chshin10 / epinet

EPINET: A Fully-Convolutional Neural Network using Epipolar Geometry for Depth from Light Field Images
MIT License
88 stars 39 forks source link

How to generate lytro 9*9 data? #4

Open ChenyangLiu opened 6 years ago

ChenyangLiu commented 6 years ago

Hi Changha,

I found that in your real-world result, you captured light-field images by a Lyrto Illum camera. And in your data dir, there are 81 images from different perspective. How did you generate these 81 images from Lytro file .lfr or .lfp? Is there any code or tools for me to generate these images on my own data?

chshin10 commented 6 years ago

You can use those tools: 1) https://sites.google.com/site/yunsubok/lf_geo_calib, 2) https://kr.mathworks.com/matlabcentral/fileexchange/49683-light-field-toolbox-v0-4

Hae-Gon commented 6 years ago

Ill

If you use a LF-toolbox, not geometry toolbox, EPINET will not work well. The Lytro and Lytro illum cameras suffer from dual distortions from main lens and microlens. The LF-toolbox only handles the main lens distortion. Thus, I recommend using the geometric toolbox for accurate depth estimation.