docs(webgpu-rt): add RTVolume example (procedural spheres + any-hit cut-out)

A 3x3x3 grid of AABB-geometry spheres rendered through an analytic
ray-sphere intersection shader, with an any-hit spherical-checkerboard
cut-out so the background shows through. Exercises both features end to
end on the WebGPU wavefront tracer.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

README.md

@@ -22,7 +22,11 @@ The two backends share the same C++ surface for the high-level pieces
 (`*Vulkan` vs `*WebGPU`) live behind `#ifdef CRAFTER_GRAPHICS_WINDOW_DOM`.
 Vulkan ray tracing is hardware (`VK_KHR_ray_tracing_pipeline`); WebGPU
 ray tracing is a library-built software path (BVH + traceRay in a
-compute pipeline composed from user-supplied WGSL stages).
+compute pipeline composed from user-supplied WGSL stages). The WebGPU
+path supports triangle and AABB (procedural, `VK_GEOMETRY_TYPE_AABBS_KHR`)
+geometry, closest-hit / miss / any-hit / intersection shaders — see
+[examples/RTVolume](examples/RTVolume/README.md) for procedural spheres
+shaded through an intersection shader with an any-hit cut-out.
 
 > **Native RT status:** reading an acceleration structure through
 > `VK_EXT_descriptor_heap` currently aborts with `VK_ERROR_DEVICE_LOST` on

examples/RTVolume/README.md

@@ -0,0 +1,26 @@
+# RTVolume
+
+WebGPU software ray tracing of **procedural (AABB) geometry** with an
+**any-hit** cut-out — the two features added for issue #13.
+
+A 3×3×3 grid of unit boxes is registered as an AABB BLAS
+(`Mesh::BuildProcedural`, the WebGPU analog of `VK_GEOMETRY_TYPE_AABBS_KHR`).
+The hit group is a `RTShaderGroupType::ProceduralHitGroup` carrying:
+
+- `intersection.wgsl` — analytic ray–sphere test that turns each box into a
+  radius-1 sphere (runs in TRACE, once per box the ray enters);
+- `anyhit.wgsl` — returns `RT_ANYHIT_IGNORE` for half the cells of a
+  spherical checkerboard, so the ray passes through and the background /
+  spheres behind show through (the visible proof any-hit runs);
+- `closesthit.wgsl` — normal-based Lambert shading, tinted per instance.
+
+The geometry is registered **non-opaque** and the instances clear their
+force-opaque flag, which is what lets the any-hit shader run. Flip the
+instance flag to `kRTGeometryInstanceForceOpaque` (or build the mesh with
+`opaque = true`) to skip any-hit and see solid spheres.
+
+WebGPU/DOM only:
+
+```
+crafter-build --target=wasm32-wasip1 -r
+```

examples/RTVolume/anyhit.wgsl

@@ -0,0 +1,24 @@
+// RTVolume any-hit shader (runs in TRACE on every candidate sphere hit,
+// because the geometry is registered non-opaque). Punches a spherical
+// checkerboard of holes: for half the cells it returns RT_ANYHIT_IGNORE,
+// so the ray passes straight through and the background / spheres behind
+// show through. Returning RT_ANYHIT_ACCEPT keeps the hit. This is the
+// visible proof the any-hit path runs — with it the spheres are perforated,
+// without it they would be solid.
+fn anyhit_main(ray: RayDesc, hit: HitInfo, payload: ptr<function, Payload>) -> u32 {
+    // Object-space hit point on the unit sphere → its normal/direction.
+    let posObj = hit.objectRayOrigin + hit.objectRayDirection * hit.t;
+    let n = normalize(posObj);
+
+    let PI = 3.14159265;
+    let longitude = atan2(n.z, n.x);                 // [-PI, PI]
+    let latitude  = asin(clamp(n.y, -1.0, 1.0));     // [-PI/2, PI/2]
+
+    let cu = i32(floor((longitude + PI) / PI * 6.0));
+    let cv = i32(floor((latitude + PI * 0.5) / PI * 6.0));
+
+    if (((cu + cv) & 1) == 0) {
+        return RT_ANYHIT_IGNORE;   // cut-out cell — see through
+    }
+    return RT_ANYHIT_ACCEPT;
+}

