示例

HTML

<script src="https://cdnjs.cloudflare.com/ajax/libs/three.js/88/three.min.js"></script>
<script id="vertexShader" type="x-shader/x-vertex">void main() {gl_Position = vec4( position, 1.0 );}
</script>
<script id="fragmentShader" type="x-shader/x-fragment">uniform vec2 u_resolution;uniform float u_time;uniform vec2 u_mouse;const int octaves = 2;const float seed = 43758.5453123;const float seed2 = 73156.8473192;// Epsilon value, used for normal gradient generationconst float eps = 0.00015;const vec3 ambientLight = 0.99 * vec3(1.0, 1.0, 1.0);const vec3 light1Pos = vec3(10., 5.0, -25.0);const vec3 light1Intensity = vec3(0.35);const vec3 light2Pos = vec3(-20., -25.0, 85.0);const vec3 light2Intensity = vec3(0.2);const vec3 colour_fog = vec3(.1, .1, .15);// movement variablesvec3 movement = vec3(.0);// Gloable variables for the raymarching algorithm.// const int maxIterations = 1024;// const int maxIterationsShad = 16;// const float stepScale = .2;// const float stopThreshold = 0.01;const int maxIterations = 512;const int maxIterationsShad = 16;const float stepScale = .8;const float stopThreshold = 0.01;float offset;mat2 rot2( float a ){ vec2 v = sin(vec2(1.570796, 0) + a);   return mat2(v, -v.y, v.x); }mat4 rotationMatrix(vec3 axis, float angle){axis = normalize(axis);float s = sin(angle);float c = cos(angle);float oc = 1.0 - c;return mat4(oc * axis.x * axis.x + c,           oc * axis.x * axis.y - axis.z * s,  oc * axis.z * axis.x + axis.y * s,  0.0,oc * axis.x * axis.y + axis.z * s,  oc * axis.y * axis.y + c,           oc * axis.y * axis.z - axis.x * s,  0.0,oc * axis.z * axis.x - axis.y * s,  oc * axis.y * axis.z + axis.x * s,  oc * axis.z * axis.z + c,           0.0,0.0,                                0.0,                                0.0,                                1.0);}float length2( vec2 p ){return sqrt( p.x*p.x + p.y*p.y );}float length6( vec2 p ){p = p*p*p; p = p*p;return pow( p.x + p.y, 1.0/6.0 );}float length8( vec2 p ){p = p*p; p = p*p; p = p*p;return pow( p.x + p.y, 1.0/8.0 );}vec3 camPath(float z) {return vec3(cos(z * .2) * sin(z * .2), sin(z * .3) * .6, z);}float tri(in float x){return abs(fract(x)-.5);}vec3 tri3(in vec3 p){return vec3( tri(p.z+tri(p.y*1.)), tri(p.z+tri(p.x*1.)), tri(p.y+tri(p.x*1.)));}mat2 m2 = mat2(0.970,  0.242, -0.242,  0.970);float triNoise3d(in vec3 p, in float spd){float z=1.4;float rz = 0.;vec3 bp = p;for (float i=0.; i<=3.; i++ ){vec3 dg = tri3(bp*2.);p += (dg+u_time*spd);bp *= 1.8;z *= 1.5;p *= 1.2;//p.xz*= m2;rz+= (tri(p.z+tri(p.x+tri(p.y))))/z;bp += 0.14;}return rz;}// smooth min// reference: http://iquilezles.org/www/articles/smin/smin.htmfloat smin(float a, float b, float k) {float res = exp(-k*a) + exp(-k*b);return -log(res)/k;}vec3 random3( vec3 p ) {return fract(sin(vec3(dot(p,vec3(127.1,311.7,319.8)),dot(p,vec3(269.5,183.3, 415.2)),dot(p,vec3(362.9,201.5,134.7))))*43758.5453);}vec2 random2( vec2 p ) {return fract(sin(vec2(dot(p,vec2(127.1,311.7)),dot(p,vec2(269.5,183.3))))*43758.5453);}// The world!float world_sdf(in vec3 p, inout float _offset) {float world = 10.;p.xy -= camPath(p.z).xy;_offset = ( triNoise3d(p * .1, 0.03) * 1.3);_offset -= smoothstep(0.5, 1., triNoise3d((p - 50.) * .01, 0.) * 2.);float n = triNoise3d(p * .01, 0.02);_offset -= smoothstep(0.8, 1., n * 2.);// _offset += smoothstep(0.3, .1, n * 1.);// p = floor(p * 5.) / 5.;float floor = p.y + .6;world = 1. - length(p.xy) * (1. + _offset);// world = 1. - length(triNoise3d(p, 0.));world = smin(world, floor, 50.);// world = min(world, floor);// world = floor;offset = _offset * .6;return world;}float world_sdf(in vec3 p) {float _offset = 0.;return world_sdf(p, _offset);}// Fuck yeah, normals!vec3 calculate_normal(in vec3 p){const vec3 small_step = vec3(0.0001, 0.0, 0.0);float gradient_x = world_sdf(vec3(p.x + eps, p.y, p.z)) - world_sdf(vec3(p.x - eps, p.y, p.z));float gradient_y = world_sdf(vec3(p.x, p.y + eps, p.z)) - world_sdf(vec3(p.x, p.y - eps, p.z));float gradient_z = world_sdf(vec3(p.x, p.y, p.z  + eps)) - world_sdf(vec3(p.x, p.y, p.z - eps));vec3 normal = vec3(gradient_x, gradient_y, gradient_z);return normalize(normal);}// Raymarching.float rayMarching( vec3 origin, vec3 dir, float start, float end, inout float field ) {float sceneDist = 1e4;float rayDepth = start;for ( int i = 0; i < maxIterations; i++ ) {sceneDist = world_sdf( origin + dir * rayDepth ); // Distance from the point along the ray to the nearest surface point in the scene.if (( sceneDist < stopThreshold ) || (rayDepth >= end)) {        break;}// We haven't hit anything, so increase the depth by a scaled factor of the minimum scene distance.rayDepth += sceneDist * stepScale;}if ( sceneDist >= stopThreshold ) rayDepth = end;else rayDepth += sceneDist;// We've used up our maximum iterations. Return the maximum distance.return rayDepth;}// The bump mapping function that gives the surface its bumpiness. Here, we're using two layers of a cheap 3D noise function. "Cheap" is a relative// term, hence only two layers. Four layers of Perlin noise would be much nicer, but that can be achieved using textures, which I'll get to soon. // I've provided an example on my site that uses a 3D Voronoi function which is nicer still.float bumpFunction(in vec3 p){return triNoise3d(p * 4., 0.05);// return texture2D(u_noise, p.xz).x * texture2D(u_noise, p.zy).x;}// Bump mapping uses similar techniques to those used when obtaining the surface normal. Use the surface point "p" and epsilon value "eps" to obtain // the gradient along each of the individual axes (f(p.x+eps)-f(p.x), etc). That should give you a scaled representation of (df/dx, df/dy, df/dz), // which can be scaled down to a desirable factor. However, this time, instead of indexing the distance function, we're indexing into a bump function - // which can be anything that takes in a 3D point and returns a value. The function can also use a texture, which speedwise, can often be // preferable, but for now we're using a 3D function.  //// After the scaled gradient of the bump mapping function is attained, we simply combine it with the surface normal "nor" using a "bump factor"// - which, oddly enough, controls the bumpiness of the surface - normalize it, then return the new bumped, or perturbed, normal.//// I should probably point out that this is not the only way to bump map a surface. In fact, there are a few. For speed and efficiency, I prefer to // index into a 2D texture (or 2D function), but that tends to involve textures and uv-mapping, which I didn't want to cover just yet, so I've taken // a different route. However, I'll be dealing with bump mapping from textures next. vec3 doBumpMap( in vec3 p, in vec3 nor, float bumpfactor ){//p = mod(p,1.0)-0.5; // If you want all the objects to be bump mapped with the same pattern, uncomment this.float ref = bumpFunction( p );// Note: To save on calculations, we're stepping to just one side of the position "p," rather than both.vec3 grad = vec3( bumpFunction(vec3(p.x+eps, p.y, p.z))-ref,bumpFunction(vec3(p.x, p.y+eps, p.z))-ref,bumpFunction(vec3(p.x, p.y, p.z+eps))-ref )/eps;grad -= nor*dot(nor, grad);return normalize( nor - bumpfactor*grad );}/*** Lighting* This stuff is way way better than the model I was using.