Answer:
The 3 parts combine into みらい (how "future" (未来) in Japanese is pronounced)
How does it work:
Every image is a hirogana character split into 3 parts - red, green and blue channels, each one is third of the image, one 120° sector. In every part these sectors "point" at their corresponding, color coded by geometric shapes, azimuths. How shape encoding works: starting from triangle, which means zero, every additional corner of a shape adds 1. Square means 1, Pentagon means 2 and so on. To see the original character one needs to increase image contrast, then undo the rotation of the sectors accordingly - in other words, make them point back "north" (top of the image).
Part 1:
mi - み. Azimuths - R: 120 G: 240 B: 360.
Unsolved:
Solved:
Part 2:
ra - ら. Azimuths - R: 90, G: 180, B: 270.
Unsolved:
Solved:
Part 3:
i - い. Azimuths - R: 60, G: 120, B: 180.
Unsolved:
Solved:
Notes:
Puzzle was made in Blender (character images) and python (shapes). The lines on the images are sector pointers; what azimuth they are pointing at. This puzzle may have suffered from lack of clues, apologies.
Also:
Material nodes used to make this pizzle: Embarassingly messy, but it was fun :)