evotars 0.0.11 → 0.0.12

package/dist/WebGPURenderer-BPb_585F.js

@@ -0,0 +1,1561 @@
+import { m as R, g as E, E as l, f as ie, S as P, B as S, b as ne, c as A, d as b, h as oe, i as ae, D as ue, M as k, j as M, w as ce, e as G, A as de } from "./index-B5GUB2He.js";
+import { S as D, c as T, a as z, g as L, r as C, b as v, d as he, U as pe, l as le, e as fe, R as ge } from "./colorToUniform-Cp4M-zOy.js";
+import { c as me, u as _e, U as be, B as ye, G as xe, e as Ge, R as Be, t as Se, S as Pe, a as Te } from "./SharedSystems-n7sA1Ui-.js";
+import { C as w } from "./CanvasPool-5ekdWScU.js";
+const y = D.for2d();
+class O {
+  init() {
+    const e = T({
+      name: "batch",
+      bits: [
+        z,
+        L(R()),
+        C
+      ]
+    });
+    this._shader = new v({
+      gpuProgram: e,
+      groups: {
+        // these will be dynamically allocated
+      }
+    });
+  }
+  start(e, t) {
+    const r = e.renderer, s = r.encoder, i = this._shader.gpuProgram;
+    this._geometry = t, s.setGeometry(t), y.blendMode = "normal", r.pipeline.getPipeline(
+      t,
+      i,
+      y
+    );
+    const n = r.globalUniforms.bindGroup;
+    s.resetBindGroup(1), s.setBindGroup(0, n, i);
+  }
+  execute(e, t) {
+    const r = this._shader.gpuProgram, s = e.renderer, i = s.encoder;
+    if (!t.bindGroup) {
+      const u = t.textures;
+      t.bindGroup = E(u.textures, u.count);
+    }
+    y.blendMode = t.blendMode;
+    const n = s.bindGroup.getBindGroup(
+      t.bindGroup,
+      r,
+      1
+    ), o = s.pipeline.getPipeline(
+      this._geometry,
+      r,
+      y
+    );
+    t.bindGroup._touch(s.textureGC.count), i.setPipeline(o), i.renderPassEncoder.setBindGroup(1, n), i.renderPassEncoder.drawIndexed(t.size, 1, t.start);
+  }
+  destroy() {
+    this._shader.destroy(!0), this._shader = null;
+  }
+}
+O.extension = {
+  type: [
+    l.WebGPUPipesAdaptor
+  ],
+  name: "batch"
+};
+class H {
+  constructor(e) {
+    this._hash = /* @__PURE__ */ Object.create(null), this._renderer = e;
+  }
+  contextChange(e) {
+    this._gpu = e;
+  }
+  getBindGroup(e, t, r) {
+    return e._updateKey(), this._hash[e._key] || this._createBindGroup(e, t, r);
+  }
+  _createBindGroup(e, t, r) {
+    const s = this._gpu.device, i = t.layout[r], n = [], o = this._renderer;
+    for (const f in i) {
+      const h = e.resources[f] ?? e.resources[i[f]];
+      let p;
+      if (h._resourceType === "uniformGroup") {
+        const d = h;
+        o.ubo.updateUniformGroup(d);
+        const g = d.buffer;
+        p = {
+          buffer: o.buffer.getGPUBuffer(g),
+          offset: 0,
+          size: g.descriptor.size
+        };
+      } else if (h._resourceType === "buffer") {
+        const d = h;
+        p = {
+          buffer: o.buffer.getGPUBuffer(d),
+          offset: 0,
+          size: d.descriptor.size
+        };
+      } else if (h._resourceType === "bufferResource") {
+        const d = h;
+        p = {
+          buffer: o.buffer.getGPUBuffer(d.buffer),
+          offset: d.offset,
+          size: d.size
+        };
+      } else if (h._resourceType === "textureSampler") {
+        const d = h;
+        p = o.texture.getGpuSampler(d);
+      } else if (h._resourceType === "textureSource") {
+        const d = h;
+        p = o.texture.getGpuSource(d).createView({});
+      }
+      n.push({
+        binding: i[f],
+        resource: p
+      });
+    }
+    const u = o.shader.getProgramData(t).bindGroups[r], c = s.createBindGroup({
+      layout: u,
+      entries: n
+    });
+    return this._hash[e._key] = c, c;
+  }
+  destroy() {
+    for (const e of Object.keys(this._hash))
+      this._hash[e] = null;
+    this._hash = null, this._renderer = null;
+  }
+}
+H.extension = {
+  type: [
+    l.WebGPUSystem
+  ],
+  name: "bindGroup"
+};
+class F {
+  constructor() {
+    this._gpuBuffers = /* @__PURE__ */ Object.create(null), this._managedBuffers = [];
+  }
+  contextChange(e) {
+    this._gpu = e;
+  }
+  getGPUBuffer(e) {
+    return this._gpuBuffers[e.uid] || this.createGPUBuffer(e);
+  }
+  updateBuffer(e) {
+    const t = this._gpuBuffers[e.uid] || this.createGPUBuffer(e), r = e.data;
+    return e._updateID && r && (e._updateID = 0, this._gpu.device.queue.writeBuffer(
+      t,
+      0,
+      r.buffer,
+      0,
+      // round to the nearest 4 bytes
+      (e._updateSize || r.byteLength) + 3 & -4
+    )), t;
+  }
+  /** dispose all WebGL resources of all managed buffers */
+  destroyAll() {
+    for (const e in this._gpuBuffers)
+      this._gpuBuffers[e].destroy();
+    this._gpuBuffers = {};
+  }
+  createGPUBuffer(e) {
+    this._gpuBuffers[e.uid] || (e.on("update", this.updateBuffer, this), e.on("change", this.onBufferChange, this), e.on("destroy", this.onBufferDestroy, this));
+    const t = this._gpu.device.createBuffer(e.descriptor);
+    return e._updateID = 0, e.data && (ie(e.data.buffer, t.getMappedRange()), t.unmap()), this._gpuBuffers[e.uid] = t, this._managedBuffers.push(e), t;
+  }
+  onBufferChange(e) {
+    this._gpuBuffers[e.uid].destroy(), e._updateID = 0, this._gpuBuffers[e.uid] = this.createGPUBuffer(e);
+  }
+  /**
+   * Disposes buffer
+   * @param buffer - buffer with data
+   */
+  onBufferDestroy(e) {
+    this._managedBuffers.splice(this._managedBuffers.indexOf(e), 1), this._destroyBuffer(e);
+  }
+  destroy() {
+    this._managedBuffers.forEach((e) => this._destroyBuffer(e)), this._managedBuffers = null, this._gpuBuffers = null;
+  }
+  _destroyBuffer(e) {
+    this._gpuBuffers[e.uid].destroy(), e.off("update", this.updateBuffer, this), e.off("change", this.onBufferChange, this), e.off("destroy", this.onBufferDestroy, this), this._gpuBuffers[e.uid] = null;
+  }
+}
+F.extension = {
+  type: [
+    l.WebGPUSystem
+  ],
+  name: "buffer"
+};
+class Ce {
+  constructor({ minUniformOffsetAlignment: e }) {
+    this._minUniformOffsetAlignment = 256, this.byteIndex = 0, this._minUniformOffsetAlignment = e, this.data = new Float32Array(65535);
+  }
+  clear() {
+    this.byteIndex = 0;
+  }
+  addEmptyGroup(e) {
+    if (e > this._minUniformOffsetAlignment / 4)
+      throw new Error(`UniformBufferBatch: array is too large: ${e * 4}`);
+    const t = this.byteIndex;
+    let r = t + e * 4;
+    if (r = Math.ceil(r / this._minUniformOffsetAlignment) * this._minUniformOffsetAlignment, r > this.data.length * 4)
+      throw new Error("UniformBufferBatch: ubo batch got too big");
+    return this.byteIndex = r, t;
+  }
+  addGroup(e) {
+    const t = this.addEmptyGroup(e.length);
+    for (let r = 0; r < e.length; r++)
+      this.data[t / 4 + r] = e[r];
+    return t;
+  }
+  destroy() {
+    this._buffer.destroy(), this._buffer = null, this.data = null;
+  }
+}
+class I {
+  constructor(e) {
+    this._colorMaskCache = 15, this._renderer = e;
+  }
+  setMask(e) {
+    this._colorMaskCache !== e && (this._colorMaskCache = e, this._renderer.pipeline.setColorMask(e));
+  }
+  destroy() {
+    this._renderer = null, this._colorMaskCache = null;
+  }
+}
+I.extension = {
+  type: [
+    l.WebGPUSystem
+  ],
+  name: "colorMask"
+};
+class U {
+  /**
+   * @param {WebGPURenderer} renderer - The renderer this System works for.