examples/RTVolume/closesthit.wgsl

@@ -0,0 +1,37 @@
+// RTVolume closest-hit (runs in SHADE). Shades the procedural sphere by
+// its surface normal with a fixed sun + ambient, tinted per instance.
+//
+// Payload declared here so the assembler sees it before wfPayload / SHADE.
+struct Payload {
+    color: vec3<f32>,
+};
+
+const SUN_DIR_TO_LIGHT: vec3<f32> = vec3<f32>(0.40, 0.85, 0.35);
+const SUN_COLOR:        vec3<f32> = vec3<f32>(1.20, 1.10, 0.95);
+const AMBIENT_COLOR:    vec3<f32> = vec3<f32>(0.16, 0.18, 0.24);
+
+fn instanceAlbedo(i: u32) -> vec3<f32> {
+    let h = i * 2654435761u;
+    return vec3<f32>(
+        0.35 + 0.6 * f32((h >>  0u) & 255u) / 255.0,
+        0.35 + 0.6 * f32((h >>  8u) & 255u) / 255.0,
+        0.35 + 0.6 * f32((h >> 16u) & 255u) / 255.0);
+}
+
+fn closesthit_main(ray: RayDesc, hit: HitInfo, payload: ptr<function, Payload>) {
+    // Object-space hit point on the unit sphere is its object-space normal.
+    let posObj = hit.objectRayOrigin + hit.objectRayDirection * hit.t;
+    let nObj   = normalize(posObj);
+    let nWorld = normalize(vec3<f32>(
+        dot(hit.objectToWorldR0.xyz, nObj),
+        dot(hit.objectToWorldR1.xyz, nObj),
+        dot(hit.objectToWorldR2.xyz, nObj)));
+
+    let albedo  = instanceAlbedo(hit.customIndex);
+    let viewDir = -ray.direction;
+    let nFacing = select(-nWorld, nWorld, dot(nWorld, viewDir) > 0.0);
+    let sunDir  = normalize(SUN_DIR_TO_LIGHT);
+    let nDotL   = max(0.0, dot(nFacing, sunDir));
+
+    rtAccumulate(albedo * (AMBIENT_COLOR + SUN_COLOR * nDotL));
+}

examples/RTVolume/intersection.wgsl

@@ -0,0 +1,33 @@
+// RTVolume intersection shader (runs in TRACE, per AABB the ray enters).
+// Analytic ray-sphere test: the unit box [-1,1]^3 is treated as the
+// bounding volume of a sphere of radius 1 centred at the box centre. The
+// ray is in object space and is NOT normalised (it is worldToObject *
+// worldRay), so the returned t is directly comparable to the world-space
+// ray parameter the tracer commits — solve the quadratic with the general
+// a = dot(d,d) form rather than assuming |d| == 1.
+fn intersection_main(ray: RayDesc, aabbMin: vec3<f32>, aabbMax: vec3<f32>,
+                     primitiveId: u32) -> IntersectionResult {
+    var r: IntersectionResult;
+    r.hit = false;
+
+    let center = (aabbMin + aabbMax) * 0.5;
+    let radius = (aabbMax.x - aabbMin.x) * 0.5;
+
+    let oc = ray.origin - center;
+    let a  = dot(ray.direction, ray.direction);
+    let b  = 2.0 * dot(oc, ray.direction);
+    let c  = dot(oc, oc) - radius * radius;
+    let disc = b * b - 4.0 * a * c;
+    if (disc < 0.0) { return r; }
+
+    let sq = sqrt(disc);
+    var t = (-b - sq) / (2.0 * a);   // near root
+    if (t < ray.tMin) { t = (-b + sq) / (2.0 * a); }   // fall back to far root
+    if (t < ray.tMin || t > ray.tMax) { return r; }
+
+    r.hit     = true;
+    r.t       = t;
+    r.attribs = vec2<f32>(0.0);
+    r.hitKind = 0u;
+    return r;
+}