* Courtesy Shane Warne* Reference: http://raymarching.com/* -------------------------------------* */// Lighting.vec3 lighting( vec3 sp, vec3 camPos, int reflectionPass, float dist, float field, vec3 rd) {// Start with black.vec3 sceneColor = vec3(0.0);vec3 objColor = vec3(1.0, .5, .5);objColor = vec3(clamp(offset, 0., 1.));// Obtain the surface normal at the scene position "sp."vec3 surfNormal = calculate_normal(sp);surfNormal = doBumpMap(sp, surfNormal, .01);// Lighting.// lp - Light position. Keeping it in the vacinity of the camera, but away from the objects in the scene.vec3 lp = vec3(0., 0.0, -1.0) + movement;// ld - Light direction.vec3 ld = lp-sp;// lcolor - Light color.vec3 lcolor = mix(vec3(.9,0.89,0.92) * .8, vec3(1., 1., 1.), offset);// Light falloff (attenuation).float len = length( ld ); // Distance from the light to the surface point.ld /= len; // Normalizing the light-to-surface, aka light-direction, vector.// float lightAtten = min( 1.0 / ( 0.15*len*len ), 1.0 ); // Removed light attenuation for this because I want the fade to whitefloat sceneLen = length(camPos - sp); // Distance of the camera to the surface pointfloat sceneAtten = min( 1.0 / ( 0.25*sceneLen ), 1.0 ); // Keeps things between 0 and 1.   float ambient = .5; //The object's ambient property.float diffuse = max( 0.0, dot(surfNormal, ld) ); //The object's diffuse value.// Bringing all the lighting components togethr to color the screen pixel.sceneColor += (objColor*(diffuse*0.8+ambient))*lcolor*1.3;sceneColor = mix(sceneColor, colour_fog, 1.-sceneAtten*sceneAtten); // fogreturn sceneColor;}void main() {// Setting up our screen coordinates.vec2 aspect = vec2(u_resolution.x/u_resolution.y, 1.0); //vec2 uv = (2.0*gl_FragCoord.xy/u_resolution.xy - 1.0)*aspect;// This just gives us a touch of fisheye// uv *= 1. + dot(uv, uv) * 0.4;// movementmovement = camPath(u_time * 2.5);// The sin in here is to make it look like a walk.vec3 lookAt = vec3(-0., 0.2, 1.);  // This is the point you look towards, or at, if you prefer.vec3 camera_position = vec3(0., 0., -1.0); // This is the point you look from, or camera you look at the scene through. Whichever way you wish to look at it.camera_position.y += abs(sin(u_time * 10.)) * .02;lookAt += camPath(u_time * 2.5 + 1. + sin(u_time) * 2.);// lookAt.z += sin(u_time / 10.) * .5;// lookAt.x += cos(u_time / 10.) * .5;camera_position += movement;vec3 forward = normalize(lookAt-camera_position); // Forward vector.vec3 right = normalize(vec3(forward.z, 0., -forward.x )); // Right vector... or is it left? Either way, so long as the correct-facing up-vector is produced.vec3 up = normalize(cross(forward,right)); // Cross product the two vectors above to get the up vector.// FOV - Field of view.float FOV = 0.4;// ro - Ray origin.vec3 ro = camera_position; // rd - Ray direction.vec3 rd = normalize(forward + FOV*uv.x*right + FOV*uv.y*up);rd.xy = rot2( camPath(lookAt.z).x * .1 )*rd.xy;// Ray marching.const float clipNear = 0.0;const float clipFar = 32.0;float field = 0.;float dist = rayMarching(ro, rd, clipNear, clipFar, field );if ( dist >= clipFar ) {gl_FragColor = vec4(colour_fog, 1.0);} else {// sp - Surface position. If we've made it this far, we've hit something.vec3 sp = ro + rd*dist;// Light the pixel that corresponds to the surface position. The last entry indicates that it's not a reflection pass// which we're not up to yet.vec3 sceneColor = lighting( sp, camera_position, 0, dist, field, rd);// Clamping the lit pixel, then put it on the screen.gl_FragColor = vec4(clamp(sceneColor, 0.0, 1.0), 1.0);}// gl_FragColor.rgb *= mat3(//   .3, 0.3, 0.2,//   0., .8, 0.,//   0.1, 0.1 - .2 * sin(u_time * .3), .8 + .2 * sin(u_time * .2)// );// gl_FragColor.rgb *= mat3(//   u_mouse.x + 1., 0., 0.,//   0., u_mouse.x + 1., 0.,//   0., 0., u_mouse.x + 1.// );// gl_FragColor.rgb *= mat3(//   1., 0., u_mouse.x,//   0., 1., 0.,//   u_mouse.x, 0., 1.// );}
</script><div id="container" touch-action="none"></div>

