WebGPU学习(八):学习“texturedCube”示例

大家好,本文学习Chrome->webgpu-samplers->texturedCube示例。

上一篇博文:
WebGPU学习(七):学习“twoCubes”和“instancedCube”示例

学习texturedCube.ts

最终渲染结果:
WebGPU学习(八):学习“texturedCube”示例

该示例绘制了有一个纹理的立方体。

与“rotatingCube”示例相比,该示例增加了下面的步骤:

  • 传输顶点的uv数据
  • 增加了sampler和sampled-texture类型的uniform数据

下面,我们打开texturedCube.ts文件,依次分析增加的步骤:

传递顶点的uv数据

  • shader加入uv attribute

代码如下:

  const vertexShaderGLSL = `#version 450
  ...
  layout(location = 0) in vec4 position;
  layout(location = 1) in vec2 uv;

  layout(location = 0) out vec2 fragUV;
  layout(location = 1) out vec4 fragPosition;

  void main() {
    fragPosition = 0.5 * (position + vec4(1.0));
    ...
    fragUV = uv;
  }
  `;
  
  const fragmentShaderGLSL = `#version 450
  layout(set = 0, binding = 1) uniform sampler mySampler;
  layout(set = 0, binding = 2) uniform texture2D myTexture;

  layout(location = 0) in vec2 fragUV;
  layout(location = 1) in vec4 fragPosition;
  layout(location = 0) out vec4 outColor;

  void main() {
    outColor =  texture(sampler2D(myTexture, mySampler), fragUV) * fragPosition;
  }
  `;

vertex shader传入了uv attribute数据,并将其传递给fragUV,从而传到fragment shader,作为纹理采样坐标

另外,这里可以顺便说明下:fragPosition用来实现与position相关的颜色渐变效果

  • uv数据包含在verticesBuffer的cubeVertexArray中

cubeVertexArray的代码如下:

cube.ts:
export const cubeUVOffset = 4 * 8;
export const cubeVertexArray = new Float32Array([
    // float4 position, float4 color, float2 uv,
    1, -1, 1, 1,   1, 0, 1, 1,  1, 1,
    -1, -1, 1, 1,  0, 0, 1, 1,  0, 1,
    -1, -1, -1, 1, 0, 0, 0, 1,  0, 0,
    1, -1, -1, 1,  1, 0, 0, 1,  1, 0,
    1, -1, 1, 1,   1, 0, 1, 1,  1, 1,
    -1, -1, -1, 1, 0, 0, 0, 1,  0, 0,

    ...
]);

创建和设置verticesBuffer的相关代码如下:

texturedCube.ts:
  const verticesBuffer = device.createBuffer({
    size: cubeVertexArray.byteLength,
    usage: GPUBufferUsage.VERTEX | GPUBufferUsage.COPY_DST
  });
  verticesBuffer.setSubData(0, cubeVertexArray);
  
  ...
  
  return function frame() {
    ...
    passEncoder.setVertexBuffer(0, verticesBuffer);
    ...
  } 
  • 创建render pipeline时指定uv attribute的相关数据

代码如下:

  const pipeline = device.createRenderPipeline({
    ...
    vertexState: {
      vertexBuffers: [{
        ...
        attributes: [
        ...
        {
          // uv
          shaderLocation: 1,
          offset: cubeUVOffset,
          format: "float2"
        }]
      }],
    },
    ...
  });    

增加了sampler和sampled-texture类型的uniform数据

WebGPU相对于WebGL1,提出了sampler,可以对它设置filter、wrap等参数,从而实现了texture和sampler*组合,同一个texture能够以不同filter、wrap来采样

  • fragment shader传入这两个uniform数据,用于纹理采样

代码如下:

  const fragmentShaderGLSL = `#version 450
  layout(set = 0, binding = 1) uniform sampler mySampler;
  layout(set = 0, binding = 2) uniform texture2D myTexture;

  layout(location = 0) in vec2 fragUV;
  layout(location = 1) in vec4 fragPosition;
  layout(location = 0) out vec4 outColor;

  void main() {
    outColor =  texture(sampler2D(myTexture, mySampler), fragUV) * fragPosition;
  }
  `;
  • 创建bind group layout时指定它们在shader中的binding位置等参数