examples/RTVolume/main.cpp

@@ -0,0 +1,199 @@
+// RTVolume — procedural (AABB) ray tracing on the WebGPU wavefront tracer.
+// Demonstrates the two features this example was written to exercise:
+//
+//   * VK_GEOMETRY_TYPE_AABBS_KHR equivalent — a BLAS built from AABBs
+//     (Mesh::BuildProcedural) whose surface is supplied by an intersection
+//     shader (here an analytic ray–sphere test). The boxes are unit cubes
+//     [-1,1]^3; the intersection shader turns each into a sphere.
+//
+//   * any-hit — the spheres are registered non-opaque, and an any-hit
+//     shader punches a spherical checkerboard of holes by returning
+//     RT_ANYHIT_IGNORE for half the cells. Without any-hit the spheres are
+//     solid; with it you can see the background (and other spheres)
+//     through the cut-out cells.
+//
+// A 3×3×3 grid of these procedural spheres is shaded by surface normal +
+// a fixed sun. WebGPU/DOM only — this is the software RT path.
+
+#ifndef CRAFTER_GRAPHICS_WINDOW_DOM
+int main() { return 0; }   // native path is hardware RT; out of scope here
+#else
+
+import Crafter.Graphics;
+import Crafter.Math;
+import Crafter.Event;
+import std;
+
+using namespace Crafter;
+namespace fs = std::filesystem;
+
+namespace {
+    constexpr int   kGrid    = 3;
+    constexpr float kSpacing = 3.0f;
+
+    struct CameraGPU {
+        float origin[3];   float pad0;
+        float right[3];    float tanHalf;
+        float up[3];       float aspect;
+        float forward[3];  float pad1;
+    };
+    static_assert(sizeof(CameraGPU) == 64);
+}
+
+int main() {
+    const int instanceCount = kGrid * kGrid * kGrid;
+    std::println("[RTVolume] grid {}^3 = {} procedural spheres", kGrid, instanceCount);
+
+    Device::Initialize();
+    static Window window(1280, 720, "RTVolume");
+    auto cmd = window.StartInit();
+
+    DescriptorHeapWebGPU heap;
+    heap.Initialize(/*images*/ 1, /*buffers*/ 2, /*samplers*/ 1);
+
+    // SBT order fixes the shader indices used by the groups below.
+    std::array<WebGPUShader, 6> shaders {{
+        WebGPUShader(fs::path("raygen.wgsl"),       "raygen_main",       WebGPURTStage::Raygen),
+        WebGPUShader(fs::path("miss.wgsl"),         "miss_main",         WebGPURTStage::Miss),
+        WebGPUShader(fs::path("closesthit.wgsl"),   "closesthit_main",   WebGPURTStage::ClosestHit),
+        WebGPUShader(fs::path("anyhit.wgsl"),       "anyhit_main",       WebGPURTStage::AnyHit),
+        WebGPUShader(fs::path("intersection.wgsl"), "intersection_main", WebGPURTStage::Intersection),
+        WebGPUShader(fs::path("resolve.wgsl"),      "resolve_main",      WebGPURTStage::Resolve),
+    }};
+    ShaderBindingTableWebGPU sbt;
+    sbt.Init(shaders);
+
+    std::array<RTShaderGroup, 1> raygenGroups {{ { .type = RTShaderGroupType::General, .generalShader = 0 } }};
+    std::array<RTShaderGroup, 1> missGroups   {{ { .type = RTShaderGroupType::General, .generalShader = 1 } }};
+    // One procedural hit group: closest-hit + any-hit + intersection.
