[whuang37]

Description

• A huge issue with depth is that texture enabling was super awkward where agents expected textures by default when calling enableDepth. Made that the default option
• Added camera projection matrices for SimulatorCamera. Since it center crops the points need to be reprojected to accurately represent world coordinates

const parameters =
  xb.core.deviceCamera?.simulatorCamera?.getCameraParameters();

if (parameters) {
  const projected = xb.projectWorldToCameraUv(worldPoint, parameters);

  if (projected?.inFrame) {
    const pixelX = projected.uv.x * parameters.width;
    const pixelY = projected.uv.y * parameters.height;
  }
}

Type of Change

• Bug fix
• New feature / enhancement
• New demo or sample
• Documentation update

Media / Screen Recordings & Screenshots (If Applicable)

Checklist

• Tested in simulator & device: Verified functionality in desktop simulator and/or physical hardware (where applicable).
• Large Assets ($\ge$ 1MB): Submitted separately to xrblocks/proprietary-assets via jsdelivr CDN.
• SDK Dynamic Dependencies: All new SDK dependencies are dynamically loaded at runtime.
• Security: Confirmed no hardcoded API keys or secrets are committed.

[dli7319]

Please check the ai reviewer comments:
🟡 enableDepth({texture: false}) / enableDepth({mesh: false}) Disables Both Outputs (src/core/Options.ts:443–448): Because useDefaults is computed as mesh === undefined && texture === undefined, passing {texture: false} (to opt out of the new default texture while keeping the mesh) or {mesh: false} sets useDefaults = false and resolves mesh ?? useDefaults / texture ?? useDefaults to false, turning both outputs off. Compute const useDefaults = mesh !== true && texture !== true; so {mesh: true} and {texture: true} still select a single output as documented in Depth.md, while {texture: false} and {mesh: false} preserve the other default output.
🟡 SimulatorCamera Custom Camera Mode (matchRenderingCamera = false, src/simulator/SimulatorCamera.ts:63–67): onBeforeSimulatorSceneRender updates this.camera.position and this.camera.quaternion via getWorldPosition/getWorldQuaternion without calling this.camera.updateMatrixWorld() (leaving this.camera.matrixWorld stale if renderScene is mocked or skips world-matrix updates, as in SimulatorCamera.test.ts:94) or syncing this.camera.aspect to sWidth / sHeight before renderScene(this.camera) (causing cropFromClip to double-scale X when sWidth !== sHeight). Call this.camera.updateMatrixWorld() and sync this.camera.aspect + this.camera.updateProjectionMatrix() before rendering.
🟡 getDeviceCameraClipFromView Signature & Shared Matrix Mutation (src/camera/CameraUtils.ts:58–70): Removing the first parameter renderCamera from exported getDeviceCameraClipFromView(deviceCamera, targetDevice) breaks external positional callers (unlike getDeviceCameraWorldFromView and getDeviceCameraWorldFromClip, which keep renderCamera first), and line 68 returns DEVICE_CAMERA_PARAMETERS[targetDevice].projectionMatrix without .clone() (risking mutation of the shared MOOHAN_PROJECTION_MATRIX / QUEST_3_PROJECTION_MATRIX singletons). Also note in Simulator.mdx:91 that xb.getCameraParametersSnapshot(...) works across both the simulator and physical headsets whereas simulatorCamera?.getCameraParameters() is simulator-only.
🟡 Unit Test Coverage (src/camera/CameraUtils.test.ts, src/simulator/SimulatorCamera.test.ts, src/core/Options.ts): Add unit tests covering projectWorldToCameraUv (center/corner UVs, inFrame: true/false, viewPoint.z >= 0 returning null, and round-trip consistency with transformRgbUvToWorld), SimulatorCamera.getCameraParameters() (undefined before first capture and after dispose(), plus non-square 1024x512 / 512x1024 center-crop clipFromView), and options.enableDepth() option combinations.