gitadmin
/
three.js
mirror of https://github.com/mrdoob/three.js.git


			
							123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480
							<!DOCTYPE html><html lang="zh"><head>
    <meta charset="utf-8">
    <title>摄像机</title>
    <meta name="viewport" content="width=device-width, user-scalable=no, minimum-scale=1.0, maximum-scale=1.0">
    <meta name="twitter:card" content="summary_large_image">
    <meta name="twitter:site" content="@threejs">
    <meta name="twitter:title" content="Three.js – 摄像机">
    <meta property="og:image" content="https://threejs.org/files/share.png">
    <link rel="shortcut icon" href="../../files/favicon_white.ico" media="(prefers-color-scheme: dark)">
    <link rel="shortcut icon" href="../../files/favicon.ico" media="(prefers-color-scheme: light)">

    <link rel="stylesheet" href="../resources/lesson.css">
    <link rel="stylesheet" href="../resources/lang.css">
<!-- Import maps polyfill -->
<!-- Remove this when import maps will be widely supported -->
<script async src="https://unpkg.com/es-module-shims@1.3.6/dist/es-module-shims.js"></script>

<script type="importmap">
{
  "imports": {
    "three": "../../build/three.module.js"
  }
}
</script>
    <link rel="stylesheet" href="/manual/zh/lang.css">
  </head>
  <body>
    <div class="container">
      <div class="lesson-title">
        <h1>摄像机</h1>
      </div>
      <div class="lesson">
        <div class="lesson-main">
          <p>本文是关于 three.js 系列文章的一部分。第一篇文章是 <a href="fundamentals.html">three.js 基础</a>。如果你还没看过而且对three.js 还不熟悉，那应该从那里开始，并且了解如何<a href="setup.html">设置开发环境</a>。上一篇文章介绍了 three.js 中的 <a href="textures.html">纹理</a>。</p>
<p>我们开始谈谈three.js中的摄像机. 我们已经在<a href="fundamentals.html">第一篇文章</a> 中涉及到了摄像机的一些知识，这里我们要更深入一些. </p>
<p>在three.js中最常用的摄像机并且之前我们一直用的摄像机是<code class="notranslate" translate="no">透视摄像机 PerspectiveCamera</code>，它可以提供一个近大远小的3D视觉效果. </p>
<p><a href="/docs/#api/zh/cameras/PerspectiveCamera"><code class="notranslate" translate="no">PerspectiveCamera</code></a> 定义了一个 <em>视锥（frustum）</em>。<a href="https://en.wikipedia.org/wiki/Frustum"><em>frustum</em> 是一个切掉顶的三角锥或者说实心金字塔型</a>。
说到<em>实心体solid</em>，在这里通常是指一个立方体、一个圆锥、一个球、一个圆柱或锥台。</p>
<div class="spread">
  <div><div data-diagram="shapeCube"></div><div>立方体</div></div>
  <div><div data-diagram="shapeCone"></div><div>圆锥</div></div>
  <div><div data-diagram="shapeSphere"></div><div>球</div></div>
  <div><div data-diagram="shapeCylinder"></div><div>圆柱</div></div>
  <div><div data-diagram="shapeFrustum"></div><div>锥台</div></div>
</div>