+    std::array<RTShaderGroup, 1> hitGroups    {{ {
+        .type               = RTShaderGroupType::ProceduralHitGroup,
+        .closestHitShader   = 2,
+        .anyHitShader       = 3,
+        .intersectionShader = 4,
+    } }};
+
+    std::array<UICustomBinding, 1> bindings {{
+        { .group = 3, .binding = 0, .kind = UICustomBindingKind::Buffer, ._pad = 0, .pushOffset = 0 },
+    }};
+
+    PipelineRTWebGPU pipeline;
+    pipeline.Init(cmd, raygenGroups, missGroups, hitGroups, sbt, bindings);
+
+    // ── One procedural unit-box BLAS. The intersection shader treats the
+    //    box as the bounding volume of a radius-1 sphere centred at the
+    //    object origin. opaque=false so the any-hit cut-out runs. ─────────
+    static std::array<RTAabb, 1> boxes {{
+        { .min = {-1.0f, -1.0f, -1.0f}, .max = {1.0f, 1.0f, 1.0f} },
+    }};
+    static Mesh sphere;
+    sphere.BuildProcedural(boxes, /*opaque*/ false, cmd);
+
+    // ── Camera buffer + handle array. ─────────────────────────────────
+    WebGPUBuffer<CameraGPU, true> cameraBuf;
+    cameraBuf.Create(1);
+    static std::array<std::uint32_t, 1> userHandles { cameraBuf.handle };
+
+    // ── Instance grid. ────────────────────────────────────────────────
+    static std::vector<RenderingElement3D> renderers;
+    renderers.reserve(static_cast<std::size_t>(instanceCount));
+    const float origin0 = -0.5f * static_cast<float>(kGrid - 1) * kSpacing;
+    for (int x = 0; x < kGrid; ++x)
+    for (int y = 0; y < kGrid; ++y)
+    for (int z = 0; z < kGrid; ++z) {
+        renderers.emplace_back();
+        RenderingElement3D& r = renderers.back();
+        auto& tx = r.instance.transform.matrix;
+        tx[0][0] = 1; tx[0][1] = 0; tx[0][2] = 0; tx[0][3] = origin0 + float(x) * kSpacing;
+        tx[1][0] = 0; tx[1][1] = 1; tx[1][2] = 0; tx[1][3] = origin0 + float(y) * kSpacing;
+        tx[2][0] = 0; tx[2][1] = 0; tx[2][2] = 1; tx[2][3] = origin0 + float(z) * kSpacing;
+        r.instance.instanceCustomIndex                    = static_cast<std::uint32_t>(renderers.size() - 1);
+        r.instance.mask                                   = 0xFF;
+        r.instance.instanceShaderBindingTableRecordOffset = 0;
+        // flags = 0: do NOT force opaque, so the any-hit shader runs.
+        r.instance.flags                                  = 0;
+        r.instance.accelerationStructureReference         = sphere.blasAddr;
+        RenderingElement3D::Add(&r);
+    }
+    RenderingElement3D::BuildTLAS(cmd, 0);
+
+    window.descriptorHeap = &heap;
+    window.FinishInit();
+
+    RTPass rtPass(&pipeline);
+    rtPass.handlesPtr   = userHandles.data();
+    rtPass.handlesCount = static_cast<std::uint32_t>(userHandles.size());
+    rtPass.maxDepth     = 1;   // primary only
+    window.passes.push_back(&rtPass);
+
+    // ── Free camera framing the grid. ─────────────────────────────────
+    const float ext = float(kGrid - 1) * kSpacing;
+    struct CamState {
+        Vector<float, 3, 4> position;
+        float yaw;
+        float pitch;
