Just processed my first 3D Gaussian Splatting model from a single 360 video I took of the cherry blossoms on the University of Canterbury campus.
The quality is a bit rough, but I was genuinely surprised it actually worked to render such a complex scene (all those tiny flowers!). It's a testament to the power of this new technology.
I'm still learning the pipeline. For those of you who've played with this, any kind suggestions for improvement? Especially around how to handle the "fuzzy" artifacts that come from 360 video inputs? 👇
This Gaussian Splat project was optimized for an immersive **first-person Walk Mode** experience.