Question about availability of the artist-designed city/village training scenes

[chendi23]

Hi, thank you for releasing the code and pretrained models for BlockFusion.

I am currently working on a research project about semantic-layout-guided 3D scene generation. In your paper, the outdoor experiments on city and village scenes are very relevant to our setting. I noticed that the paper mentions that the city and village scenes are designed by artists, and that 10K training blocks are cropped from these scenes.

I would like to ask whether the original artist-designed city/village scenes, the cropped training blocks, or the corresponding semantic layouts are publicly available or planned to be released.

More specifically, I am interested in the following data if available:

• the city/village scene meshes used for training;
• the cropped 3D blocks used to train BlockFusion;
• the semantic layout labels for the outdoor scenes, such as road, tree, solar panels, cars, houses, pine, cypress tree, and ground;
• any preprocessing scripts or instructions for preparing similar outdoor scene data.

If the dataset cannot be released, would you be able to share any guidance on how to construct a compatible outdoor training dataset for BlockFusion-style experiments?

Thank you very much for your time and for the great work.