<p>重新讲一遍这些东西是因为我好久没有在意过了。很多书或者文章提到<em>锥台</em>这个东西的时候我扫一眼就过去了。再了解一下不同几何体会让下面的一些表述变得更为感性...吧😅</p>
<p><a href="/docs/#api/zh/cameras/PerspectiveCamera"><code class="notranslate" translate="no">PerspectiveCamera</code></a>通过四个属性来定义一个视锥。<code class="notranslate" translate="no">near</code>定义了视锥的前端，<code class="notranslate" translate="no">far</code>定义了后端，<code class="notranslate" translate="no">fov</code>是视野，通过计算正确的高度来从摄像机的位置获得指定的以<code class="notranslate" translate="no">near</code>为单位的视野，定义的是视锥的前端和后端的高度。<code class="notranslate" translate="no">aspect</code>间接地定义了视锥前端和后端的宽度，实际上视锥的宽度是通过高度乘以 aspect 来得到的。</p>
<p><img src="../resources/frustum-3d.svg" width="500" class="threejs_center"></p>
<p>我们借用<a href="lights.html">上一篇文章</a>的场景. 其中包含一个地平面，一个球和一个立方体，我们可以在其中调整摄像机的设置。 
我们通过<code class="notranslate" translate="no">MinMaxGUIHelper</code>来调整<code class="notranslate" translate="no">near</code>，<code class="notranslate" translate="no">far</code>的设置。显然<code class="notranslate" translate="no">near</code>应该总是比<code class="notranslate" translate="no">far</code>要小。lil-gui 有<code class="notranslate" translate="no">min</code>和<code class="notranslate" translate="no">max</code>两个属性可调，然后这两个属性将决定摄像机的设置。</p>
<pre class="prettyprint showlinemods notranslate lang-js" translate="no">class MinMaxGUIHelper {
  constructor(obj, minProp, maxProp, minDif) {
    this.obj = obj;
    this.minProp = minProp;
    this.maxProp = maxProp;
    this.minDif = minDif;
  }
  get min() {
    return this.obj[this.minProp];
  }
  set min(v) {
    this.obj[this.minProp] = v;
    this.obj[this.maxProp] = Math.max(this.obj[this.maxProp], v + this.minDif);
  }
  get max() {
    return this.obj[this.maxProp];
  }
  set max(v) {
    this.obj[this.maxProp] = v;
    this.min = this.min;  // 这将调用min的setter
  }
}
</pre>
<p>现在我们可以将GUI设置为：</p>
<pre class="prettyprint showlinemods notranslate lang-js" translate="no">function updateCamera() {
  camera.updateProjectionMatrix();
}

const gui = new GUI();
gui.add(camera, 'fov', 1, 180).onChange(updateCamera);
const minMaxGUIHelper = new MinMaxGUIHelper(camera, 'near', 'far', 0.1);
gui.add(minMaxGUIHelper, 'min', 0.1, 50, 0.1).name('near').onChange(updateCamera);
gui.add(minMaxGUIHelper, 'max', 0.1, 50, 0.1).name('far').onChange(updateCamera);
</pre>
<p>任何时候摄像机的设置变动，我们需要调用摄像机的<a href="/docs/#api/zh/cameras/PerspectiveCamera#updateProjectionMatrix"><code class="notranslate" translate="no">updateProjectionMatrix</code></a>来更新设置。我们写一个函数<code class="notranslate" translate="no">updataCamera</code>，当lil-gui改变了属性的时候会调用它来更新参数。</p>
<p></p><div translate="no" class="threejs_example_container notranslate">
  <div><iframe class="threejs_example notranslate" translate="no" style=" " src="/manual/examples/resources/editor.html?url=/manual/examples/cameras-perspective.html"></iframe></div>
  <a class="threejs_center" href="/manual/examples/cameras-perspective.html" target="_blank">点击此处在新标签页中打开</a>
</div>

<p></p>
<p>现在可以调整这些数值来观察这些参数是如何影响摄像机的。注意我们并没有改变<code class="notranslate" translate="no">aspect</code>，因为这个参数来自于窗口的大小. 如果想调整<code class="notranslate" translate="no">aspect</code>，只需要开个新窗口然后调整窗口大小就可以了。</p>
<p>即便是这样，观察参数对视野的影响还是挺麻烦的. 所以我们来设置两台摄像机吧! 一台是跟上面一样展现出摄像机中看到的实际场景，另一个则是用来观察这个实际工作的摄像机，然后画出摄像机的视锥. </p>
<p>我们需要用到three.js的剪函数(scissor function)来画两个场景和两个摄像机. </p>
<p>首先让我们用HTML和CSS来定义两个肩并肩的元素. 这也将帮助我们将两个摄像机赋予不同的<a href="/docs/#examples/controls/OrbitControls"><code class="notranslate" translate="no">OrbitControls</code></a>。</p>
<pre class="prettyprint showlinemods notranslate lang-html" translate="no">&lt;body&gt;
  &lt;canvas id="c"&gt;&lt;/canvas&gt;
+  &lt;div class="split"&gt;
+     &lt;div id="view1" tabindex="1"&gt;&lt;/div&gt;
+     &lt;div id="view2" tabindex="2"&gt;&lt;/div&gt;
+  &lt;/div&gt;
&lt;/body&gt;
</pre>
<p>CSS将控制两个视窗并排显示在 canvas 中：</p>
<pre class="prettyprint showlinemods notranslate lang-css" translate="no">.split {
  position: absolute;
  left: 0;
  top: 0;
  width: 100%;
  height: 100%;
  display: flex;
}
.split&gt;div {
  width: 100%;
  height: 100%;
}
</pre>
<p>接下来将添加一个<a href="/docs/#api/zh/helpers/CameraHelper"><code class="notranslate" translate="no">CameraHelper</code></a>, 它可以把摄像机的视锥画出来：</p>
<pre class="prettyprint showlinemods notranslate lang-js" translate="no">const cameraHelper = new THREE.CameraHelper(camera);

