---
tags:
- novel view synthesis
- NeRF
- 3D Gaussian Splatting
- 3D Vision
- Content Generation
- text-to-3d
- image-to-3d
size_categories:
- 100B<n<1T
---
# DL3DV-10K Dataset Sample 
Here is the repo for a sample (11 scenes/videos) of the full DL3DV-10K dataset. Each scene has a raw 4K video and the processed colmap results. 

## File Structure Example
``` bash
5c3af581028068a3c402c7cbe16ecf9471ddf2897c34ab634b7b1b6cf81aba00
├── colmap
│   ├── colmap
│   │   ├── features.h5
│   │   ├── global-feats-netvlad.h5
│   │   ├── matches.h5
│   │   ├── pairs-netvlad.txt
│   │   └── sparse
│   │       └── 0
│   │           ├── cameras.bin
│   │           ├── database.db
│   │           ├── images.bin
│   │           ├── models
│   │           │   └── 0
│   │           └── points3D.bin
│   ├── images
│   │   ├── frame_00001.png
│   │   ├── frame_00002.png
│   │   ├── ....
│   ├── images_2
│   │   ├── frame_00001.png
│   │   ├── frame_00002.png
│   │   ├── ....
│   ├── images_4
│   │   ├── frame_00001.png
│   │   ├── frame_00002.png
│   │   ├── ....
│   ├── images_8
│   │   ├── frame_00001.png
│   │   ├── frame_00002.png
│   │   ├── ....
│   ├── temp_images
│   └── transforms.json
└── video.mp4


```


# DL3DV-10K Benchmark 
Please refer to this [repo](https://github.com/DL3DV-10K/Dataset) for instruction.