search

DL3DV-10K / Dataset

News: the 10k dataset is ready for download.
https://dl3dv-10k.github.io/DL3DV-10K/
Other
330 stars 5 forks source link

How this dataset was made？ #16

Closed shenyehui closed 6 months ago

shenyehui commented 6 months ago

Thank you very much for your dataset, but I still need to make a similar dataset myself for my work, may I ask how your dataset was made and especially how the transforms.json file was obtained?

LuLing06 commented 6 months ago

Feel free to check our 'data preparation' section in the main page.

shenyehui commented 6 months ago

Feel free to check our 'data preparation' section in the main page.

Hello, I understand how you collect the data, but what tool is used to generate the transforms.json file?

LuLing06 commented 6 months ago

You can use COLMAP to estimate the camera parameters. The transforms.json is the same format as Nerfstudio. Refer to the issue4 for details.

shenyehui commented 6 months ago

You can use COLMAP to estimate the camera parameters. The transforms.json is the same format as Nerfstudio. Refer to the issue4 for details.

Thank you!

shenyehui commented 6 months ago

You can use COLMAP to estimate the camera parameters. The transforms.json is the same format as Nerfstudio. Refer to the issue4 for details.

I used nerfstudio to get all the poses of the images successfully, but the order of saving to transforms.json is quite confusing (not from frame_00001 to frame_00300), while the order of DL3DV's transforms.json is from frame_00001 to frame The order of DL3DV's transforms.json is from frame_00001 to frame_00300, is this something you guys adjusted the messy order to normal order by yourselves?（The command I use is (ns-process-data video --data data/nerfstudio/room.mp4 --output-dir outputs/room)）