...

scene.add(cameraHelper);
</pre>
<p>我们现在需要查找到刚刚定义的两个元素：</p>
<pre class="prettyprint showlinemods notranslate lang-js" translate="no">const view1Elem = document.querySelector('#view1');
const view2Elem = document.querySelector('#view2');
</pre>
<p>现在只给第一个视窗中的摄像机分配<a href="/docs/#examples/controls/OrbitControls"><code class="notranslate" translate="no">OrbitControls</code></a>：</p>
<pre class="prettyprint showlinemods notranslate lang-js" translate="no">-const controls = new OrbitControls(camera, canvas);
+const controls = new OrbitControls(camera, view1Elem);
</pre>
<p>我们定义第二个<a href="/docs/#api/zh/cameras/PerspectiveCamera"><code class="notranslate" translate="no">PerspectiveCamera</code></a>和<a href="/docs/#examples/controls/OrbitControls"><code class="notranslate" translate="no">OrbitControls</code></a>：</p>
<pre class="prettyprint showlinemods notranslate lang-js" translate="no">const camera2 = new THREE.PerspectiveCamera(
  60,  // fov
  2,   // aspect
  0.1, // near
  500, // far
);
camera2.position.set(40, 10, 30);
camera2.lookAt(0, 5, 0);

const controls2 = new OrbitControls(camera2, view2Elem);
controls2.target.set(0, 5, 0);
controls2.update();
</pre>
<p>最后，我们需要使用剪刀功能从每个摄影机的视角渲染场景，以仅渲染画布的一部分。
这个函数接受一个元素，计算这个元素在canvas上的重叠面积，这将设置剪刀函数和视角长宽并返回 aspect ：</p>
<pre class="prettyprint showlinemods notranslate lang-js" translate="no">function setScissorForElement(elem) {
  const canvasRect = canvas.getBoundingClientRect();
  const elemRect = elem.getBoundingClientRect();

  // 计算canvas的尺寸
  const right = Math.min(elemRect.right, canvasRect.right) - canvasRect.left;
  const left = Math.max(0, elemRect.left - canvasRect.left);
  const bottom = Math.min(elemRect.bottom, canvasRect.bottom) - canvasRect.top;
  const top = Math.max(0, elemRect.top - canvasRect.top);

  const width = Math.min(canvasRect.width, right - left);
  const height = Math.min(canvasRect.height, bottom - top);

  // 设置剪函数以仅渲染一部分场景
  const positiveYUpBottom = canvasRect.height - bottom;
  renderer.setScissor(left, positiveYUpBottom, width, height);
  renderer.setViewport(left, positiveYUpBottom, width, height);

  // 返回aspect
  return width / height;
}
</pre>
<p>我们用这个函数在<code class="notranslate" translate="no">render</code>中绘制两遍场景</p>
<pre class="prettyprint showlinemods notranslate lang-js" translate="no">  function render() {

-    if (resizeRendererToDisplaySize(renderer)) {
-      const canvas = renderer.domElement;
-      camera.aspect = canvas.clientWidth / canvas.clientHeight;
-      camera.updateProjectionMatrix();
-    }

+    resizeRendererToDisplaySize(renderer);
+
+    // 启用剪刀函数
+    renderer.setScissorTest(true);
+
+    // 渲染主视野
+    {
+      const aspect = setScissorForElement(view1Elem);
+
+      // 用计算出的aspect修改摄像机参数
+      camera.aspect = aspect;
+      camera.updateProjectionMatrix();
+      cameraHelper.update();
+
+      // 来原视野中不要绘制cameraHelper
+      cameraHelper.visible = false;
+
+      scene.background.set(0x000000);
+
+      // 渲染
+      renderer.render(scene, camera);
+    }
+
+    // 渲染第二台摄像机
+    {
+      const aspect = setScissorForElement(view2Elem);
+
+      // 调整aspect
+      camera2.aspect = aspect;
+      camera2.updateProjectionMatrix();
+
+      // 在第二台摄像机中绘制cameraHelper
+      cameraHelper.visible = true;
+
+      scene.background.set(0x000040);
+
+      renderer.render(scene, camera2);
+    }