+   */
+  constructor(e) {
+    this._renderer = e;
+  }
+  async init(e) {
+    return this._initPromise ? this._initPromise : (this._initPromise = this._createDeviceAndAdaptor(e).then((t) => {
+      this.gpu = t, this._renderer.runners.contextChange.emit(this.gpu);
+    }), this._initPromise);
+  }
+  /**
+   * Handle the context change event
+   * @param gpu
+   */
+  contextChange(e) {
+    this._renderer.gpu = e;
+  }
+  /**
+   * Helper class to create a WebGL Context
+   * @param {object} options - An options object that gets passed in to the canvas element containing the
+   *    context attributes
+   * @see https://developer.mozilla.org/en/docs/Web/API/HTMLCanvasElement/getContext
+   * @returns {WebGLRenderingContext} the WebGL context
+   */
+  async _createDeviceAndAdaptor(e) {
+    const t = await navigator.gpu.requestAdapter({
+      powerPreference: e.powerPreference,
+      forceFallbackAdapter: e.forceFallbackAdapter
+    }), r = [
+      "texture-compression-bc",
+      "texture-compression-astc",
+      "texture-compression-etc2"
+    ].filter((i) => t.features.has(i)), s = await t.requestDevice({
+      requiredFeatures: r
+    });
+    return { adapter: t, device: s };
+  }
+  destroy() {
+    this.gpu = null, this._renderer = null;
+  }
+}
+U.extension = {
+  type: [
+    l.WebGPUSystem
+  ],
+  name: "device"
+};
+U.defaultOptions = {
+  /**
+   * {@link WebGPUOptions.powerPreference}
+   * @default default
+   */
+  powerPreference: void 0,
+  /**
+   * Force the use of the fallback adapter
+   * @default false
+   */
+  forceFallbackAdapter: !1
+};
+class W {
+  constructor(e) {
+    this._boundBindGroup = /* @__PURE__ */ Object.create(null), this._boundVertexBuffer = /* @__PURE__ */ Object.create(null), this._renderer = e;
+  }
+  renderStart() {
+    this.commandFinished = new Promise((e) => {
+      this._resolveCommandFinished = e;
+    }), this.commandEncoder = this._renderer.gpu.device.createCommandEncoder();
+  }
+  beginRenderPass(e) {
+    this.endRenderPass(), this._clearCache(), this.renderPassEncoder = this.commandEncoder.beginRenderPass(e.descriptor);
+  }
+  endRenderPass() {
+    this.renderPassEncoder && this.renderPassEncoder.end(), this.renderPassEncoder = null;
+  }
+  setViewport(e) {
+    this.renderPassEncoder.setViewport(e.x, e.y, e.width, e.height, 0, 1);
+  }
+  setPipelineFromGeometryProgramAndState(e, t, r, s) {
+    const i = this._renderer.pipeline.getPipeline(e, t, r, s);
+    this.setPipeline(i);
+  }
+  setPipeline(e) {
+    this._boundPipeline !== e && (this._boundPipeline = e, this.renderPassEncoder.setPipeline(e));
+  }
+  _setVertexBuffer(e, t) {
+    this._boundVertexBuffer[e] !== t && (this._boundVertexBuffer[e] = t, this.renderPassEncoder.setVertexBuffer(e, this._renderer.buffer.updateBuffer(t)));
+  }
+  _setIndexBuffer(e) {
+    if (this._boundIndexBuffer === e)
+      return;
+    this._boundIndexBuffer = e;
+    const t = e.data.BYTES_PER_ELEMENT === 2 ? "uint16" : "uint32";
+    this.renderPassEncoder.setIndexBuffer(this._renderer.buffer.updateBuffer(e), t);
+  }
+  resetBindGroup(e) {
+    this._boundBindGroup[e] = null;
+  }
+  setBindGroup(e, t, r) {
+    if (this._boundBindGroup[e] === t)
+      return;
+    this._boundBindGroup[e] = t, t._touch(this._renderer.textureGC.count);
+    const s = this._renderer.bindGroup.getBindGroup(t, r, e);
+    this.renderPassEncoder.setBindGroup(e, s);
+  }
+  setGeometry(e) {
+    for (const t in e.attributes) {
+      const r = e.attributes[t];
+      this._setVertexBuffer(r.location, r.buffer);
+    }
+    e.indexBuffer && this._setIndexBuffer(e.indexBuffer);
+  }
+  _setShaderBindGroups(e, t) {
+    for (const r in e.groups) {
+      const s = e.groups[r];
+      t || this._syncBindGroup(s), this.setBindGroup(r, s, e.gpuProgram);
+    }
+  }
+  _syncBindGroup(e) {
+    for (const t in e.resources) {
+      const r = e.resources[t];
+      r.isUniformGroup && this._renderer.ubo.updateUniformGroup(r);
+    }
+  }
+  draw(e) {
+    const { geometry: t, shader: r, state: s, topology: i, size: n, start: o, instanceCount: u, skipSync: c } = e;
+    this.setPipelineFromGeometryProgramAndState(t, r.gpuProgram, s, i), this.setGeometry(t), this._setShaderBindGroups(r, c), t.indexBuffer ? this.renderPassEncoder.drawIndexed(
+      n || t.indexBuffer.data.length,
+      u || t.instanceCount,
+      o || 0
+    ) : this.renderPassEncoder.draw(n || t.getSize(), u || t.instanceCount, o || 0);
+  }
+  finishRenderPass() {
+    this.renderPassEncoder && (this.renderPassEncoder.end(), this.renderPassEncoder = null);
+  }
+  postrender() {
+    this.finishRenderPass(), this._gpu.device.queue.submit([this.commandEncoder.finish()]), this._resolveCommandFinished(), this.commandEncoder = null;
+  }
+  // restores a render pass if finishRenderPass was called
+  // not optimised as really used for debugging!
+  // used when we want to stop drawing and log a texture..
