Many dyes are charge separated species (zwitterions, etc..) with the color arising from a charge transfer absorption in the UV-Visible spectrum. These types of transitions are highly sensitive to polarity, and can shift hugely (red to green, etc...) depending on the solvent they are in. I think that's probably what you are seeing. To test this theory, I'd dissolve the dye in methanol, and slowly add in a non-polar antisolvent to see if the shift in color happens. Alternatively, you could dissolve the materials in a series of alcohols of decreasing polarity (trifluoromethanol, methanol, ethanol, isopropanol, t-butanol, etc..) and see if you get a color shift. Nice video - pretty cool.