Hello,
I am new to iOS/iPadOS and Apple development.
I am interested in transforming and saving pointcloud data without rendering and using the shaders in the SceneDepthPointcloud example.
To that extent I have been able to successfuly transform the depthmap to local pointcloud data, like so:
If I capture the data without moving the device the pointcloud looks reasonable.
However I am not sure what's the correct way to transform it to world coordinates.
First I tried to follow the same implementation as in the shader in the SceneDepthPointcoud example, and that didn't seem to work when I moved the device.
I also tried the following:
I am still getting jumbled data when I move the device.
I would greatly appreciate any suggestions or tips.
I am new to iOS/iPadOS and Apple development.
I am interested in transforming and saving pointcloud data without rendering and using the shaders in the SceneDepthPointcloud example.
To that extent I have been able to successfuly transform the depthmap to local pointcloud data, like so:
Code Block /* x and y are the coordinates in the depth buffer and widthResFactor and heightResFactor are used to scale up the depth resolution to that of the camera */ let cameraPoint = simd_float2(Float(x*widthResFactor),Float(y*heightResFactor)) let localPoint = cameraIntrinsicsInversed * simd_float3(cameraPoint, 1) * depthValue; // from camera.intrinsics.inverse
If I capture the data without moving the device the pointcloud looks reasonable.
However I am not sure what's the correct way to transform it to world coordinates.
First I tried to follow the same implementation as in the shader in the SceneDepthPointcoud example, and that didn't seem to work when I moved the device.
I also tried the following:
Code Block var worldPoint = transform * simd_float4(localPoint, 1) // transform is taken from frame.camera.transform worldPoint = worldPoint / worldPoint.w;
I am still getting jumbled data when I move the device.
I would greatly appreciate any suggestions or tips.