+  restoreRenderPass() {
+    const e = this._renderer.renderTarget.adaptor.getDescriptor(
+      this._renderer.renderTarget.renderTarget,
+      !1,
+      [0, 0, 0, 1]
+    );
+    this.renderPassEncoder = this.commandEncoder.beginRenderPass(e);
+    const t = this._boundPipeline, r = { ...this._boundVertexBuffer }, s = this._boundIndexBuffer, i = { ...this._boundBindGroup };
+    this._clearCache();
+    const n = this._renderer.renderTarget.viewport;
+    this.renderPassEncoder.setViewport(n.x, n.y, n.width, n.height, 0, 1), this.setPipeline(t);
+    for (const o in r)
+      this._setVertexBuffer(o, r[o]);
+    for (const o in i)
+      this.setBindGroup(o, i[o], null);
+    this._setIndexBuffer(s);
+  }
+  _clearCache() {
+    for (let e = 0; e < 16; e++)
+      this._boundBindGroup[e] = null, this._boundVertexBuffer[e] = null;
+    this._boundIndexBuffer = null, this._boundPipeline = null;
+  }
+  destroy() {
+    this._renderer = null, this._gpu = null, this._boundBindGroup = null, this._boundVertexBuffer = null, this._boundIndexBuffer = null, this._boundPipeline = null;
+  }
+  contextChange(e) {
+    this._gpu = e;
+  }
+}
+W.extension = {
+  type: [l.WebGPUSystem],
+  name: "encoder",
+  priority: 1
+};
+class V {
+  constructor(e) {
+    this._renderTargetStencilState = /* @__PURE__ */ Object.create(null), this._renderer = e, e.renderTarget.onRenderTargetChange.add(this);
+  }
+  onRenderTargetChange(e) {
+    let t = this._renderTargetStencilState[e.uid];
+    t || (t = this._renderTargetStencilState[e.uid] = {
+      stencilMode: P.DISABLED,
+      stencilReference: 0
+    }), this._activeRenderTarget = e, this.setStencilMode(t.stencilMode, t.stencilReference);
+  }
+  setStencilMode(e, t) {
+    const r = this._renderTargetStencilState[this._activeRenderTarget.uid];
+    r.stencilMode = e, r.stencilReference = t;
+    const s = this._renderer;
+    s.pipeline.setStencilMode(e), s.encoder.renderPassEncoder.setStencilReference(t);
+  }
+  destroy() {
+    this._renderer.renderTarget.onRenderTargetChange.remove(this), this._renderer = null, this._activeRenderTarget = null, this._renderTargetStencilState = null;
+  }
+}
+V.extension = {
+  type: [
+    l.WebGPUSystem
+  ],
+  name: "stencil"
+};
+const x = {
+  i32: { align: 4, size: 4 },
+  u32: { align: 4, size: 4 },
+  f32: { align: 4, size: 4 },
+  f16: { align: 2, size: 2 },
+  "vec2<i32>": { align: 8, size: 8 },
+  "vec2<u32>": { align: 8, size: 8 },
+  "vec2<f32>": { align: 8, size: 8 },
+  "vec2<f16>": { align: 4, size: 4 },
+  "vec3<i32>": { align: 16, size: 12 },
+  "vec3<u32>": { align: 16, size: 12 },
+  "vec3<f32>": { align: 16, size: 12 },
+  "vec3<f16>": { align: 8, size: 6 },
+  "vec4<i32>": { align: 16, size: 16 },
+  "vec4<u32>": { align: 16, size: 16 },
+  "vec4<f32>": { align: 16, size: 16 },
+  "vec4<f16>": { align: 8, size: 8 },
+  "mat2x2<f32>": { align: 8, size: 16 },
+  "mat2x2<f16>": { align: 4, size: 8 },
+  "mat3x2<f32>": { align: 8, size: 24 },
+  "mat3x2<f16>": { align: 4, size: 12 },
+  "mat4x2<f32>": { align: 8, size: 32 },
+  "mat4x2<f16>": { align: 4, size: 16 },
+  "mat2x3<f32>": { align: 16, size: 32 },
+  "mat2x3<f16>": { align: 8, size: 16 },
+  "mat3x3<f32>": { align: 16, size: 48 },
+  "mat3x3<f16>": { align: 8, size: 24 },
+  "mat4x3<f32>": { align: 16, size: 64 },
+  "mat4x3<f16>": { align: 8, size: 32 },
+  "mat2x4<f32>": { align: 16, size: 32 },
+  "mat2x4<f16>": { align: 8, size: 16 },
+  "mat3x4<f32>": { align: 16, size: 48 },
+  "mat3x4<f16>": { align: 8, size: 24 },
+  "mat4x4<f32>": { align: 16, size: 64 },
+  "mat4x4<f16>": { align: 8, size: 32 }
+};
+function ve(a) {
+  const e = a.map((r) => ({
+    data: r,
+    offset: 0,
+    size: 0
+  }));
+  let t = 0;
+  for (let r = 0; r < e.length; r++) {
+    const s = e[r];
+    let i = x[s.data.type].size;
+    const n = x[s.data.type].align;
+    if (!x[s.data.type])
+      throw new Error(`[Pixi.js] WebGPU UniformBuffer: Unknown type ${s.data.type}`);
+    s.data.size > 1 && (i = Math.max(i, n) * s.data.size), t = Math.ceil(t / n) * n, s.size = i, s.offset = t, t += i;
+  }
+  return t = Math.ceil(t / 16) * 16, { uboElements: e, size: t };
+}
+function Ue(a, e) {
+  const { size: t, align: r } = x[a.data.type], s = (r - t) / 4;
+  return `
+         v = uv.${a.data.name};
+         ${e !== 0 ? `offset += ${e};` : ""}
+
+         arrayOffset = offset;
+
+         t = 0;
+
+         for(var i=0; i < ${a.data.size * (t / 4)}; i++)
+         {
+             for(var j = 0; j < ${t / 4}; j++)
+             {
+                 data[arrayOffset++] = v[t++];
+             }
+             ${s !== 0 ? `arrayOffset += ${s};` : ""}
+         }
+     `;
+}
+function Me(a) {
+  return me(
+    a,
+    "uboWgsl",
+    Ue,
+    _e
+  );
+}
+class j extends be {
+  constructor() {
+    super({
+      createUboElements: ve,
+      generateUboSync: Me
+    });
+  }
+}
+j.extension = {
+  type: [l.WebGPUSystem],
+  name: "ubo"
+};
+const _ = 128;
+class N {
+  constructor(e) {
+    this._bindGroupHash = /* @__PURE__ */ Object.create(null), this._buffers = [], this._bindGroups = [], this._bufferResources = [], this._renderer = e, this._batchBuffer = new Ce({ minUniformOffsetAlignment: _ });
+    const t = 256 / _;
+    for (let r = 0; r < t; r++) {
+      let s = S.UNIFORM | S.COPY_DST;
+      r === 0 && (s |= S.COPY_SRC), this._buffers.push(new ne({
+        data: this._batchBuffer.data,
+        usage: s
+      }));
+    }
+  }
+  renderEnd() {
+    this._uploadBindGroups(), this._resetBindGroups();
+  }
+  _resetBindGroups() {
+    for (const e in this._bindGroupHash)
+      this._bindGroupHash[e] = null;
+    this._batchBuffer.clear();
+  }
+  // just works for single bind groups for now
+  getUniformBindGroup(e, t) {
+    if (!t && this._bindGroupHash[e.uid])
+      return this._bindGroupHash[e.uid];
+    this._renderer.ubo.ensureUniformGroup(e);
+    const r = e.buffer.data, s = this._batchBuffer.addEmptyGroup(r.length);
+    return this._renderer.ubo.syncUniformGroup(e, this._batchBuffer.data, s / 4), this._bindGroupHash[e.uid] = this._getBindGroup(s / _), this._bindGroupHash[e.uid];
+  }
+  getUboResource(e) {
+    this._renderer.ubo.updateUniformGroup(e);
+    const t = e.buffer.data, r = this._batchBuffer.addGroup(t);
+    return this._getBufferResource(r / _);
+  }
+  getArrayBindGroup(e) {
+    const t = this._batchBuffer.addGroup(e);
+    return this._getBindGroup(t / _);
+  }
+  getArrayBufferResource(e) {
