- A+
这里给大家分享我在网上总结出来的一些知识,希望对大家有所帮助
建模
首先我们需要一些贴图素材
贴图素材一般可以在3dtextures网站上找到,这里我找了2份,包含了墙的法线贴图和潮湿地面的法线、透明度、粗糙度贴图
通过kokomi.AssetManager将贴图素材一次性全部加载出来,将它们应用到Mesh
上,加上基本的环境光照,即可完成最基本的建模
// 光照 const pointLight1 = new THREE.PointLight(config.color, 0.5, 17, 0.8); pointLight1.position.set(0, 2, 0); this.scene.add(pointLight1); ... // 网格 const aspTex = am.items["asphalt-normal"]; aspTex.rotation = THREE.MathUtils.degToRad(90); aspTex.wrapS = aspTex.wrapT = THREE.RepeatWrapping; aspTex.repeat.set(5, 8); const wallMat = new THREE.MeshPhongMaterial({ color: new THREE.Color("#111111"), normalMap: aspTex, normalScale: new THREE.Vector2(0.5, 0.5), shininess: 200, }); const wall = new THREE.Mesh(new THREE.BoxGeometry(25, 20, 0.5), wallMat); this.scene.add(wall); wall.position.y = 10; wall.position.z = -10.3; ... // 文字 const t3d = new kokomi.Text3D(this, config.text, font, { size: 3, height: 0.2, curveSegments: 120, bevelEnabled: false, }); t3d.mesh.geometry.center(); const tm = new THREE.Mesh( t3d.mesh.geometry, new THREE.MeshBasicMaterial({ color: config.color, }) ); this.scene.add(tm); tm.position.y = 1.54;
积水地面
地面上的积水能反射出周围的景色,因此我们将选用kokomi.Reflector来实现反射效果
const mirror = new kokomi.Reflector(new THREE.PlaneGeometry(25, 100)); mirror.position.z = -25; mirror.rotation.x = -Math.PI / 2;
普通的反射器仅仅是一面镜子,因此我们要自定义反射器的Shader
涟漪效果
之前逛shadertoy时看到了一个很棒的涟漪特效,就直接拿来用了
// https://www.shadertoy.com/view/4djSRW float hash12(vec2 p){ vec3 p3=fract(vec3(p.xyx)*.1031); p3+=dot(p3,p3.yzx+19.19); return fract((p3.x+p3.y)*p3.z); } vec2 hash22(vec2 p){ vec3 p3=fract(vec3(p.xyx)*vec3(.1031,.1030,.0973)); p3+=dot(p3,p3.yzx+19.19); return fract((p3.xx+p3.yz)*p3.zy); } // https://gist.github.com/companje/29408948f1e8be54dd5733a74ca49bb9 float map(float value,float min1,float max1,float min2,float max2){ return min2+(value-min1)*(max2-min2)/(max1-min1); } vec2 rippleUv=75.*p*uTexScale; vec2 p0=floor(rippleUv); float rainStrength=map(uRainCount,0.,10000.,3.,.5); if(rainStrength==3.){ rainStrength=50.; } vec2 circles=vec2(0.); for(int j=-MAX_RADIUS;j<=MAX_RADIUS;++j) { for(int i=-MAX_RADIUS;i<=MAX_RADIUS;++i) { vec2 pi=p0+vec2(i,j); #if DOUBLE_HASH vec2 hsh=hash22(pi); #else vec2 hsh=pi; #endif vec2 p=pi+hash22(hsh); float t=fract(.8*iTime+hash12(hsh)); vec2 v=p-rippleUv; float d=length(v)-(float(MAX_RADIUS)+1.)*t+(rainStrength*.1*t); float h=1e-3; float d1=d-h; float d2=d+h; float p1=sin(31.*d1)*smoothstep(-.6,-.3,d1)*smoothstep(0.,-.3,d1); float p2=sin(31.*d2)*smoothstep(-.6,-.3,d2)*smoothstep(0.,-.3,d2); circles+=.5*normalize(v)*((p2-p1)/(2.*h)*(1.-t)*(1.-t)); } } circles/=float((MAX_RADIUS*2+1)*(MAX_RADIUS*2+1)); float intensity=.05*floorOpacity; vec3 n=vec3(circles,sqrt(1.-dot(circles,circles))); vec2 rainUv=intensity*n.xy;
与地面结合
光有涟漪效果也不够,要将它与地面的贴图相结合起来
这里采用了自定义mipmap
技术,利用kokomi.PackedMipMapGenerator生成了多个贴图的mipmap
自定义mipmap除了能捆绑贴图外,还有个好处就是可以动态控制贴图的模糊程度
const mipmapper = new kokomi.PackedMipMapGenerator(); const mirrorFBO = mirror.getRenderTarget(); const mipmapFBO = new kokomi.FBO(this); mirror.material.uniforms.tDiffuse.value = mipmapFBO.rt.texture; this.update(() => { mipmapper.update(mirrorFBO.texture, mipmapFBO.rt, this.renderer); });
vec2 p=vUv; vec2 texUv=p*uTexScale; texUv+=uTexOffset; float floorOpacity=texture(uOpacityTexture,texUv).r; vec3 floorNormal=texture(uNormalTexture,texUv).rgb*2.-1.; floorNormal=normalize(floorNormal); float roughness=texture(uRoughnessTexture,texUv).r; vec2 finalUv=reflectionUv+floorNormal.xy*uDistortionAmount-rainUv; float level=roughness*uBlurStrength; vec3 col=packedTexture2DLOD(tDiffuse,finalUv,level,uMipmapTextureSize).rgb;
下雨动画
生成雨滴
雨滴数量会很多,因此要用到THREE.InstancedMesh来生成实例化网格对象
const rain = new THREE.InstancedMesh(new THREE.PlaneGeometry(), rainMat, count); rain.instanceMatrix.needsUpdate = true; const dummy = new THREE.Object3D(); for (let i = 0; i < rain.count; i++) { dummy.position.set( THREE.MathUtils.randFloat(-10, 10), 0, THREE.MathUtils.randFloat(-20, 10) ); dummy.scale.set(0.03, THREE.MathUtils.randFloat(0.3, 0.5), 0.03); dummy.updateMatrix(); rain.setMatrixAt(i, dummy.matrix); } rain.rotation.set(-0.1, 0, 0.1); rain.position.set(0, 4, 4);