代码如下:

  const bindGroupLayout = device.createBindGroupLayout({
    bindings: [
    ...
    {
      // Sampler
      binding: 1,
      visibility: GPUShaderStage.FRAGMENT,
      type: "sampler"
    }, {
      // Texture view
      binding: 2,
      visibility: GPUShaderStage.FRAGMENT,
      type: "sampled-texture"
    }]
  });
  • 拷贝图片到texture,返回texture

代码如下,后面会进一步研究:

  const cubeTexture = await createTextureFromImage(device, 'assets/img/Di-3d.png', GPUTextureUsage.SAMPLED);
  • 创建sampler,指定filter

代码如下:

  const sampler = device.createSampler({
    magFilter: "linear",
    minFilter: "linear",
  });
  

我们看一下相关定义:

GPUSampler createSampler(optional GPUSamplerDescriptor descriptor = {});

...

dictionary GPUSamplerDescriptor : GPUObjectDescriptorBase {
    GPUAddressMode addressModeU = "clamp-to-edge";
    GPUAddressMode addressModeV = "clamp-to-edge";
    GPUAddressMode addressModeW = "clamp-to-edge";
    GPUFilterMode magFilter = "nearest";
    GPUFilterMode minFilter = "nearest";
    GPUFilterMode mipmapFilter = "nearest";
    float lodMinClamp = 0;
    float lodMaxClamp = 0xffffffff;
    GPUCompareFunction compare = "never";
};

GPUSamplerDescriptor的addressMode指定了texture在u、v、w方向的wrap mode(u、v方向的wrap相当于WebGL1的wrapS、wrapT)(w方向是给3d texture用的)

mipmapFilter与mipmap有关,lodXXX与texture lod有关,compare与软阴影的Percentage Closer Filtering技术有关,我们不讨论它们

  • 创建uniform bind group时传入sampler和texture的view
  const uniformBindGroup = device.createBindGroup({
    layout: bindGroupLayout,
    bindings: [
    ...
    {
      binding: 1,
      resource: sampler,
    }, {
      binding: 2,
      resource: cubeTexture.createView(),
    }],
  });

参考资料

Sampler Object

详细分析“拷贝图片到texture”步骤

相关代码如下:

  const cubeTexture = await createTextureFromImage(device, 'assets/img/Di-3d.png', GPUTextureUsage.SAMPLED);

该步骤可以分解为两步:
1.解码图片
2.拷贝解码后的类型为HTMLImageElement的图片到GPU的texture中

下面依次分析:

解码图片

打开helper.ts文件,查看createTextureFromImage对应代码:

  const img = document.createElement('img');
  img.src = src;
  await img.decode();

这里使用decode api来解码图片,也可以使用img.onload来实现:

  const img = document.createElement('img');
  img.src = src;
  img.onload = (img) => {
    ...
  };

根据Pre-Loading and Pre-Decoding Images with Javascript for Better Performance的说法,图片的加载过程有两个步骤:
1.从服务器加载图片
2.解码图片

第1步都是在其它线程上并行执行;
如果用onload,则浏览器会在主线程上同步执行第2步,会阻塞主线程;
如果用decode api,则浏览器会在其它线程上并行执行第2步,不会阻塞主线程。

chrome和firefox浏览器都支持decode api,因此加载图片应该优先使用该API:
WebGPU学习(八):学习“texturedCube”示例

参考资料

Pre-Loading and Pre-Decoding Images with Javascript for Better Performance
Chrome 图片解码与 Image.decode API

拷贝图片

WebGL1直接使用texImage2D将图片上传到GPU texture中,而WebGPU能让我们更加灵活地控制上传过程。

WebGPU有两种方法上传:

  • 创建图片对应的imageBitmap,将其拷贝到GPU texture中

该方法要用到copyImageBitmapToTexture函数。虽然WebGPU规范已经定义了该函数,但目前Chrome Canary不支持它,所以暂时不能用该方法上传。

