ayushtewari.com

What It Is

Ayush Tewari’s personal academic homepage is not an online course platform in the traditional sense, but rather an academic archive maintained in his role as an Assistant Professor at the University of Cambridge. The site brings together his long-running research in visual perception, 3D reconstruction, and neural rendering. For learners in computer vision and graphics, it functions more like an advanced self-study resource for “paper reading and code reproduction.”

Core Dimensions

• Subject Area: Deeply focused on the intersection of computer vision and graphics, especially neural rendering, 3D Gaussian Splatting, neural radiance fields, and generative models (GANs).
• Format: Mainly top-tier conference papers (SIGGRAPH, CVPR, NeurIPS, ICLR, etc.) and corresponding open-source code repositories. Some papers include project pages with intuitive visual demos, but there are no video lectures or slide decks.
• Instructor / Institutional Background: Extremely strong. The author is currently an Assistant Professor at the University of Cambridge; previously, he was a postdoc at MIT CSAIL working with Bill Freeman and others, and earned his PhD from the Max Planck Institute under Christian Theobalt. He is a leading scholar in this field.
• Certification / Certificate: No course completion certificate or certification is provided.

Pricing

Completely free. All papers and open-source code are freely accessible, reflecting the spirit of open science.

Pros and Cons

• Pros: The research is highly forward-looking and coherent, spanning from earlier 3DMM-based face reconstruction to recent 3D Gaussian Splatting, and it has covered nearly every major hotspot in the CV field. A high proportion of the work is open source, with a large amount of high-quality code that is very helpful for researchers reproducing algorithms.
• Cons: The learning curve is extremely steep. Reading top-conference papers and reproducing frontier code requires a very strong foundation in mathematics and programming. There is little systematic teaching guidance, so it cannot replace foundational courses. There is also no interactive Q&A support.

Who It’s For

This resource is only suitable for master’s or PhD students in computer vision and graphics, as well as researchers working on cutting-edge algorithms. If you are looking for the latest baselines and code in areas such as NeRF, 3DGS, or differentiable rendering, this is a treasure trove. But if you are a beginner hoping to get started with 3D vision from scratch, this site is not a good fit.

Access from China

As a personal academic homepage, it is directly accessible from mainland China with normal loading speeds. Some code repositories hosted on GitHub may require a proxy to access.