+    } cam {
+        Vector<float, 3, 4>{ ext * 1.1f, ext * 0.8f, ext * 1.6f },
+        0.0f, 0.0f,
+    };
+    {
+        Vector<float, 3, 4> d { -cam.position.x, -cam.position.y, -cam.position.z };
+        const float len = std::sqrt(d.x*d.x + d.y*d.y + d.z*d.z);
+        cam.yaw   = std::atan2(d.z, d.x);
+        cam.pitch = std::asin(d.y / len);
+    }
+
+    Input::Map inputMap;
+    Input::Action& moveAct = inputMap.AddAction("Move", Input::ActionType::Vector2);
+    Input::Action& lookAct = inputMap.AddAction("Look", Input::ActionType::Vector2);
+    moveAct.bindings = { Input::WASDBind{
+        Key(CrafterKeys::W), Key(CrafterKeys::S), Key(CrafterKeys::A), Key(CrafterKeys::D) } };
+    lookAct.bindings = { Input::MouseDeltaBind{ 1.0f } };
+    inputMap.Attach(window);
+
+    const float kMoveSpeed = ext * 0.8f + 1.0f;
+    const float kLookSens  = 0.05f;
+    const float kDt        = 1.0f / 60.0f;
+
+    EventListener<void> camTick(&window.onBeforeUpdate, [&]() {
+        inputMap.Tick();
+        cam.yaw   += lookAct.vector2.x * kLookSens;
+        cam.pitch -= lookAct.vector2.y * kLookSens;
+        cam.pitch  = std::clamp(cam.pitch, -1.55f, 1.55f);
+
+        const float cp = std::cos(cam.pitch), sp = std::sin(cam.pitch);
+        const float cy = std::cos(cam.yaw),   sy = std::sin(cam.yaw);
+        Vector<float, 3, 4> forward { cp * cy, sp, cp * sy };
+        Vector<float, 3, 4> worldUp { 0.0f, 1.0f, 0.0f };
+        Vector<float, 3, 4> right { forward.y*worldUp.z - forward.z*worldUp.y,
+                                    forward.z*worldUp.x - forward.x*worldUp.z,
+                                    forward.x*worldUp.y - forward.y*worldUp.x };
+        const float rLen = std::sqrt(right.x*right.x + right.y*right.y + right.z*right.z);
+        right.x /= rLen; right.y /= rLen; right.z /= rLen;
+        Vector<float, 3, 4> up { right.y*forward.z - right.z*forward.y,
+                                 right.z*forward.x - right.x*forward.z,
+                                 right.x*forward.y - right.y*forward.x };
+
+        const float dx = moveAct.vector2.x * kMoveSpeed * kDt;
+        const float dy = moveAct.vector2.y * kMoveSpeed * kDt;
+        cam.position.x += right.x*dx + forward.x*dy;
+        cam.position.y += right.y*dx + forward.y*dy;
+        cam.position.z += right.z*dx + forward.z*dy;
+
+        CameraGPU& g = cameraBuf.value[0];
+        g.origin[0]=cam.position.x; g.origin[1]=cam.position.y; g.origin[2]=cam.position.z; g.pad0=0;
+        g.right[0]=right.x;   g.right[1]=right.y;   g.right[2]=right.z;
+        g.up[0]=up.x;         g.up[1]=up.y;         g.up[2]=up.z;
+        g.forward[0]=forward.x; g.forward[1]=forward.y; g.forward[2]=forward.z;
+        g.aspect  = float(window.width) / float(window.height);
+        g.tanHalf = std::tan(70.0f * 3.14159265f / 360.0f);
+        g.pad1    = 0;
+        cameraBuf.FlushDevice();
+    });
+
+    window.Render();
+    window.StartUpdate();
+    window.StartSync();
+    return 0;
+}
+#endif

