Task 1¶

Prompt: Succinctly explain why I should take 16-825: Learning for 3D Vision at CMU

Response:

You should take 16-825: Learning for 3D Vision if you want to understand how modern AI turns 2D observations into useful 3D representations—a core problem behind robotics, autonomous driving, AR/VR, and generative 3D. (Computer Vision @ CMU)

The biggest reasons:

  • Highly relevant: Covers neural 3D representations, differentiable rendering, NeRFs/neural rendering, single-view 3D, generative 3D, meshes, and point clouds. (Learning for 3D Vision)
  • Research-oriented: It sits directly at the intersection of modern deep learning and 3D vision—a particularly active area of computer vision research.
  • Hands-on: Assignments implement these ideas rather than just discussing them, with a substantial final project. (Learning for 3D Vision)
  • Good CMU leverage: It complements rather than duplicates 16-824 (broader visual learning) and 16-822 (classical geometric vision), giving you a more modern learning-based 3D toolkit. (Computer Vision @ CMU)
  • Career value: If you're interested in robotics, computer vision, graphics, embodied AI, or generative models, 3D perception is one of the most transferable technical foundations you can have.

In one sentence: take it if you want to move beyond “recognizing what's in an image” toward teaching machines to reconstruct, represent, and reason about the 3D world.

Task 2¶

image

source: https://www.sfmoma.org/exhibition/alexander-calder-scaling-up/

Task 3¶

image

source: https://eu.36kr.com/en/p/3540030668353416