这里要注意一点:雨滴的方向是始终朝向用户的,为了达成这点就要用billboard方案来实现
vec3 billboard(vec3 v,mat4 view){ vec3 up=vec3(view[0][1],view[1][1],view[2][1]); vec3 right=vec3(view[0][0],view[1][0],view[2][0]); vec3 pos=right*v.x+up*v.y; return pos; } vec3 billboardPos=billboard(transformed,modelViewMatrix); transformed=billboardPos;
下落动画
我们可以给雨滴赋予随机的高度和速度attribute,并在顶点着色器中让它动起来
const progressArr = []; const speedArr = []; for (let i = 0; i < rain.count; i++) { ... progressArr.push(Math.random()); speedArr.push(dummy.scale.y * 10); } rain.geometry.setAttribute( "aProgress", new THREE.InstancedBufferAttribute(new Float32Array(progressArr), 1) ); rain.geometry.setAttribute( "aSpeed", new THREE.InstancedBufferAttribute(new Float32Array(speedArr), 1) );
attribute float aProgress; attribute float aSpeed; uniform float uSpeed; uniform float uHeightRange; vec3 distort(vec3 p){ float y=mod(aProgress-iTime*aSpeed*.25*uSpeed,1.)*uHeightRange-(uHeightRange*.5); p.y+=y; return p; } transformed=distort(transformed);
反射效果
创建背景的离屏渲染FBO,将其作为反射的主要材质
const bgFBO = new kokomi.FBO(this, { width: window.innerWidth * 0.1, height: window.innerHeight * 0.1, }); rainMat.uniforms.uBgRt.value = bgFBO.rt.texture; const fboCamera = this.camera.clone(); this.update(() => { rain.visible = false; this.renderer.setRenderTarget(bgFBO.rt); this.renderer.render(this.scene, fboCamera); this.renderer.setRenderTarget(null); rain.visible = true; });
在顶点着色器中获取屏幕空间vScreenspace
// https://github.com/Samsy/glsl-screenspace vec2 screenspace(mat4 projectionmatrix,mat4 modelviewmatrix,vec3 position){ vec4 temp=projectionmatrix*modelviewmatrix*vec4(position,1.); temp.xyz/=temp.w; temp.xy=(.5)+(temp.xy)*.5; return temp.xy; } vScreenspace=screenspace(projectionMatrix,modelViewMatrix,transformed);
在片元着色器中采样反射材质
uniform sampler2D uNormalTexture; uniform sampler2D uBgRt; uniform float uRefraction; uniform float uBaseBrightness; varying vec2 vScreenspace; void main(){ vec2 p=vUv; vec4 normalColor=texture(uNormalTexture,p); if(normalColor.a<.5){ discard; } vec3 normal=normalize(normalColor.rgb); vec2 bgUv=vScreenspace+normal.xy*uRefraction; vec4 bgColor=texture(uBgRt,bgUv); float brightness=uBaseBrightness*pow(normal.b,10.); vec3 col=bgColor.rgb+vec3(brightness); col=vec3(p,0.); gl_FragColor=vec4(col,1.); }
这里有一点要注意:积水地面中要把雨滴的反射去掉,不然会看着很乱
rainFloor.mirror.ignoreObjects.push(rain);
灯光闪烁
用setInterval
来间歇地设置文字和灯光材质的颜色即可
// flicker const turnOffLight = () => { tm.material.color.copy(new THREE.Color("black")); pointLight1.color.copy(new THREE.Color("black")); }; const turnOnLight = () => { tm.material.color.copy(new THREE.Color(config.color)); pointLight1.color.copy(new THREE.Color(config.color)); }; let flickerTimer = null; const flicker = () => { flickerTimer = setInterval(async () => { const rate = Math.random(); if (rate < 0.5) { turnOffLight(); await kokomi.sleep(200 * Math.random()); turnOnLight(); await kokomi.sleep(200 * Math.random()); turnOffLight(); await kokomi.sleep(200 * Math.random()); turnOnLight(); } }, 3000); }; flicker();
后期处理
为了让文字灯光看上去更加明亮,可以用Bloom
滤镜来照亮文字
由于后期处理中原先renderer的抗锯齿会失效,故用SMAA
滤镜来实现抗锯齿
// postprocessing const composer = new POSTPROCESSING.EffectComposer(this.renderer); this.composer = composer; composer.addPass(new POSTPROCESSING.RenderPass(this.scene, this.camera)); // bloom const bloom = new POSTPROCESSING.BloomEffect({ luminanceThreshold: 0.4, luminanceSmoothing: 0, mipmapBlur: true, intensity: 2, radius: 0.4, }); composer.addPass(new POSTPROCESSING.EffectPass(this.camera, bloom)); // antialiasing const smaa = new POSTPROCESSING.SMAAEffect(); composer.addPass(new POSTPROCESSING.EffectPass(this.camera, smaa));
待优化
效果算是基本实现了,但也有很多待优化的点
- 添加现实中的雨声
- 实现更棒的相机交互
- 添加更多的物体