参考资料
Proposal for copyImageBitmapToTexture
ImageBitmapToTexture design

  • 将图片绘制到canvas中,通过getImageData获得数据->将其设置到buffer中->把buffer数据拷贝到GPU texture中

我们来看下createTextureFromImage对应代码:

  const imageCanvas = document.createElement('canvas');
  imageCanvas.width = img.width;
  imageCanvas.height = img.height;

  const imageCanvasContext = imageCanvas.getContext('2d');
  
  //flipY
  imageCanvasContext.translate(0, img.height);
  imageCanvasContext.scale(1, -1);
  
  imageCanvasContext.drawImage(img, 0, 0, img.width, img.height);
  const imageData = imageCanvasContext.getImageData(0, 0, img.width, img.height);

这里创建canvas->绘制图片->获得图片数据。
(注:在绘制图片时将图片在Y方向反转了)

接着看代码:

  let data = null;

  const rowPitch = Math.ceil(img.width * 4 / 256) * 256;
  if (rowPitch == img.width * 4) {
    data = imageData.data;
  } else {
    data = new Uint8Array(rowPitch * img.height);
    for (let y = 0; y < img.height; ++y) {
      for (let x = 0; x < img.width; ++x) {
        let i = x * 4 + y * rowPitch;
        data[i] = imageData.data[i];
        data[i + 1] = imageData.data[i + 1];
        data[i + 2] = imageData.data[i + 2];
        data[i + 3] = imageData.data[i + 3];
      }
    }
  }

  const texture = device.createTexture({
    size: {
      width: img.width,
      height: img.height,
      depth: 1,
    },
    format: "rgba8unorm",
    usage: GPUTextureUsage.COPY_DST | usage,
  });

  const textureDataBuffer = device.createBuffer({
    size: data.byteLength,
    usage: GPUBufferUsage.COPY_DST | GPUBufferUsage.COPY_SRC,
  });

  textureDataBuffer.setSubData(0, data);

rowPitch需要为256的倍数(也就是说,图片的宽度需要为64px的倍数),这是因为Dx12对此做了限制(参考Copies investigation):

RowPitch must be aligned to D3D12_TEXTURE_DATA_PITCH_ALIGNMENT.
Offset must be aligned to D3D12_TEXTURE_DATA_PLACEMENT_ALIGNMENT, which is 512.

另外,关于纹理尺寸,可以参考WebGPU-6

第一个问题是关于纹理尺寸的,回答是WebGPU没有对尺寸有特别明确的要求。sample code中最多不能比4kor8k大就行。这个也不是太难理解,OpenGL对纹理和FBO的尺寸总是有上限的。

根据我的测试,buffer(代码中的textureDataBuffer)中的图片数据需要为未压缩的图片数据(它的类型为Uint8Array,length=img.width * img.height * 4(因为每个像素有r、g、b、a这4个值)),否则会报错(在我的测试中,“通过canvas->toDataURL得到图片的base64->将其转为Uint8Array,得到压缩后的图片数据->将其设置到buffer中”会报错)

继续看代码:

  const commandEncoder = device.createCommandEncoder({});
  commandEncoder.copyBufferToTexture({
    buffer: textureDataBuffer,
    rowPitch: rowPitch,
    imageHeight: 0,
  }, {
    texture: texture,
  }, {
    width: img.width,
    height: img.height,
    depth: 1,
  });

  device.defaultQueue.submit([commandEncoder.finish()]);

  return texture;

这里提交了copyBufferToTexture这个command到GPU,并返回texture
(注:这个command此时并没有执行,会由GPU决定什么时候执行)

WebGPU支持buffer与buffer、buffer与texture、texture与texture之间互相拷贝。

参考资料
3 channel formats
Copies investigation (+ proposals)

参考资料

WebGPU规范
webgpu-samplers Github Repo
WebGPU-6

上一篇:WebGPU学习(十):介绍“GPU实现粒子效果”


下一篇:WebGPU学习(七):学习“twoCubes”和“instancedCube”示例