Javascript 使用回收帧缓冲区的threejs片段着色器

我正在尝试制作一个模拟长曝光摄影的应用程序。我的想法是从网络摄像头抓取当前帧并将其合成到画布上。随着时间的推移，照片将“曝光”，变得越来越亮。（见附件）

我有一个完美的着色器。这就像photoshop中的“添加”混合模式。问题是我无法让它回收上一帧

我认为它应该是一些简单的东西，比如

renderer.autoClear=false
下面是使用THREE.EffectComposer应用着色器的代码

        onWebcamInit: function () {    
            var $stream = $("#user-stream"),
                width = $stream.width(),
                height = $stream.height(),
                near = .1,
                far = 10000;

            this.renderer = new THREE.WebGLRenderer();
            this.renderer.setSize(width, height);
            this.renderer.autoClear = false;
            this.scene = new THREE.Scene();

            this.camera = new THREE.OrthographicCamera(width / -2, width / 2, height / 2, height / -2, near, far);
            this.scene.add(this.camera);

            this.$el.append(this.renderer.domElement);

            this.frameTexture = new THREE.Texture(document.querySelector("#webcam"));
            this.compositeTexture = new THREE.Texture(this.renderer.domElement);

            this.composer = new THREE.EffectComposer(this.renderer);

            // same effect with or without this line
            // this.composer.addPass(new THREE.RenderPass(this.scene, this.camera));

            var addEffect = new THREE.ShaderPass(addShader);
            addEffect.uniforms[ 'exposure' ].value = .5;
            addEffect.uniforms[ 'frameTexture' ].value = this.frameTexture;
            addEffect.renderToScreen = true;
            this.composer.addPass(addEffect);

            this.plane = new THREE.Mesh(new THREE.PlaneGeometry(width, height, 1, 1), new THREE.MeshBasicMaterial({map: this.compositeTexture}));
            this.scene.add(this.plane);

            this.frameTexture.needsUpdate = true;
            this.compositeTexture.needsUpdate = true;

            new FrameImpulse(this.renderFrame);

        },
        renderFrame: function () {
            this.frameTexture.needsUpdate = true;
            this.compositeTexture.needsUpdate = true;
            this.composer.render();
        }

这是着色器。没什么特别的

        uniforms: {
            "tDiffuse": { type: "t", value: null },
            "frameTexture": { type: "t", value: null },
            "exposure": { type: "f", value: 1.0 }
        },

        vertexShader: [
            "varying vec2 vUv;",
            "void main() {",
            "vUv = uv;",
            "gl_Position = projectionMatrix * modelViewMatrix * vec4( position, 1.0 );",

            "}"
        ].join("\n"),

        fragmentShader: [

            "uniform sampler2D frameTexture;",
            "uniform sampler2D tDiffuse;",
            "uniform float exposure;",
            "varying vec2 vUv;",

            "void main() {",
            "vec4 n = texture2D(frameTexture, vUv);",
            "vec4 o = texture2D(tDiffuse, vUv);",
            "vec3 sum = n.rgb + o.rgb;",
            "gl_FragColor = vec4(mix(o.rgb, sum.rgb, exposure), 1.0);",
            "}"

        ].join("\n")

要实现这种反馈效果，必须交替编写
WebGLRenderTarget
的单独实例。否则，将覆盖帧缓冲区。不完全确定为什么会发生这种情况。。。但这是解决办法

初始化：

呈现：

    this.renderer.render(this.scene, this.camera);
    this.renderer.render(this.scene, this.camera, this.rt1, false);

    // swap buffers
    var a = this.rt2;
    this.rt2 = this.rt1;
    this.rt1 = a;
    this.shaders.add.uniforms.tDiffuse.value = this.rt2;

composer.render();
composer.swapTargets();

试试这个：

this.renderer = new THREE.WebGLRenderer( { preserveDrawingBuffer: true } );

这在本质上相当于posit labs answer，但我已经成功地使用了一个更精简的解决方案—我创建了一个EffectComposer，只使用我想要回收的ShaderPass，然后在每个渲染中为该composer交换renderTargets

初始化：

THREE.EffectComposer.prototype.swapTargets = function() {
    var tmp = this.renderTarget2;
    this.renderTarget2 = this.renderTarget1;
    this.renderTarget1 = tmp;
};

...

composer = new THREE.EffectComposer(renderer,  
    new THREE.WebGLRenderTarget(512, 512, { minFilter: THREE.LinearFilter, magFilter: THREE.NearestFilter, format: THREE.RGBFormat })
);

var addEffect = new THREE.ShaderPass(addShader, 'frameTexture');
addEffect.renderToScreen = true;
this.composer.addPass(addEffect);

呈现：

    this.renderer.render(this.scene, this.camera);
    this.renderer.render(this.scene, this.camera, this.rt1, false);

    // swap buffers
    var a = this.rt2;
    this.rt2 = this.rt1;
    this.rt1 = a;
    this.shaders.add.uniforms.tDiffuse.value = this.rt2;

composer.render();
composer.swapTargets();

然后，次要EffectComposer可以获取两个渲染目标中的一个，并将其推送到屏幕上或对其进行进一步变换

还要注意，在初始化着色器类时，我将“frameTexture”声明为textureID。这让ShaderPass知道如何使用上一个过程的结果更新帧纹理一致性。
似乎表明您应该创建WebGLRenderer，如下所示：
新建WebGLRenderer（{preserveDrawingBuffer:true}）
，并将
renderer.autoClearColor
设置为false。这将用于创建几何体效果（如运动模糊），但它对几何体上的实际纹理没有影响。这正是我试图瞄准的目标。感谢您的响应，doob。不过，恐怕这对我不起作用。我发布的作为答案的“渲染到纹理”技术非常有效……但我仍然不确定为什么会起作用。从它看来，渲染过程可能会被标记为“needsSwap”可自动执行此操作。