * Convolution shader
 * ported from o3d sample to WebGL / GLSL
 * http://o3d.googlecode.com/svn/trunk/samples/convolution.html
 */

THREE.ConvolutionShader = {
  defines: {
    "KERNEL_SIZE_FLOAT": "25.0",
    "KERNEL_SIZE_INT": "25"
  },
  uniforms: {
    "tDiffuse": {
      value: null
    },
    "uImageIncrement": {
      value: new THREE.Vector2(0.001953125, 0.0)
    },
    "cKernel": {
      value: []
    }
  },
  vertexShader: ["uniform vec2 uImageIncrement;", "varying vec2 vUv;", "void main() {", "	vUv = uv - ( ( KERNEL_SIZE_FLOAT - 1.0 ) / 2.0 ) * uImageIncrement;", "	gl_Position = projectionMatrix * modelViewMatrix * vec4( position, 1.0 );", "}"].join("\n"),
  fragmentShader: ["uniform float cKernel[ KERNEL_SIZE_INT ];", "uniform sampler2D tDiffuse;", "uniform vec2 uImageIncrement;", "varying vec2 vUv;", "void main() {", "	vec2 imageCoord = vUv;", "	vec4 sum = vec4( 0.0, 0.0, 0.0, 0.0 );", "	for( int i = 0; i < KERNEL_SIZE_INT; i ++ ) {", "		sum += texture2D( tDiffuse, imageCoord ) * cKernel[ i ];", "		imageCoord += uImageIncrement;", "	}", "	gl_FragColor = sum;", "}"].join("\n"),
  buildKernel: function buildKernel(sigma) {
    // We lop off the sqrt(2 * pi) * sigma term, since we're going to normalize anyway.
    function gauss(x, sigma) {
      return Math.exp(-(x * x) / (2.0 * sigma * sigma));
    }

    var i,
        values,
        sum,
        halfWidth,
        kMaxKernelSize = 25,
        kernelSize = 2 * Math.ceil(sigma * 3.0) + 1;
    if (kernelSize > kMaxKernelSize) kernelSize = kMaxKernelSize;
    halfWidth = (kernelSize - 1) * 0.5;
    values = new Array(kernelSize);
    sum = 0.0;

    for (i = 0; i < kernelSize; ++i) {
      values[i] = gauss(i - halfWidth, sigma);
      sum += values[i];
    } // normalize the kernel


    for (i = 0; i < kernelSize; ++i) {
      values[i] /= sum;
    }

    return values;
  }
};