NLP is not happy with more or less uniform, low contrast images. Results can be anything and don’t really matter…unless you want to see the dust in the setup 
From left: Light panel captured wide open; NLP conversion; NLP conversion of wide open capture with radial mask; panel as captured with smaller aperture; NLP conversion of small aperture photo.
We can see that the radial mask does a fairly good job and could be improved by adding more masks, taking into account that falloff is not linear but “curved”. Falloff is reduced by smaller aperture…and we also see the dirt in the setup ;-(
All conversions were done with the same settings of NLP and due to its way of working, results vary from black to bluish to brownish hues.
Here is the description of how I set the radial mask I applied above.