-    renderer.render(scene, camera);

    requestAnimationFrame(render);
  }

  requestAnimationFrame(render);
}
</pre>
<p>上面的代码还将主辅摄像机的背景色区分开以利观察。</p>
<p>我们可以移除<code class="notranslate" translate="no">updateCamera</code>了，因为所有的东西在<code class="notranslate" translate="no">render</code>中更新过了。</p>
<pre class="prettyprint showlinemods notranslate lang-js" translate="no">-function updateCamera() {
-  camera.updateProjectionMatrix();
-}

const gui = new GUI();
-gui.add(camera, 'fov', 1, 180).onChange(updateCamera);
+gui.add(camera, 'fov', 1, 180);
const minMaxGUIHelper = new MinMaxGUIHelper(camera, 'near', 'far', 0.1);
-gui.add(minMaxGUIHelper, 'min', 0.1, 50, 0.1).name('near').onChange(updateCamera);
-gui.add(minMaxGUIHelper, 'max', 0.1, 50, 0.1).name('far').onChange(updateCamera);
+gui.add(minMaxGUIHelper, 'min', 0.1, 50, 0.1).name('near');
+gui.add(minMaxGUIHelper, 'max', 0.1, 50, 0.1).name('far');
</pre>
<p>现在我们就可以在辅摄像机中观察到主摄像机的视锥轮廓了。</p>
<p></p><div translate="no" class="threejs_example_container notranslate">
  <div><iframe class="threejs_example notranslate" translate="no" style=" " src="/manual/examples/resources/editor.html?url=/manual/examples/cameras-perspective-2-scenes.html"></iframe></div>
  <a class="threejs_center" href="/manual/examples/cameras-perspective-2-scenes.html" target="_blank">点击此处在新标签页中打开</a>
</div>

<p></p>
<p>左侧可以看到主摄像机的视角，右侧则是辅摄像机观察主摄像机和主摄像机的视锥轮廓。可以调整<code class="notranslate" translate="no">near</code>，<code class="notranslate" translate="no">far</code>，<code class="notranslate" translate="no">fov</code>和用鼠标移动摄像机来观察视锥轮廓和场景之间的关系. </p>
<p>将<code class="notranslate" translate="no">near</code>调整到大概20左右，前景就会在视锥中消失。<code class="notranslate" translate="no">far</code>低于35时，远景也不复存在. </p>
<p>这带来一个问题，为什么不把<code class="notranslate" translate="no">near</code>设置到0.0000000001然后将<code class="notranslate" translate="no">far</code>设置成100000000，使得一切都可以尽收眼底? 原因是你的GPU没有足够的精度来决定某个东西是另一个东西的前面还是后面。更糟的是，在默认情况下，离摄像机近的将会更清晰，离摄像机远的模糊，从<code class="notranslate" translate="no">near</code>到<code class="notranslate" translate="no">far</code>逐渐过渡。</p>
<p>从上面的例子出发，我们向场景中添加20个球：</p>
<pre class="prettyprint showlinemods notranslate lang-js" translate="no">{
  const sphereRadius = 3;
  const sphereWidthDivisions = 32;
  const sphereHeightDivisions = 16;
  const sphereGeo = new THREE.SphereBufferGeometry(sphereRadius, sphereWidthDivisions, sphereHeightDivisions);
  const numSpheres = 20;
  for (let i = 0; i &lt; numSpheres; ++i) {
    const sphereMat = new THREE.MeshPhongMaterial();
    sphereMat.color.setHSL(i * .73, 1, 0.5);
    const mesh = new THREE.Mesh(sphereGeo, sphereMat);
    mesh.position.set(-sphereRadius - 1, sphereRadius + 2, i * sphereRadius * -2.2);
    scene.add(mesh);
  }
}
</pre>
<p>把 <code class="notranslate" translate="no">near</code> 设置成0.00001</p>
<pre class="prettyprint showlinemods notranslate lang-js" translate="no">const fov = 45;
const aspect = 2;  // canvas 默认
-const near = 0.1;
+const near = 0.00001;
const far = 100;
const camera = new THREE.PerspectiveCamera(fov, aspect, near, far);
</pre>
<p>调整一下GUI使得能设置到0.00001</p>
<pre class="prettyprint showlinemods notranslate lang-js" translate="no">-gui.add(minMaxGUIHelper, 'min', 0.1, 50, 0.1).name('near').onChange(updateCamera);
+gui.add(minMaxGUIHelper, 'min', 0.00001, 50, 0.00001).name('near').onChange(updateCamera);
</pre>
<p>你觉得会发生什么？</p>
<p></p><div translate="no" class="threejs_example_container notranslate">
  <div><iframe class="threejs_example notranslate" translate="no" style=" " src="/manual/examples/resources/editor.html?url=/manual/examples/cameras-z-fighting.html"></iframe></div>
  <a class="threejs_center" href="/manual/examples/cameras-z-fighting.html" target="_blank">点击此处在新标签页中打开</a>
</div>