CSS

body {margin: 0;padding: 0;
}#container {position: fixed;touch-action: none;
}

JS

/*
Most of the stuff in here is just bootstrapping. Essentially it's just
setting ThreeJS up so that it renders a flat surface upon which to draw
the shader. The only thing to see here really is the uniforms sent to
the shader. Apart from that all of the magic happens in the HTML view
under the fragment shader.
*/let container;
let camera, scene, renderer;
let uniforms;let loader=new THREE.TextureLoader();
let texture;
loader.setCrossOrigin("anonymous");
loader.load('https://s3-us-west-2.amazonaws.com/s.cdpn.io/982762/noise.png',function do_something_with_texture(tex) {texture = tex;texture.wrapS = THREE.RepeatWrapping;texture.wrapT = THREE.RepeatWrapping;texture.minFilter = THREE.LinearFilter;init();animate();}
);function init() {container = document.getElementById( 'container' );camera = new THREE.Camera();camera.position.z = 1;scene = new THREE.Scene();var geometry = new THREE.PlaneBufferGeometry( 2, 2 );uniforms = {u_time: { type: "f", value: 1.0 },u_resolution: { type: "v2", value: new THREE.Vector2() },u_noise: { type: "t", value: texture },u_mouse: { type: "v2", value: new THREE.Vector2() }};var material = new THREE.ShaderMaterial( {uniforms: uniforms,vertexShader: document.getElementById( 'vertexShader' ).textContent,fragmentShader: document.getElementById( 'fragmentShader' ).textContent} );material.extensions.derivatives = true;var mesh = new THREE.Mesh( geometry, material );scene.add( mesh );renderer = new THREE.WebGLRenderer();// renderer.setPixelRatio( window.devicePixelRatio );container.appendChild( renderer.domElement );onWindowResize();window.addEventListener( 'resize', onWindowResize, false );document.addEventListener('pointermove', (e)=> {let ratio = window.innerHeight / window.innerWidth;uniforms.u_mouse.value.x = (e.pageX - window.innerWidth / 2) / window.innerWidth / ratio;uniforms.u_mouse.value.y = (e.pageY - window.innerHeight / 2) / window.innerHeight * -1;e.preventDefault();});
}function onWindowResize( event ) {renderer.setSize( window.innerWidth, window.innerHeight );uniforms.u_resolution.value.x = renderer.domElement.width;uniforms.u_resolution.value.y = renderer.domElement.height;
}function animate(delta) {requestAnimationFrame( animate );render(delta);
}let capturer = new CCapture( { verbose: true, framerate: 30,// motionBlurFrames: 4,quality: 90,format: 'webm',workersPath: 'js/'} );
let capturing = false;isCapturing = function(val) {if(val === false && window.capturing === true) {capturer.stop();capturer.save();} else if(val === true && window.capturing === false) {capturer.start();}capturing = val;
}
toggleCapture = function() {isCapturing(!capturing);
}window.addEventListener('keyup', function(e) { if(e.keyCode == 68) toggleCapture(); });let then = 0;
function render(delta) {uniforms.u_time.value = delta * 0.0005;renderer.render( scene, camera );if(capturing) {capturer.capture( renderer.domElement );}
}

穿越（pass through）相关推荐

配置Open***使用User/Pass方式验证登录
Open***和PPTP ***相比存在诸多的优势,最明显的是Open***支持NAT穿越,也就是说在nat环境下使用open***只需要一个在路由器上做一个端口映射即可!不需要其他路由的支持,要知道 ...
loginform登录表单 vs_【VSRC唯科普】穿越网页表单与登录窗口的采集（第9/14篇）...
原标题:[VSRC唯科普]穿越网页表单与登录窗口的采集(第9/14篇) 鸣谢 VSRC感谢业界小伙伴-- Mils投稿精品科普类文章.VSRC欢迎精品原创类文章投稿,优秀文章一旦采纳发布,将有好礼相 ...
NAT会话穿越应用程序（STUN）（RFC-5389）
RFC文档链接摘要 NAT会话穿越应用程序 (STUN) 是一种协议,在处理网络地址转换(Network Address Translator, NAT)穿越时充当其他协议的工具.终端可以使用它来确 ...
穿越NAT的p2p通信方法研究
穿越NAT的p2p通信方法研究日期:2008-12-08 来源:P2P网作者:未知字体:大中小 <script src="http://www.ppcn.net/ads/b ...
pass基础架构分析
pass基础架构分析 Relay 和 TVM IR,包含一系列优化passes,可提高模型的性能指标,如平均推理,内存占用,或特定设备的功耗.有一套标准优化,及特定机器学习的优化,包括常量折叠,死代码 ...
将编译器pass添加到Relay
将编译器pass添加到Relay 编译器pass是扩展Relay功能集和对Relay程序执行优化的主要接口.通过编写编译器pass,可以修改AST或收集有关AST的信息,具体取决于目标.事实上,Rel ...
AI中pass架构设计优化
AI中pass架构设计优化 Relay 和 TVM IR,包含一系列优化passes,可提高模型的性能指标,例如平均推理,内存占用,或特定设备的功耗.有一套标准优化,及特定机器学习的优化,包括常量折叠 ...
TVM，Relay，Pass
TVM,Relay,Pass Relay介绍主要结合TVM的文档(https://tvm.apache.org/docs/dev/relay_intro.html),介绍一下NNVM的第二代Rela ...
Pass算子python 函数
Pass算子python 函数函数 • 函数是代码的一种组织形式 • 函数应该能完成一项特定的工作,而且一般一个函数只完成一项工作 • 有些语言,分函数和过程两个概念,通俗解释是,有返回结果的是函数 ...

穿越（pass through）

穿越（pass through）

示例

HTML

CSS

JS

穿越（pass through）相关推荐

最新文章

热门文章