+    const r = this._batchBuffer.addGroup(e) / _;
+    return this._getBufferResource(r);
+  }
+  _getBufferResource(e) {
+    if (!this._bufferResources[e]) {
+      const t = this._buffers[e % 2];
+      this._bufferResources[e] = new ye({
+        buffer: t,
+        offset: (e / 2 | 0) * 256,
+        size: _
+      });
+    }
+    return this._bufferResources[e];
+  }
+  _getBindGroup(e) {
+    if (!this._bindGroups[e]) {
+      const t = new A({
+        0: this._getBufferResource(e)
+      });
+      this._bindGroups[e] = t;
+    }
+    return this._bindGroups[e];
+  }
+  _uploadBindGroups() {
+    const e = this._renderer.buffer, t = this._buffers[0];
+    t.update(this._batchBuffer.byteIndex), e.updateBuffer(t);
+    const r = this._renderer.gpu.device.createCommandEncoder();
+    for (let s = 1; s < this._buffers.length; s++) {
+      const i = this._buffers[s];
+      r.copyBufferToBuffer(
+        e.getGPUBuffer(t),
+        _,
+        e.getGPUBuffer(i),
+        0,
+        this._batchBuffer.byteIndex
+      );
+    }
+    this._renderer.gpu.device.queue.submit([r.finish()]);
+  }
+  destroy() {
+    for (let e = 0; e < this._bindGroups.length; e++)
+      this._bindGroups[e].destroy();
+    this._bindGroups = null, this._bindGroupHash = null;
+    for (let e = 0; e < this._buffers.length; e++)
+      this._buffers[e].destroy();
+    this._buffers = null;
+    for (let e = 0; e < this._bufferResources.length; e++)
+      this._bufferResources[e].destroy();
+    this._bufferResources = null, this._batchBuffer.destroy(), this._bindGroupHash = null, this._renderer = null;
+  }
+}
+N.extension = {
+  type: [
+    l.WebGPUPipes
+  ],
+  name: "uniformBatch"
+};
+const we = {
+  "point-list": 0,
+  "line-list": 1,
+  "line-strip": 2,
+  "triangle-list": 3,
+  "triangle-strip": 4
+};
+function Re(a, e, t, r, s) {
+  return a << 24 | e << 16 | t << 10 | r << 5 | s;
+}
+function Ee(a, e, t, r) {
+  return t << 6 | a << 3 | r << 1 | e;
+}
+class K {
+  constructor(e) {
+    this._moduleCache = /* @__PURE__ */ Object.create(null), this._bufferLayoutsCache = /* @__PURE__ */ Object.create(null), this._pipeCache = /* @__PURE__ */ Object.create(null), this._pipeStateCaches = /* @__PURE__ */ Object.create(null), this._colorMask = 15, this._multisampleCount = 1, this._renderer = e;
+  }
+  contextChange(e) {
+    this._gpu = e, this.setStencilMode(P.DISABLED), this._updatePipeHash();
+  }
+  setMultisampleCount(e) {
+    this._multisampleCount !== e && (this._multisampleCount = e, this._updatePipeHash());
+  }
+  setRenderTarget(e) {
+    this._multisampleCount = e.msaaSamples, this._depthStencilAttachment = e.descriptor.depthStencilAttachment ? 1 : 0, this._updatePipeHash();
+  }
+  setColorMask(e) {
+    this._colorMask !== e && (this._colorMask = e, this._updatePipeHash());
+  }
+  setStencilMode(e) {
+    this._stencilMode !== e && (this._stencilMode = e, this._stencilState = xe[e], this._updatePipeHash());
+  }
+  setPipeline(e, t, r, s) {
+    const i = this.getPipeline(e, t, r);
+    s.setPipeline(i);
+  }
+  getPipeline(e, t, r, s) {
+    e._layoutKey || (Ge(e, t.attributeData), this._generateBufferKey(e)), s = s || e.topology;
+    const i = Re(
+      e._layoutKey,
+      t._layoutKey,
+      r.data,
+      r._blendModeId,
+      we[s]
+    );
+    return this._pipeCache[i] ? this._pipeCache[i] : (this._pipeCache[i] = this._createPipeline(e, t, r, s), this._pipeCache[i]);
+  }
+  _createPipeline(e, t, r, s) {
+    const i = this._gpu.device, n = this._createVertexBufferLayouts(e), o = this._renderer.state.getColorTargets(r);
+    o[0].writeMask = this._stencilMode === P.RENDERING_MASK_ADD ? 0 : this._colorMask;
+    const u = this._renderer.shader.getProgramData(t).pipeline, c = {
+      // TODO later check if its helpful to create..
+      // layout,
+      vertex: {
+        module: this._getModule(t.vertex.source),
+        entryPoint: t.vertex.entryPoint,
+        // geometry..
+        buffers: n
+      },
+      fragment: {
+        module: this._getModule(t.fragment.source),
+        entryPoint: t.fragment.entryPoint,
+        targets: o
+      },
+      primitive: {
+        topology: s,
+        cullMode: r.cullMode
+      },
+      layout: u,
+      multisample: {
+        count: this._multisampleCount
+      },
+      // depthStencil,
+      label: "PIXI Pipeline"
+    };
+    return this._depthStencilAttachment && (c.depthStencil = {
+      ...this._stencilState,
+      format: "depth24plus-stencil8",
+      depthWriteEnabled: r.depthTest,
+      depthCompare: r.depthTest ? "less" : "always"
+    }), i.createRenderPipeline(c);
+  }
+  _getModule(e) {
+    return this._moduleCache[e] || this._createModule(e);
+  }
+  _createModule(e) {
+    const t = this._gpu.device;
+    return this._moduleCache[e] = t.createShaderModule({
+      code: e
+    }), this._moduleCache[e];
+  }
+  _generateBufferKey(e) {
+    const t = [];
+    let r = 0;
+    const s = Object.keys(e.attributes).sort();
+    for (let n = 0; n < s.length; n++) {
+      const o = e.attributes[s[n]];
+      t[r++] = o.location, t[r++] = o.offset, t[r++] = o.format, t[r++] = o.stride;
+    }
+    const i = t.join("");
+    return e._layoutKey = he(i, "geometry"), e._layoutKey;
+  }
+  _createVertexBufferLayouts(e) {
+    if (this._bufferLayoutsCache[e._layoutKey])
+      return this._bufferLayoutsCache[e._layoutKey];
+    const t = [];
+    return e.buffers.forEach((r) => {
+      const s = {
+        arrayStride: 0,
+        stepMode: "vertex",
+        attributes: []
+      }, i = s.attributes;
+      for (const n in e.attributes) {
+        const o = e.attributes[n];
+        o.buffer === r && (s.arrayStride = o.stride, s.stepMode = o.instance ? "instance" : "vertex", i.push({
+          shaderLocation: o.location,
+          offset: o.offset,
+          format: o.format
+        }));
+      }
+      i.length && t.push(s);
+    }), this._bufferLayoutsCache[e._layoutKey] = t, t;
+  }
+  _updatePipeHash() {
+    const e = Ee(
+      this._stencilMode,
+      this._multisampleCount,
+      this._colorMask,
+      this._depthStencilAttachment
+    );
+    this._pipeStateCaches[e] || (this._pipeStateCaches[e] = /* @__PURE__ */ Object.create(null)), this._pipeCache = this._pipeStateCaches[e];
+  }
+  destroy() {
+    this._renderer = null, this._bufferLayoutsCache = null;
+  }
+}
+K.extension = {
+  type: [l.WebGPUSystem],
+  name: "pipeline"
+};
+class Ae {
+  constructor() {
