It would be good to have a full-sized 3D for testing various things, beginning with how the 2D and 3D implementations of FFT scale up.

• Download the full 6000 milliseconds Penobscot project and access 3D volume
• Use OpendTect to export the full milliseconds
• Use segyio to load it and save it a .npy file (example code in this notebook and also the array with samples (values of times in milliseconds for each time slice)
• Store on Dropbox
• Use pooch (here) for retrieval, as suggested by Matt Hall on a discussion on SWUNG, if we want to do everything programmatically

Recording below shows Reset only working for the FFT spectrum plot, not the filter

Once enhancement 2 from #6 (adding dropdown menu for seismic volume choice) is completed:

1. record a demo of the app
2. add as a gif to the readme
3. adjust all the text to reflect we are now in post-hackathon, with a working prototype to move forward

The goal is to be able to apply this workflow to larger cubes. One idea is to use Dask to do this as it is possible to apply python functions.

Dask works by chunking the dataset into smaller cubes and carrying out the analysis on each chunk before recombining.

We would have to check the results to make sure the FFT are being carried out correctly

Assigned everyone to this collaborative effort.
Let's go back to the app and heavily comment it. I'd like to see it all the way to ELI5 level (explain it like I am 5). Egoistically for my benefit, but also anyone who wants to dig deeper into the code.

1. Download of segy seismic files from AWS
   Build on #4 to provide support for programmatic access to seismic volumes for the App
   This notebook is now the working prototype:
   https://github.com/mycarta/t21-hack-footprint/blob/main/dev/in-out%20work/Example_Seismic_IN-OUT_using_segyio.ipynb
   It only needs an option to allow user to store on a directory of choice

We are using later features of some of the packages, notably templates in panel which are only available in panel > 10.X it would be good to update the environment files so that it is easier to install

In the line of making the tool easier #12 to install a docker file would probably make this more portable! IT would also facilitate deployment.

Why not also do a GitHub CICD to build the image automatically so people can run docker run -p 5006:5006 Frida to get the whole tool to work!

Currently the code works on 2D time slices and when applying to a cube we iterate over each time slice one by one.

The objective is to run this code in native 3D

• Pad
• Run FFTs
• Unpad

The padding and un-adding can be extended from the current code.

For the FFT bumpy proposes an fftN version of their fft2. From what I understand this generalizes the FFTs to a N dimensional space.

However fft2 can be applied to a data cube (or N dimensions for that matter) but ONLY works along 2 axes.

Question: Do we need/want an N dimensional implementation or does a 2D ftt along each time slice work fine?

Using tools like Panel it may be possible to create an interactive app to create a visualise the results of at the filter before applying it to all levels

There are starting to have multiple bits of code being reused in different notebooks.

So we can make sure that we are always using the latest version of the code (unless reason not to) I propose to write the common functions inside a small package that can be called from the notebooks to start consolidating some of the code.