Same dataset and Colmap reconstruction as used here: https://superspl.at/scene/997b1a59
But this time trained with Spirula Studio: https://github.com/harry7557558/spirula-studio
3M splats, ~30k iterations.
How do you compare the detail rendering to the other one? Slang-Splat was definitely faster and I would say more detailed on the nameplate text. Spirula seems to have better rendered the Victor Hugo page.