I have a question about the most appropriate way to normalize camera extrinsic translation vectors, and I'd appreciate your thoughts. I'm considering two main approaches: On the one hand, we could ...