+    this.contexts = [], this.msaaTextures = [], this.msaaSamples = 1;
+  }
+}
+class ke {
+  init(e, t) {
+    this._renderer = e, this._renderTargetSystem = t;
+  }
+  copyToTexture(e, t, r, s, i) {
+    const n = this._renderer, o = this._getGpuColorTexture(
+      e
+    ), u = n.texture.getGpuSource(
+      t.source
+    );
+    return n.encoder.commandEncoder.copyTextureToTexture(
+      {
+        texture: o,
+        origin: r
+      },
+      {
+        texture: u,
+        origin: i
+      },
+      s
+    ), t;
+  }
+  startRenderPass(e, t = !0, r, s) {
+    const n = this._renderTargetSystem.getGpuRenderTarget(e), o = this.getDescriptor(e, t, r);
+    n.descriptor = o, this._renderer.pipeline.setRenderTarget(n), this._renderer.encoder.beginRenderPass(n), this._renderer.encoder.setViewport(s);
+  }
+  finishRenderPass() {
+    this._renderer.encoder.endRenderPass();
+  }
+  /**
+   * returns the gpu texture for the first color texture in the render target
+   * mainly used by the filter manager to get copy the texture for blending
+   * @param renderTarget
+   * @returns a gpu texture
+   */
+  _getGpuColorTexture(e) {
+    const t = this._renderTargetSystem.getGpuRenderTarget(e);
+    return t.contexts[0] ? t.contexts[0].getCurrentTexture() : this._renderer.texture.getGpuSource(
+      e.colorTextures[0].source
+    );
+  }
+  getDescriptor(e, t, r) {
+    typeof t == "boolean" && (t = t ? b.ALL : b.NONE);
+    const s = this._renderTargetSystem, i = s.getGpuRenderTarget(e), n = e.colorTextures.map(
+      (c, f) => {
+        const h = i.contexts[f];
+        let p, d;
+        h ? p = h.getCurrentTexture().createView() : p = this._renderer.texture.getGpuSource(c).createView({
+          mipLevelCount: 1
+        }), i.msaaTextures[f] && (d = p, p = this._renderer.texture.getTextureView(
+          i.msaaTextures[f]
+        ));
+        const g = t & b.COLOR ? "clear" : "load";
+        return r ?? (r = s.defaultClearColor), {
+          view: p,
+          resolveTarget: d,
+          clearValue: r,
+          storeOp: "store",
+          loadOp: g
+        };
+      }
+    );
+    let o;
+    if ((e.stencil || e.depth) && !e.depthStencilTexture && (e.ensureDepthStencilTexture(), e.depthStencilTexture.source.sampleCount = i.msaa ? 4 : 1), e.depthStencilTexture) {
+      const c = t & b.STENCIL ? "clear" : "load", f = t & b.DEPTH ? "clear" : "load";
+      o = {
+        view: this._renderer.texture.getGpuSource(e.depthStencilTexture.source).createView(),
+        stencilStoreOp: "store",
+        stencilLoadOp: c,
+        depthClearValue: 1,
+        depthLoadOp: f,
+        depthStoreOp: "store"
+      };
+    }
+    return {
+      colorAttachments: n,
+      depthStencilAttachment: o
+    };
+  }
+  clear(e, t = !0, r, s) {
+    if (!t)
+      return;
+    const { gpu: i, encoder: n } = this._renderer, o = i.device;
+    if (n.commandEncoder === null) {
+      const c = o.createCommandEncoder(), f = this.getDescriptor(e, t, r), h = c.beginRenderPass(f);
+      h.setViewport(s.x, s.y, s.width, s.height, 0, 1), h.end();
+      const p = c.finish();
+      o.queue.submit([p]);
+    } else
+      this.startRenderPass(e, t, r, s);
+  }
+  initGpuRenderTarget(e) {
+    e.isRoot = !0;
+    const t = new Ae();
+    return e.colorTextures.forEach((r, s) => {
+      if (oe.test(r.resource)) {
+        const i = r.resource.getContext(
+          "webgpu"
+        ), n = r.transparent ? "premultiplied" : "opaque";
+        try {
+          i.configure({
+            device: this._renderer.gpu.device,
+            // eslint-disable-next-line max-len
+            usage: GPUTextureUsage.TEXTURE_BINDING | GPUTextureUsage.COPY_DST | GPUTextureUsage.RENDER_ATTACHMENT | GPUTextureUsage.COPY_SRC,
+            format: "bgra8unorm",
+            alphaMode: n
+          });
+        } catch (o) {
+          console.error(o);
+        }
+        t.contexts[s] = i;
+      }
+      if (t.msaa = r.source.antialias, r.source.antialias) {
+        const i = new ae({
+          width: 0,
+          height: 0,
+          sampleCount: 4
+        });
+        t.msaaTextures[s] = i;
+      }
+    }), t.msaa && (t.msaaSamples = 4, e.depthStencilTexture && (e.depthStencilTexture.source.sampleCount = 4)), t;
+  }
+  destroyGpuRenderTarget(e) {
+    e.contexts.forEach((t) => {
+      t.unconfigure();
+    }), e.msaaTextures.forEach((t) => {
+      t.destroy();
+    }), e.msaaTextures.length = 0, e.contexts.length = 0;
+  }
+  ensureDepthStencilTexture(e) {
+    const t = this._renderTargetSystem.getGpuRenderTarget(e);
+    e.depthStencilTexture && t.msaa && (e.depthStencilTexture.source.sampleCount = 4);
+  }
+  resizeGpuRenderTarget(e) {
+    const t = this._renderTargetSystem.getGpuRenderTarget(e);
+    t.width = e.width, t.height = e.height, t.msaa && e.colorTextures.forEach((r, s) => {
+      const i = t.msaaTextures[s];
+      i == null || i.resize(
+        r.source.width,
+        r.source.height,
+        r.source._resolution
+      );
+    });
+  }
+}
+class q extends Be {
+  constructor(e) {
+    super(e), this.adaptor = new ke(), this.adaptor.init(e, this);
+  }
+}
+q.extension = {
+  type: [l.WebGPUSystem],
+  name: "renderTarget"
+};
+class Y {
+  constructor() {
+    this._gpuProgramData = /* @__PURE__ */ Object.create(null);
+  }
+  contextChange(e) {
+    this._gpu = e;
+  }
+  getProgramData(e) {
+    return this._gpuProgramData[e._layoutKey] || this._createGPUProgramData(e);
+  }
+  _createGPUProgramData(e) {
+    const t = this._gpu.device, r = e.gpuLayout.map((i) => t.createBindGroupLayout({ entries: i })), s = { bindGroupLayouts: r };
+    return this._gpuProgramData[e._layoutKey] = {
+      bindGroups: r,
+      pipeline: t.createPipelineLayout(s)
+    }, this._gpuProgramData[e._layoutKey];
+  }
+  destroy() {
+    this._gpu = null, this._gpuProgramData = null;
+  }
+}
+Y.extension = {
+  type: [
+    l.WebGPUSystem
+  ],
+  name: "shader"
+};
+const m = {};
+m.normal = {
+  alpha: {
+    srcFactor: "one",
+    dstFactor: "one-minus-src-alpha",
+    operation: "add"
+  },
+  color: {
+    srcFactor: "one",
+    dstFactor: "one-minus-src-alpha",
+    operation: "add"
+  }
+};
+m.add = {
+  alpha: {
+    srcFactor: "src-alpha",
+    dstFactor: "one-minus-src-alpha",
+    operation: "add"
+  },
+  color: {
+    srcFactor: "one",
+    dstFactor: "one",
+    operation: "add"
+  }
+};
+m.multiply = {
+  alpha: {
+    srcFactor: "one",
+    dstFactor: "one-minus-src-alpha",
+    operation: "add"
