Nui(縫い)Scene43 Dataset

Han-Hung Lee1, Qinghong Han1, Angel X. Chang1,2,
1Simon Fraser University 2Canada CIFAR AI Chair, Amii
arXiv Model Dataset Project Page

Description

The NuiScene43 dataset offers a curated collection of moderate to large artist-created outdoor scenes, filtered from Objaverse, for training and testing unbounded scene generation methods. We manually unified the scene scales and ground geometry to enable consistent joint training across all 43 scenes. Please see the paper for more details and github repos for the method and dataset above. If you find this dataset useful, please consider citing our work and Objaverse.

NuiScene43 statistics.
Category # Scenes # Sampleable Quad Chunks
Rural/Medieval 16 6.2 M
Low Poly City 19 4.9 M
Japanese Buildings 4 2.9 M
Other 4 1.2 M

Note: Each ground truth scene includes a depth image, full-scene RGB render, and two zoomed-in views. Note that occupancies are later unified by setting all values below ground-level to 1. The visualizations shown are before this step; the meshes are obtained from the processed occupancy using marching cubes.

Other

Scene ID: 857673bc44c8411ca8aca7cab3be7091

Scene ID: 0f73d55509644181937da3d41ba5623f

Scene ID: 6ca71306801744c48cdb66f6d573258b

Scene ID: 5fc65fd24ca647388d055dbc122b2c53

Japanese Buildings

Scene ID: 4a2536980b404fdcb42cb7eff9617e65

Scene ID: ee086eadd38744819350d73b37f0c2fe

Scene ID: 5f1822bbb40c43b097c4c98ecc697ed2

Scene ID: f82faef8e00d48948f12627b7dd4a836

Rural/Medieval

Scene ID: 19b198bb18be49498db3b647abebc755

Scene ID: 885f519b5a2e417c850736d3d9cd0ef3

Scene ID: e068ea6ae6bb41b4bdb51fda3091e5c4

Scene ID: e28598347055419db23d959dfb13a6f3

Scene ID: 590b2cae8e524f17a9ddda685cd3d8c3

Scene ID: 49e6306b29c74c679de8cec69473535b

Scene ID: 2e18c1baa9164093ad2e99e0a904363a

Scene ID: 3c9f5bd1b5fe416c931ef0b1e284cf0c

Scene ID: 8a512d2ca3d546319bbc7aaa696faa45

Scene ID: 6cdd292634f145d58a06c9bd9e349a14

Scene ID: 4c5e4490c71344ffaa34bc4abd0cbc11

Scene ID: 1a97e66594d241e9bf2d7e8ff7c967e8

Scene ID: 3cf5cfe07eec41fd928f1933696b35ff

Scene ID: 801103c094c04f45897be69ef6b27bf9

Scene ID: c5b0a81b3f254606b775134123a3e1e2

Scene ID: 63dd568f5bc64e8694d5f4252924c99e

Low Poly City

Scene ID: 3b61335c2a004a9ea31c8dab59471222

Scene ID: 1e2f91cf7085455baea97aa9e975cea0

Scene ID: 7f671f35e9ad4149b83451a3a92a6e2e

Scene ID: 040805292d1d4c929cfebdbe0007fff4

Scene ID: 19919b4a2518487b9c05024eb4277ae2

Scene ID: deb4dc75e62346c19c117bf61334eeb5

Scene ID: 41697300a4c643d089784b8688b2ed2c

Scene ID: 66e3384074574d58ba6e9d6969f6eae0

Scene ID: 1414092814104f0d841da6fe02f67c97

Scene ID: 6382d8d3bd054e05b4d8fc97e15e86af

Scene ID: 9580f0eb13814a93985ebc7c051fc910

Scene ID: 675a63829c2f475da42b187cc86d7ea0

Scene ID: 4a72d39a88264d02af6dde464bbcbb5c

Scene ID: 0e07ba9a139b4346a302b17aa85f2ee4

Scene ID: 6d0e0cd1b1314c4badfcceb5ccbcb5d5

Scene ID: 27c79cd96eba4561ab037d01be8828ec

Scene ID: dc39444ce3904ea5bca379b188f65ae0

Scene ID: da1728266f3149af9d5ac54316d28c8e

Scene ID: 5907c0fd28aa49daa2887ed2f3c02abb

Acknowledgement

We thank the authors of Objaverse for releasing their dataset, which made this research possible.

BibTeX


      @InProceedings{Lee_2025_ICCV,
          author    = {Lee, Han-Hung and Han, Qinghong and Chang, Angel X.},
          title     = {NuiScene: Exploring Efficient Generation of Unbounded Outdoor Scenes},
          booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)},
          month     = {October},
          year      = {2025},
          pages     = {26509-26518}
      }