Radiance Fields Techniques - Review

PhD Review - 20/01/2026 | 3D Reconstruction - Deep Learning - Computer Graphics

The presentation below was a PhD state-of-the-art review on 3D reconstruction with radiance fields techniques. Although these topics are outside the scope of my PhD research, these algorithms are highly influential in the robotics community. Recognizing their importance, and as a result of my personal curiosity and fascination with these novel techniques, I provided a gentle introduction to the topic during one of my ongoing work update meetings.
(This pdf version of the pptx slides is linked to the projects webpages and github repo, just click on the GitHub logo or the “World” logo)

Some Experimental 3D Gaussian Splat Reconstructions

Explore Some of my Gaussian Splat reconstruction obtained with COLMAP SfM and nerfstudio gsplat. These models are obtained from few (200-300) phone images, extracted from video sequences. The purpose was mainly to experiment with the available radiance fields tools rather than obtaining optimal reconstruction results.