+  },
+  color: {
+    srcFactor: "dst",
+    dstFactor: "one-minus-src-alpha",
+    operation: "add"
+  }
+};
+m.screen = {
+  alpha: {
+    srcFactor: "one",
+    dstFactor: "one-minus-src-alpha",
+    operation: "add"
+  },
+  color: {
+    srcFactor: "one",
+    dstFactor: "one-minus-src",
+    operation: "add"
+  }
+};
+m.overlay = {
+  alpha: {
+    srcFactor: "one",
+    dstFactor: "one-minus-src-alpha",
+    operation: "add"
+  },
+  color: {
+    srcFactor: "one",
+    dstFactor: "one-minus-src",
+    operation: "add"
+  }
+};
+m.none = {
+  alpha: {
+    srcFactor: "one",
+    dstFactor: "one-minus-src-alpha",
+    operation: "add"
+  },
+  color: {
+    srcFactor: "zero",
+    dstFactor: "zero",
+    operation: "add"
+  }
+};
+m["normal-npm"] = {
+  alpha: {
+    srcFactor: "one",
+    dstFactor: "one-minus-src-alpha",
+    operation: "add"
+  },
+  color: {
+    srcFactor: "src-alpha",
+    dstFactor: "one-minus-src-alpha",
+    operation: "add"
+  }
+};
+m["add-npm"] = {
+  alpha: {
+    srcFactor: "one",
+    dstFactor: "one",
+    operation: "add"
+  },
+  color: {
+    srcFactor: "src-alpha",
+    dstFactor: "one",
+    operation: "add"
+  }
+};
+m["screen-npm"] = {
+  alpha: {
+    srcFactor: "one",
+    dstFactor: "one-minus-src-alpha",
+    operation: "add"
+  },
+  color: {
+    srcFactor: "src-alpha",
+    dstFactor: "one-minus-src",
+    operation: "add"
+  }
+};
+m.erase = {
+  alpha: {
+    srcFactor: "zero",
+    dstFactor: "one-minus-src-alpha",
+    operation: "add"
+  },
+  color: {
+    srcFactor: "zero",
+    dstFactor: "one-minus-src",
+    operation: "add"
+  }
+};
+class $ {
+  constructor() {
+    this.defaultState = new D(), this.defaultState.blend = !0;
+  }
+  contextChange(e) {
+    this.gpu = e;
+  }
+  /**
+   * Gets the blend mode data for the current state
+   * @param state - The state to get the blend mode from
+   */
+  getColorTargets(e) {
+    return [
+      {
+        format: "bgra8unorm",
+        writeMask: 0,
+        blend: m[e.blendMode] || m.normal
+      }
+    ];
+  }
+  destroy() {
+    this.gpu = null;
+  }
+}
+$.extension = {
+  type: [
+    l.WebGPUSystem
+  ],
+  name: "state"
+};
+const De = {
+  type: "image",
+  upload(a, e, t) {
+    const r = a.resource, s = (a.pixelWidth | 0) * (a.pixelHeight | 0), i = r.byteLength / s;
+    t.device.queue.writeTexture(
+      { texture: e },
+      r,
+      {
+        offset: 0,
+        rowsPerImage: a.pixelHeight,
+        bytesPerRow: a.pixelHeight * i
+      },
+      {
+        width: a.pixelWidth,
+        height: a.pixelHeight,
+        depthOrArrayLayers: 1
+      }
+    );
+  }
+}, X = {
+  "bc1-rgba-unorm": { blockBytes: 8, blockWidth: 4, blockHeight: 4 },
+  "bc2-rgba-unorm": { blockBytes: 16, blockWidth: 4, blockHeight: 4 },
+  "bc3-rgba-unorm": { blockBytes: 16, blockWidth: 4, blockHeight: 4 },
+  "bc7-rgba-unorm": { blockBytes: 16, blockWidth: 4, blockHeight: 4 },
+  "etc1-rgb-unorm": { blockBytes: 8, blockWidth: 4, blockHeight: 4 },
+  "etc2-rgba8unorm": { blockBytes: 16, blockWidth: 4, blockHeight: 4 },
+  "astc-4x4-unorm": { blockBytes: 16, blockWidth: 4, blockHeight: 4 }
+}, ze = { blockBytes: 4, blockWidth: 1, blockHeight: 1 }, Le = {
+  type: "compressed",
+  upload(a, e, t) {
+    let r = a.pixelWidth, s = a.pixelHeight;
+    const i = X[a.format] || ze;
+    for (let n = 0; n < a.resource.length; n++) {
+      const o = a.resource[n], u = Math.ceil(r / i.blockWidth) * i.blockBytes;
+      t.device.queue.writeTexture(
+        {
+          texture: e,
+          mipLevel: n
+        },
+        o,
+        {
+          offset: 0,
+          bytesPerRow: u
+        },
+        {
+          width: Math.ceil(r / i.blockWidth) * i.blockWidth,
+          height: Math.ceil(s / i.blockHeight) * i.blockHeight,
+          depthOrArrayLayers: 1
+        }
+      ), r = Math.max(r >> 1, 1), s = Math.max(s >> 1, 1);
+    }
+  }
+}, Z = {
+  type: "image",
+  upload(a, e, t) {
+    const r = a.resource;
+    if (!r)
+      return;
+    const s = Math.min(e.width, a.resourceWidth || a.pixelWidth), i = Math.min(e.height, a.resourceHeight || a.pixelHeight), n = a.alphaMode === "premultiply-alpha-on-upload";
+    t.device.queue.copyExternalImageToTexture(
+      { source: r },
+      { texture: e, premultipliedAlpha: n },
+      {
+        width: s,
+        height: i
+      }
+    );
+  }
+}, Oe = {
+  type: "video",
+  upload(a, e, t) {
+    Z.upload(a, e, t);
+  }
+};
+class He {
+  constructor(e) {
+    this.device = e, this.sampler = e.createSampler({ minFilter: "linear" }), this.pipelines = {};
+  }
+  _getMipmapPipeline(e) {
+    let t = this.pipelines[e];
+    return t || (this.mipmapShaderModule || (this.mipmapShaderModule = this.device.createShaderModule({
+      code: (
+        /* wgsl */
+        `
+                        var<private> pos : array<vec2<f32>, 3> = array<vec2<f32>, 3>(
+                        vec2<f32>(-1.0, -1.0), vec2<f32>(-1.0, 3.0), vec2<f32>(3.0, -1.0));
+
+                        struct VertexOutput {
+                        @builtin(position) position : vec4<f32>,
+                        @location(0) texCoord : vec2<f32>,
+                        };
+
+                        @vertex
+                        fn vertexMain(@builtin(vertex_index) vertexIndex : u32) -> VertexOutput {
+                        var output : VertexOutput;
+                        output.texCoord = pos[vertexIndex] * vec2<f32>(0.5, -0.5) + vec2<f32>(0.5);
+                        output.position = vec4<f32>(pos[vertexIndex], 0.0, 1.0);
+                        return output;
+                        }
+
+                        @group(0) @binding(0) var imgSampler : sampler;
+                        @group(0) @binding(1) var img : texture_2d<f32>;
+
+                        @fragment
+                        fn fragmentMain(@location(0) texCoord : vec2<f32>) -> @location(0) vec4<f32> {
+                        return textureSample(img, imgSampler, texCoord);
+                        }
+                    `
+      )
+    })), t = this.device.createRenderPipeline({
+      layout: "auto",
+      vertex: {
+        module: this.mipmapShaderModule,
+        entryPoint: "vertexMain"
+      },
+      fragment: {
+        module: this.mipmapShaderModule,
+        entryPoint: "fragmentMain",
+        targets: [{ format: e }]
+      }
+    }), this.pipelines[e] = t), t;
+  }
+  /**
+   * Generates mipmaps for the given GPUTexture from the data in level 0.