<p></p>
<p>这就是一个典型的<em>z冲突</em>的例子。GPU没有足够的精度来决定哪个像素在前哪个在后。</p>
<p>如果你的机器太好可能不会出现我说的情况，我把我看到的截图放在这：</p>
<div class="threejs_center"><img src="../resources/images/z-fighting.png" style="width: 570px;"></div>

<p>解决的方法之一是告诉three.js使用不同的方法计算像素的前后关系。我们可以在创建<a href="/docs/#api/zh/renderers/WebGLRenderer"><code class="notranslate" translate="no">WebGLRenderer</code></a>时开启<code class="notranslate" translate="no">logarithmicDepthBuffer</code>。</p>
<pre class="prettyprint showlinemods notranslate lang-js" translate="no">-const renderer = new THREE.WebGLRenderer({antialias: true, canvas});
+const renderer = new THREE.WebGLRenderer({
+  antialias: true,
+  canvas,
+  logarithmicDepthBuffer: true,
+});
</pre>
<p>这看起来就行了</p>
<p></p><div translate="no" class="threejs_example_container notranslate">
  <div><iframe class="threejs_example notranslate" translate="no" style=" " src="/manual/examples/resources/editor.html?url=/manual/examples/cameras-logarithmic-depth-buffer.html"></iframe></div>
  <a class="threejs_center" href="/manual/examples/cameras-logarithmic-depth-buffer.html" target="_blank">点击此处在新标签页中打开</a>
</div>

<p></p>
<p>如果上面的方案不行的话，那你就遇到了<em>为什么不能无脑使用这种解决方案</em>的情况了。截止2018年9月，绝大多数台式机可以，但几乎没有移动设备支持这个功能。</p>
<p>另外，最好别用这种解决方案，因为这会大大降低运行速度。</p>
<p>即便是现在跑得好好地，选择太小的<code class="notranslate" translate="no">near</code>和太大的<code class="notranslate" translate="no">far</code>最终也会遇到同样的问题。</p>
<p>所以说你需要选择好好抉择<code class="notranslate" translate="no">near</code>和<code class="notranslate" translate="no">far</code>的设置，来和你的场景配合。既不丢失重要的近景，也不让远处的东西消失不见。如果你想渲染一个巨大的场景，不但能看清面前的人的眼睫毛又想看到50公里以外的玩意，你得自己想一个<em>厉害的</em>方案，这里就不涉及了。现在，好好地选个适合的参数就行。</p>
<p>第二种常见的摄像机是<code class="notranslate" translate="no">正交摄像机 OrthographicCamera</code>，和指定一个视锥不同的是，它需要设置<code class="notranslate" translate="no">left</code>，<code class="notranslate" translate="no">right</code>
<code class="notranslate" translate="no">top</code>，<code class="notranslate" translate="no">bottom</code>，<code class="notranslate" translate="no">near</code>，和<code class="notranslate" translate="no">far</code>指定一个长方体，使得视野是平行的而不是透视的。</p>
<p>我们来把上面的例子改成<a href="/docs/#api/zh/cameras/OrthographicCamera"><code class="notranslate" translate="no">OrthographicCamera</code></a>, 首先来设置摄像机</p>
<pre class="prettyprint showlinemods notranslate lang-js" translate="no">const left = -1;
const right = 1;
const top = 1;
const bottom = -1;
const near = 5;
const far = 50;
const camera = new THREE.OrthographicCamera(left, right, top, bottom, near, far);
camera.zoom = 0.2;
</pre>
<p>我们将<code class="notranslate" translate="no">left</code>和<code class="notranslate" translate="no">bottom</code>设置成 -1 <code class="notranslate" translate="no">right</code> 和 <code class="notranslate" translate="no">top</code>设成 1，这样就使盒子宽为两个单位，高两个单位。我们接下来通过调整<code class="notranslate" translate="no">left</code>和<code class="notranslate" translate="no">top</code>来选择其 aspect 。我们将用<code class="notranslate" translate="no">zoom</code>属性来调整相机到底展现多少的单位大小。</p>
<p>给GUI添加<code class="notranslate" translate="no">zoom</code>设置：</p>
<pre class="prettyprint showlinemods notranslate lang-js" translate="no">const gui = new GUI();
+gui.add(camera, 'zoom', 0.01, 1, 0.01).listen();
</pre>
<p><code class="notranslate" translate="no">listen</code>调用告诉lil-gui去监视属性的变化。写在这里是因为<a href="/docs/#examples/controls/OrbitControls"><code class="notranslate" translate="no">OrbitControls</code></a>同样可以控制缩放。在这个例子中，鼠标滚轮将会通过<a href="/docs/#examples/controls/OrbitControls"><code class="notranslate" translate="no">OrbitControls</code></a>控件来控制缩放。</p>
<p>最后更改aspect然后更新摄像机：</p>
<pre class="prettyprint showlinemods notranslate lang-js" translate="no">{
  const aspect = setScissorForElement(view1Elem);

