If max color is on 2bytes (using 4 hex digit) then it means the max value is FFFF = 1111 1111 1111 1111
if it's RGBA it's 4 bit per color
if it's RGB it's 5 bit per color +1 unused 0 11111 11111 11111 = 01111111 11111111 = 0BBBBBGG GGGRRRRR = 2 bytes. Each byte store a mix of two colors, contrary to 3 bytes with one color per byte)
known as
15-bit color
11111 = 0x1f = full color, but you need to set them to binary to find the correct color.
White is not 0x1f1f1f but 0x7FFF (11111 11111 11111)
PS: it's only a supposition on how the color can be coded in your current game
You'll have to test it to know if it's working, and if it's in
big or little endian (the unused bit at the start or the end, and the direction of the color reading)
0RGB
RGB0
0BGR <- more likely to be this one
BGR0
and high chance it's in little endian (reversed order) 0x7FFF must be written FF 7F in your Action replay code.
Read the link above to learn how to convert 24bit to 15bit
Edit:
It could be that it's not using a color definition, but a palette position.
For example you have a palette with 32 positions (I randomly picked that number). each position has it's own color definition, and the sprite use the palette position instead of a color definition for of its each pixel.