+   * @param {module:External.GPUTexture} texture - Texture to generate mipmaps for.
+   * @returns {module:External.GPUTexture} - The originally passed texture
+   */
+  generateMipmap(e) {
+    const t = this._getMipmapPipeline(e.format);
+    if (e.dimension === "3d" || e.dimension === "1d")
+      throw new Error("Generating mipmaps for non-2d textures is currently unsupported!");
+    let r = e;
+    const s = e.depthOrArrayLayers || 1, i = e.usage & GPUTextureUsage.RENDER_ATTACHMENT;
+    if (!i) {
+      const u = {
+        size: {
+          width: Math.ceil(e.width / 2),
+          height: Math.ceil(e.height / 2),
+          depthOrArrayLayers: s
+        },
+        format: e.format,
+        usage: GPUTextureUsage.TEXTURE_BINDING | GPUTextureUsage.COPY_SRC | GPUTextureUsage.RENDER_ATTACHMENT,
+        mipLevelCount: e.mipLevelCount - 1
+      };
+      r = this.device.createTexture(u);
+    }
+    const n = this.device.createCommandEncoder({}), o = t.getBindGroupLayout(0);
+    for (let u = 0; u < s; ++u) {
+      let c = e.createView({
+        baseMipLevel: 0,
+        mipLevelCount: 1,
+        dimension: "2d",
+        baseArrayLayer: u,
+        arrayLayerCount: 1
+      }), f = i ? 1 : 0;
+      for (let h = 1; h < e.mipLevelCount; ++h) {
+        const p = r.createView({
+          baseMipLevel: f++,
+          mipLevelCount: 1,
+          dimension: "2d",
+          baseArrayLayer: u,
+          arrayLayerCount: 1
+        }), d = n.beginRenderPass({
+          colorAttachments: [{
+            view: p,
+            storeOp: "store",
+            loadOp: "clear",
+            clearValue: { r: 0, g: 0, b: 0, a: 0 }
+          }]
+        }), g = this.device.createBindGroup({
+          layout: o,
+          entries: [{
+            binding: 0,
+            resource: this.sampler
+          }, {
+            binding: 1,
+            resource: c
+          }]
+        });
+        d.setPipeline(t), d.setBindGroup(0, g), d.draw(3, 1, 0, 0), d.end(), c = p;
+      }
+    }
+    if (!i) {
+      const u = {
+        width: Math.ceil(e.width / 2),
+        height: Math.ceil(e.height / 2),
+        depthOrArrayLayers: s
+      };
+      for (let c = 1; c < e.mipLevelCount; ++c)
+        n.copyTextureToTexture({
+          texture: r,
+          mipLevel: c - 1
+        }, {
+          texture: e,
+          mipLevel: c
+        }, u), u.width = Math.ceil(u.width / 2), u.height = Math.ceil(u.height / 2);
+    }
+    return this.device.queue.submit([n.finish()]), i || r.destroy(), e;
+  }
+}
+class J {
+  constructor(e) {
+    this.managedTextures = [], this._gpuSources = /* @__PURE__ */ Object.create(null), this._gpuSamplers = /* @__PURE__ */ Object.create(null), this._bindGroupHash = /* @__PURE__ */ Object.create(null), this._textureViewHash = /* @__PURE__ */ Object.create(null), this._uploads = {
+      image: Z,
+      buffer: De,
+      video: Oe,
+      compressed: Le
+    }, this._renderer = e;
+  }
+  contextChange(e) {
+    this._gpu = e;
+  }
+  initSource(e) {
+    if (e.autoGenerateMipmaps) {
+      const u = Math.max(e.pixelWidth, e.pixelHeight);
+      e.mipLevelCount = Math.floor(Math.log2(u)) + 1;
+    }
+    let t = GPUTextureUsage.TEXTURE_BINDING | GPUTextureUsage.COPY_DST;
+    e.uploadMethodId !== "compressed" && (t |= GPUTextureUsage.RENDER_ATTACHMENT, t |= GPUTextureUsage.COPY_SRC);
+    const r = X[e.format] || { blockBytes: 4, blockWidth: 1, blockHeight: 1 }, s = Math.ceil(e.pixelWidth / r.blockWidth) * r.blockWidth, i = Math.ceil(e.pixelHeight / r.blockHeight) * r.blockHeight, n = {
+      label: e.label,
+      size: { width: s, height: i },
+      format: e.format,
+      sampleCount: e.sampleCount,
+      mipLevelCount: e.mipLevelCount,
+      dimension: e.dimension,
+      usage: t
+    }, o = this._gpu.device.createTexture(n);
+    return this._gpuSources[e.uid] = o, this.managedTextures.includes(e) || (e.on("update", this.onSourceUpdate, this), e.on("resize", this.onSourceResize, this), e.on("destroy", this.onSourceDestroy, this), e.on("unload", this.onSourceUnload, this), e.on("updateMipmaps", this.onUpdateMipmaps, this), this.managedTextures.push(e)), this.onSourceUpdate(e), o;
+  }
+  onSourceUpdate(e) {
+    const t = this.getGpuSource(e);
+    t && (this._uploads[e.uploadMethodId] && this._uploads[e.uploadMethodId].upload(e, t, this._gpu), e.autoGenerateMipmaps && e.mipLevelCount > 1 && this.onUpdateMipmaps(e));
+  }
+  onSourceUnload(e) {
+    const t = this._gpuSources[e.uid];
+    t && (this._gpuSources[e.uid] = null, t.destroy());
+  }
+  onUpdateMipmaps(e) {
+    this._mipmapGenerator || (this._mipmapGenerator = new He(this._gpu.device));
+    const t = this.getGpuSource(e);
+    this._mipmapGenerator.generateMipmap(t);
+  }
+  onSourceDestroy(e) {
+    e.off("update", this.onSourceUpdate, this), e.off("unload", this.onSourceUnload, this), e.off("destroy", this.onSourceDestroy, this), e.off("resize", this.onSourceResize, this), e.off("updateMipmaps", this.onUpdateMipmaps, this), this.managedTextures.splice(this.managedTextures.indexOf(e), 1), this.onSourceUnload(e);
+  }
+  onSourceResize(e) {
+    const t = this._gpuSources[e.uid];
+    t ? (t.width !== e.pixelWidth || t.height !== e.pixelHeight) && (this._textureViewHash[e.uid] = null, this._bindGroupHash[e.uid] = null, this.onSourceUnload(e), this.initSource(e)) : this.initSource(e);
+  }
+  _initSampler(e) {
+    return this._gpuSamplers[e._resourceId] = this._gpu.device.createSampler(e), this._gpuSamplers[e._resourceId];
+  }
+  getGpuSampler(e) {
+    return this._gpuSamplers[e._resourceId] || this._initSampler(e);
+  }
+  getGpuSource(e) {
+    return this._gpuSources[e.uid] || this.initSource(e);
+  }
+  getTextureBindGroup(e) {
