How to run on different set of images?

[kajalsanklecha]

The given readme shows how to run the code on the DTU dataset but if we want to run it on some other set of images, how do we do it?
What all data is needed other than the images to make the code run?

[eduardramon]

Hi @kajalsanklecha! I created this dummy example for data generation that shows how to use the current code with any data.
https://github.com/eram1205/idr/blob/main/data/generate_dummy.py

[lioryariv]

Thank you @eram1205 that's amazing!
To run the code under the fixed cameras setup you need the images, the corresponding silhouettes (object masks), and cameras parameters.
Note that for DTU we re-normalize the cameras so that their visual hulls are contained in the unit sphere.
We plan to publish the re-normalization code in the near future.

[eduardramon]

Thanks @lioryariv ! Let me know in case you'd like to include it in the project, I'd create a clean PR for that. The scene looks like below. Simple but works :)

[nsiemeni]

Hello, I was trying to run IDR on new data. I made sure to supply images, masks and an initial cameras.npz but i was not able to run the pre-process script to calculate the new cameras.npz file.
How do you initially create the cameras.npz file and calculate the normalization matrix?
Thank you.

[ykasten]

Please see my answer here: #15 (comment) and let me know if you have any additional issues.

[chunyangL7]

Hi, I used colmap to generate camera parameters, and processed the camera parameters with the scripts from NeuS to get the cameras_sphere.npz file. The script preprocess_cameras.py provided by IDR just use world mat to calculate the normalization, so I used preprocess_cameras.py to process cameras_sphere.npz, and got the normalization matrix.