  // 使用aspect更新摄像机
-  camera.aspect = aspect;
+  camera.left   = -aspect;
+  camera.right  =  aspect;
  camera.updateProjectionMatrix();
  cameraHelper.update();

  // 在主摄像机中不绘制视野辅助线
  cameraHelper.visible = false;

  scene.background.set(0x000000);
  renderer.render(scene, camera);
}
</pre>
<p>现在就可以看到<a href="/docs/#api/zh/cameras/OrthographicCamera"><code class="notranslate" translate="no">OrthographicCamera</code></a>工作了。</p>
<p></p><div translate="no" class="threejs_example_container notranslate">
  <div><iframe class="threejs_example notranslate" translate="no" style=" " src="/manual/examples/resources/editor.html?url=/manual/examples/cameras-orthographic-2-scenes.html"></iframe></div>
  <a class="threejs_center" href="/manual/examples/cameras-orthographic-2-scenes.html" target="_blank">点击此处在新标签页中打开</a>
</div>

<p></p>
<p>大多数情况下，绘制2D图像的时候会用到<a href="/docs/#api/zh/cameras/OrthographicCamera"><code class="notranslate" translate="no">OrthographicCamera</code></a>。你可以自己决定摄像机的视野大小。比如说你想让 canvas 的一个像素匹配摄像机的一个单位，你可以这么做：</p>
<p>将原点置于中心，令一个像素等于一个单位</p>
<pre class="prettyprint showlinemods notranslate lang-js" translate="no">camera.left = -canvas.width / 2;
camera.right = canvas.width / 2;
camera.top = canvas.height / 2;
camera.bottom = -canvas.height / 2;
camera.near = -1;
camera.far = 1;
camera.zoom = 1;
</pre>
<p>或者如果我们想让原点在左上，就像是2D canvas</p>
<pre class="prettyprint showlinemods notranslate lang-js" translate="no">camera.left = 0;
camera.right = canvas.width;
camera.top = 0;
camera.bottom = canvas.height;
camera.near = -1;
camera.far = 1;
camera.zoom = 1;
</pre>
<p>这样左上角就成了0,0</p>
<p>试试，这样设置摄像机</p>
<pre class="prettyprint showlinemods notranslate lang-js" translate="no">const left = 0;
const right = 300;  // 默认的canvas大小
const top = 0;
const bottom = 150;  // 默认的canvas大小
const near = -1;
const far = 1;
const camera = new THREE.OrthographicCamera(left, right, top, bottom, near, far);
camera.zoom = 1;
</pre>
<p>然后我们载入六个材质，生成六个平面，一一对应。把每一个平面绑定到父对象<a href="/docs/#api/zh/core/Object3D"><code class="notranslate" translate="no">THREE.Object3D</code></a>上，以便调整每个平面和左上角原点的相对关系</p>
<pre class="prettyprint showlinemods notranslate lang-js" translate="no">const loader = new THREE.TextureLoader();
const textures = [
  loader.load('resources/images/flower-1.jpg'),
  loader.load('resources/images/flower-2.jpg'),
  loader.load('resources/images/flower-3.jpg'),
  loader.load('resources/images/flower-4.jpg'),
  loader.load('resources/images/flower-5.jpg'),
  loader.load('resources/images/flower-6.jpg'),
];
const planeSize = 256;
const planeGeo = new THREE.PlaneGeometry(planeSize, planeSize);
const planes = textures.map((texture) =&gt; {
  const planePivot = new THREE.Object3D();
  scene.add(planePivot);
  texture.magFilter = THREE.NearestFilter;
  const planeMat = new THREE.MeshBasicMaterial({
    map: texture,
    side: THREE.DoubleSide,
  });
  const mesh = new THREE.Mesh(planeGeo, planeMat);
  planePivot.add(mesh);
  // 调整平面使得左上角为原点
  mesh.position.set(planeSize / 2, planeSize / 2, 0);
  return planePivot;
});
</pre>
<p>然后当 canvas 更新后我们更新摄像机设置</p>
<pre class="prettyprint showlinemods notranslate lang-js" translate="no">function render() {