+    return this._bindGroupHash[e.uid] ?? this._createTextureBindGroup(e);
+  }
+  _createTextureBindGroup(e) {
+    const t = e.source, r = t.uid;
+    return this._bindGroupHash[r] = new A({
+      0: t,
+      1: t.style
+    }), this._bindGroupHash[r];
+  }
+  getTextureView(e) {
+    const t = e.source;
+    return this._textureViewHash[t.uid] ?? this._createTextureView(t);
+  }
+  _createTextureView(e) {
+    return this._textureViewHash[e.uid] = this.getGpuSource(e).createView(), this._textureViewHash[e.uid];
+  }
+  generateCanvas(e) {
+    const t = this._renderer, r = t.gpu.device.createCommandEncoder(), s = ue.get().createCanvas();
+    s.width = e.source.pixelWidth, s.height = e.source.pixelHeight;
+    const i = s.getContext("webgpu");
+    return i.configure({
+      device: t.gpu.device,
+      // eslint-disable-next-line max-len
+      usage: GPUTextureUsage.COPY_DST | GPUTextureUsage.COPY_SRC,
+      format: navigator.gpu.getPreferredCanvasFormat(),
+      alphaMode: "premultiplied"
+    }), r.copyTextureToTexture({
+      texture: t.texture.getGpuSource(e.source),
+      origin: {
+        x: 0,
+        y: 0
+      }
+    }, {
+      texture: i.getCurrentTexture()
+    }, {
+      width: s.width,
+      height: s.height
+    }), t.gpu.device.queue.submit([r.finish()]), s;
+  }
+  getPixels(e) {
+    const t = this.generateCanvas(e), r = w.getOptimalCanvasAndContext(t.width, t.height), s = r.context;
+    s.drawImage(t, 0, 0);
+    const { width: i, height: n } = t, o = s.getImageData(0, 0, i, n), u = new Uint8ClampedArray(o.data.buffer);
+    return w.returnCanvasAndContext(r), { pixels: u, width: i, height: n };
+  }
+  destroy() {
+    this.managedTextures.slice().forEach((e) => this.onSourceDestroy(e)), this.managedTextures = null;
+    for (const e of Object.keys(this._bindGroupHash)) {
+      const t = Number(e), r = this._bindGroupHash[t];
+      r == null || r.destroy(), this._bindGroupHash[t] = null;
+    }
+    this._gpu = null, this._mipmapGenerator = null, this._gpuSources = null, this._bindGroupHash = null, this._textureViewHash = null, this._gpuSamplers = null;
+  }
+}
+J.extension = {
+  type: [
+    l.WebGPUSystem
+  ],
+  name: "texture"
+};
+class Q {
+  init() {
+    const e = new pe({
+      uTransformMatrix: { value: new k(), type: "mat3x3<f32>" },
+      uColor: { value: new Float32Array([1, 1, 1, 1]), type: "vec4<f32>" },
+      uRound: { value: 0, type: "f32" }
+    }), t = T({
+      name: "graphics",
+      bits: [
+        z,
+        L(R()),
+        le,
+        C
+      ]
+    });
+    this.shader = new v({
+      gpuProgram: t,
+      resources: {
+        // added on the fly!
+        localUniforms: e
+      }
+    });
+  }
+  execute(e, t) {
+    const r = t.context, s = r.customShader || this.shader, i = e.renderer, n = i.graphicsContext, {
+      geometry: o,
+      instructions: u
+    } = n.getContextRenderData(r), c = i.encoder;
+    c.setPipelineFromGeometryProgramAndState(
+      o,
+      s.gpuProgram,
+      e.state
+    ), c.setGeometry(o);
+    const f = i.globalUniforms.bindGroup;
+    c.setBindGroup(0, f, s.gpuProgram);
+    const h = i.renderPipes.uniformBatch.getUniformBindGroup(s.resources.localUniforms, !0);
+    c.setBindGroup(2, h, s.gpuProgram);
+    const p = u.instructions;
+    for (let d = 0; d < u.instructionSize; d++) {
+      const g = p[d];
+      if (s.groups[1] = g.bindGroup, !g.gpuBindGroup) {
+        const B = g.textures;
+        g.bindGroup = E(B.textures, B.count), g.gpuBindGroup = i.bindGroup.getBindGroup(
+          g.bindGroup,
+          s.gpuProgram,
+          1
+        );
+      }
+      c.setBindGroup(1, g.bindGroup, s.gpuProgram), c.renderPassEncoder.drawIndexed(g.size, 1, g.start);
+    }
+  }
+  destroy() {
+    this.shader.destroy(!0), this.shader = null;
+  }
+}
+Q.extension = {
+  type: [
+    l.WebGPUPipesAdaptor
+  ],
+  name: "graphics"
+};
+class ee {
+  init() {
+    const e = T({
+      name: "mesh",
+      bits: [
+        fe,
+        Se,
+        C
+      ]
+    });
+    this._shader = new v({
+      gpuProgram: e,
+      resources: {
+        uTexture: M.EMPTY._source,
+        uSampler: M.EMPTY._source.style,
+        textureUniforms: {
+          uTextureMatrix: { type: "mat3x3<f32>", value: new k() }
+        }
+      }
+    });
+  }
+  execute(e, t) {
+    const r = e.renderer;
+    let s = t._shader;
+    if (!s)
+      s = this._shader, s.resources.uTexture = t.texture.source, s.resources.uSampler = t.texture.source.style, s.resources.textureUniforms.uniforms.uTextureMatrix = t.texture.textureMatrix.mapCoord;
+    else if (!s.gpuProgram) {
+      ce("Mesh shader has no gpuProgram", t.shader);
+      return;
+    }
+    const i = s.gpuProgram;
+    if (i.autoAssignGlobalUniforms && (s.groups[0] = r.globalUniforms.bindGroup), i.autoAssignLocalUniforms) {
+      const n = e.localUniforms;
+      s.groups[1] = r.renderPipes.uniformBatch.getUniformBindGroup(n, !0);
+    }
+    r.encoder.draw({
+      geometry: t._geometry,
+      shader: s,
+      state: t.state
+    });
+  }
+  destroy() {
+    this._shader.destroy(!0), this._shader = null;
+  }
+}
+ee.extension = {
+  type: [
+    l.WebGPUPipesAdaptor
+  ],
+  name: "mesh"
+};
+const Fe = [
+  ...Pe,
+  j,
+  W,
+  U,
+  F,
+  J,
+  q,
+  Y,
+  $,
+  K,
+  I,
+  V,
+  H
+], Ie = [...Te, N], We = [O, ee, Q], te = [], re = [], se = [];
+G.handleByNamedList(l.WebGPUSystem, te);
+G.handleByNamedList(l.WebGPUPipes, re);
+G.handleByNamedList(l.WebGPUPipesAdaptor, se);
+G.add(...Fe, ...Ie, ...We);
+class Ye extends de {
+  constructor() {
+    const e = {
+      name: "webgpu",
+      type: ge.WEBGPU,
+      systems: te,
+      renderPipes: re,
+      renderPipeAdaptors: se
+    };
+    super(e);
+  }
+}
+export {
+  Ye as WebGPURenderer
+};