关于向Autodesk Reality Capture API传入相机拍摄角度参数的问询

Great question! When working with the Autodesk Reality Capture API, there is indeed support for providing explicit camera position/extrinsic parameters to boost reconstruction accuracy—perfect for your controlled circular capture setup (24 shots at 15° intervals).

Here's what you need to know:

• The API accepts pre-defined camera pose data through parameters typically named extrinsics or cameraPositions in your job submission payload. This lets you pass in rotation and translation values for every image you upload.
• For your circular capture scenario, you can structure each camera's data using either:
  • Rotation matrices (3x3) + translation vectors (3x1)
  • Euler angles (pitch, yaw, roll) + X/Y/Z translation coordinates — just make sure to confirm the API's expected coordinate system (e.g., right-handed vs. left-handed) to avoid alignment issues
• Example snippet of how this might look in a JSON job request:

{
  "input": {
    "images": [
      {
        "url": "your-image-1-url",
        "extrinsics": {
          "rotation": [1.0, 0.0, 0.0, 0.0, 0.866, -0.5, 0.0, 0.5, 0.866],
          "translation": [5.0, 0.0, 2.0]
        }
      },
      {
        "url": "your-image-2-url",
        "extrinsics": {
          "rotation": [0.866, 0.0, 0.5, 0.0, 1.0, 0.0, -0.5, 0.0, 0.866],
          "translation": [4.33, 2.5, 2.0]
        }
      }
      // ... repeat this structure for all 24 images
    ]
  },
  "output": {
    "format": "obj"
  }
}

• A few key tips:
  • If you didn't spot this in the docs initially, check under advanced or calibration sections—sometimes these parameters are tucked away in specialized settings
  • Verify the API's latest version documentation; Autodesk occasionally adds or refines these features in updates
  • Double-check that your camera coordinate system matches the API's requirements to avoid skewed reconstructions

内容的提问来源于stack exchange，提问作者Peter