  if (resizeRendererToDisplaySize(renderer)) {
    camera.right = canvas.width;
    camera.bottom = canvas.height;
    camera.updateProjectionMatrix();
  }

  ...
</pre>
<p><code class="notranslate" translate="no">planes</code>是<a href="/docs/#api/zh/objects/Mesh"><code class="notranslate" translate="no">THREE.Mesh</code></a>的数组，每一个对应一个平面。
现在让它随着时间移动</p>
<pre class="prettyprint showlinemods notranslate lang-js" translate="no">function render(time) {
  time *= 0.001;  // 转换为秒;

  ...

  const distAcross = Math.max(20, canvas.width - planeSize);
  const distDown = Math.max(20, canvas.height - planeSize);

  // 来回运动的总距离
  const xRange = distAcross * 2;
  const yRange = distDown * 2;
  const speed = 180;

  planes.forEach((plane, ndx) =&gt; {
    // 为每个平面单独计算时间
    const t = time * speed + ndx * 300;

    // 在0到最远距离之间获取一个值
    const xt = t % xRange;
    const yt = t % yRange;

    // 0到距离的一半, 向前运动
    // 另一半的时候往回运动
    const x = xt &lt; distAcross ? xt : xRange - xt;
    const y = yt &lt; distDown   ? yt : yRange - yt;

    plane.position.set(x, y, 0);
  });

  renderer.render(scene, camera);
</pre>
<p>你可以看到图片在其中弹跳，和边际完美契合，就是2D canvas的效果一样</p>
<p></p><div translate="no" class="threejs_example_container notranslate">
  <div><iframe class="threejs_example notranslate" translate="no" style=" " src="/manual/examples/resources/editor.html?url=/manual/examples/cameras-orthographic-canvas-top-left-origin.html"></iframe></div>
  <a class="threejs_center" href="/manual/examples/cameras-orthographic-canvas-top-left-origin.html" target="_blank">点击此处在新标签页中打开</a>
</div>

<p></p>
<p>另一个常见的用途是用<a href="/docs/#api/zh/cameras/OrthographicCamera"><code class="notranslate" translate="no">OrthographicCamera</code></a>来展示模型的三视图。</p>
<div class="threejs_center"><img src="../resources/images/quad-viewport.png" style="width: 574px;"></div>

<p>上面的截图展示了一个透视图和三个正交视角。</p>
<p>这就是摄像机的基础. 我们在其他的文章中会介绍另外的一些摄像机用法。现在，我们移步到<a href="shadows.html">阴影</a>。</p>
<p><canvas id="c"></canvas></p>
<script type="module" src="../resources/threejs-cameras.js"></script>

        </div>
      </div>
    </div>
  
  <script src="../resources/prettify.js"></script>
  <script src="../resources/lesson.js"></script>


</body></html>