The comparison is a 15 stop ND. Based on the cloud positions it appears they exposed the long exposure first, then when it was finished exposed the second exposure. 15 stops means that even if the short exposure was fairly quick (say a 60th of the second), the 15 stop exposure would have been 4 minutes. I assume the sunset colors were more apparent when that exposure started.
While there is some color there (from the sunset), there isn’t a noticeable color shift which is a big problem with heavy ND filters. If you don’t filter out the IR light pollution on really long exposures the colors shift dramatically.
I’ve used the Lee Big Stopper and Super Stopper, color shift is obvious and you can’t always get it right in post. I’ll be ordering these.
Yes, that's the whole point. IR contamination affects all Bayer CFA filters (which dyes are transparent to IR), throwing off the ability to 'correctly' White Balance. So whatever the scene color is, it gets desaturated by an addition of non-existing color, unless IR is better controlled.
The only optical question I'd have is how much vignetting does the 2 mm thickness produce. After all, more oblique rays (towards the corners of the FoV) travel a longer distance through the medium than those on the optical axis. Glass that's dyed in mass, causes more vignetting than metal evaporation on one surface.
I also wonder about whether the NDIR-10 and NDIR-15 have proper mechanical gaskets/protection (like the Big-stoppers) against light creeping in at the rear of the filter, or does one have to use some additional IR-blocking (cloth won't do) cover material?
Cheers,
Bart