The models look extremely detailed. Are these 3D scans or self-made assets?
Most of the 3D objects are my own photogrammetry scans. The knight and the horse was my first attempt at creating a 3D model using AI, which created a 3D model from just a single photo. Post-processing is then necessary, of course. The landscape I also created with Photogrammetry. I made hundreds of screenshots with Google Earth and then let the Photogrammetry software create the 3D model of the landscape. And I use my own "planting script" to place the trees on the landscape.
I experimented with local AI based photo-to-3D-model-conversion in the past. I think, currently the best models available are probably Hunyuan 3D 2.1 and Trellis.