examples/RTVolume/miss.wgsl

@@ -0,0 +1,7 @@
+// RTVolume miss (runs in SHADE). Vertical sky gradient — also what shows
+// through the any-hit cut-out cells.
+fn miss_main(ray: RayDesc, payload: ptr<function, Payload>) {
+    let t = clamp(ray.direction.y * 0.5 + 0.5, 0.0, 1.0);
+    rtAccumulate(mix(vec3<f32>(0.05, 0.07, 0.12),
+                     vec3<f32>(0.45, 0.60, 0.85), t));
+}

examples/RTVolume/project.cpp

@@ -0,0 +1,48 @@
+import std;
+import Crafter.Build;
+namespace fs = std::filesystem;
+using namespace Crafter;
+
+extern "C" Configuration CrafterBuildProject(std::span<const std::string_view> args) {
+    bool isWasm = false;
+    for (std::string_view a : args) {
+        if (a.starts_with("--target=") && a.find("wasm") != std::string_view::npos) {
+            isWasm = true;
+            break;
+        }
+    }
+
+    std::vector<std::string> graphicsArgs(args.begin(), args.end());
+    Configuration* graphics = LocalProject({
+        .projectFile = "../../project.cpp",
+        .args = graphicsArgs,
+    });
+
+    Configuration cfg;
+    cfg.path = "./";
+    cfg.name = "RTVolume";
+    cfg.outputName = "RTVolume";
+    cfg.type = ConfigurationType::Executable;
+    if (isWasm) {
+        cfg.target = "wasm32-wasip1";
+        cfg.defines.push_back({"CRAFTER_GRAPHICS_WINDOW_DOM", ""});
+        cfg.compileFlags.push_back("-msimd128");
+    }
+    ApplyStandardArgs(cfg, args);
+    cfg.dependencies = { graphics };
+
+    std::array<fs::path, 0> ifaces = {};
+    std::array<fs::path, 1> impls = { "main" };
+    cfg.GetInterfacesAndImplementations(ifaces, impls);
+
+    if (isWasm) {
+        cfg.files.emplace_back(fs::path("raygen.wgsl"));
+        cfg.files.emplace_back(fs::path("intersection.wgsl"));
+        cfg.files.emplace_back(fs::path("anyhit.wgsl"));
+        cfg.files.emplace_back(fs::path("closesthit.wgsl"));
+        cfg.files.emplace_back(fs::path("miss.wgsl"));
+        cfg.files.emplace_back(fs::path("resolve.wgsl"));
+        EnableWasiBrowserRuntime(cfg);
+    }
+    return cfg;
+}

examples/RTVolume/raygen.wgsl

@@ -0,0 +1,34 @@
+// RTVolume raygen (runs in GENERATE). Host-driven pinhole camera at
+// @group(3) (groups 0..2 are reserved by the wavefront pipeline:
+// 0 = WfParams, 1 = data heaps, 2 = indirect args).
+struct Camera {
+    origin:  vec3<f32>,
+    pad0:    f32,
+    right:   vec3<f32>,
+    tanHalf: f32,
+    up:      vec3<f32>,
+    aspect:  f32,
+    forward: vec3<f32>,
+    pad1:    f32,
+};
+@group(3) @binding(0) var<storage, read> camera : Camera;
+
+fn raygen_main(gid: vec3<u32>) {
+    if (gid.x >= wfParams.surfaceW || gid.y >= wfParams.surfaceH) { return; }
+
+    let pixelf = vec2<f32>(f32(gid.x), f32(gid.y));
+    let res    = vec2<f32>(f32(wfParams.surfaceW), f32(wfParams.surfaceH));
+    let uv     = (pixelf + vec2<f32>(0.5)) / res;
+    let ndc    = uv * 2.0 - vec2<f32>(1.0);
+
+    let direction = normalize(
+        camera.right   * (ndc.x  * camera.aspect * camera.tanHalf) +
+        camera.up      * (-ndc.y * camera.tanHalf) +
+        camera.forward);
+
+    var p: Payload;
+    p.color = vec3<f32>(0.0);
+
+    rtEmitPrimaryRay(camera.origin, 0.01, direction, 100000.0,
+                     0u, 0xFFu, 0u, 0u, p);
+}

examples/RTVolume/resolve.wgsl

@@ -0,0 +1,7 @@
+// RTVolume RESOLVE-stage tonemap: Reinhard + gamma 2.2 over the linear
+// accumulator.
+fn resolve_main(coord: vec2<u32>, hdr: vec4<f32>) -> vec4<f32> {
+    let mapped = hdr.rgb / (hdr.rgb + vec3<f32>(1.0));
+    let g      = pow(mapped, vec3<f32>(1.0 / 2.2));
+    return vec4<f32>(g, 1.0);
+}