Concept clearity on intrinsics affter depth alignment

So, according to this issue https://github.com/IntelRealSense/librealsense/issues/9001  you had mentioned that "My understanding is that depth and color would still have separate intrinsic, but when aligning you can use the intrinsics of the stream that you have aligned to (i.e if you are using depth to color alignment then use the color intrinsics).”

So from what I understand from your statement is that, assuming I have done my alignment from depth->color. Then in this case, 

• You warp the depth map to match the color image's pixel grid.
• Use the intrinsic of the color camera because the final output will be in the color camera’s perspective.

So for projecting back the pixel coordinates ( using YOLO-> centroid) to get a 3d object position in camera frame, I use the intrinsic of color and not the depth. The reason being, I have aligned/transformed my depth frame to color frame, thereby while projecting the pixel coordinates to get depth, I use color intrinsic .

Summary

• Depth and color cameras have separate intrinsics.
• When aligning one stream to another, use the intrinsics of the target stream.
• Depth → Color Alignment → Use color intrinsics.
• Color → Depth Alignment → Use depth intrinsics
  
  

Is my understanding/concept correct?