==== using DIC 1 ====
this is based on variance. works ok.
======using DIC 2 ========
based on floodfilling DIC outside. color intensity increases near embryo
so a lot of the outside comes with it. bg is not homogenous.
========== using RFP ==========
using floodfill. works quite well. suggests that a 3d threshold would do the job well.
======= using RFP =========
using percentile thres 2d.
* works well later
* a lot of noise early, need something in addition
====== using DIC ===========
uses abs-grad on dic directly
fill holes
needs smoothing to fix jpeg artifacts
==== using DIC ==========
bilateral filter. dim 2,2, interval 10