---
title: README
emoji: ⚡
colorFrom: pink
colorTo: indigo
sdk: static
pinned: false
---

This is the space for DL3DV-10K organization. If you are looking for the DL3DV benchmark or DL3DV-10K dataset, please find a version that fits your needs and **accept the term-of-use** in the Dataset card section. Otherwise you cannot download the dataset. **We provide the preview of the [DL3DV-10K dataset](https://dl3dv-10k.github.io/DL3DV-10K-Preview/), the [DL3DV-140 benchmark](https://dl3dv-10k.github.io/DL3DV-Benchmark-Preview/) and the [DL3DV-Evaluation](https://dl3dv-10k.github.io/DL3DV-Testing-Split-Preview/)**.

The whole dataset is extremly large. Here are different versions for different needs.
  - 480P resolution frames with poses (~730G). [Dataset link](https://huggingface.co/datasets/DL3DV/DL3DV-ALL-480P).
  - 960P resolution frames with poses (~2.8T).[Dataset link](https://huggingface.co/datasets/DL3DV/DL3DV-ALL-960P).
  - 2K resolution frames with poses (~11T). [Dataset link](https://huggingface.co/datasets/DL3DV/DL3DV-ALL-2K).
  - 4K resolution frames with poses (~44T). [Dataset link](https://huggingface.co/datasets/DL3DV/DL3DV-ALL-4K).
  - 4K videos (~7T). [Dataset link](https://huggingface.co/datasets/DL3DV/DL3DV-ALL-video).
  - COLMAP cache folders (may be useful if you need post-processing based on those caches). [Dataset link](https://huggingface.co/datasets/DL3DV/DL3DV-ALL-ColmapCache).
  - DL3DV-benchmark provides 140 scenes that are sampled from DL3DV-10K. It contains nerfstudio and gaussian splatting formats. [Dataset link](https://huggingface.co/datasets/DL3DV/DL3DV-Benchmark)
  - **_DL3DV-Evaluation provides 55 scenes that are not covered in the DL3DV-10K. It is used for evaluation purpose_**. [Dataset link](https://huggingface.co/datasets/DL3DV/DL3DV-Testing-Split)


You can either use git to download the whole dataset or use a download script to download a subset. Details can be found in each repo. 

If you find any problems, feel free to raise an issue in [github repo](https://github.com/DL3DV-10K/Dataset/issues).