Hmm, my English has big problem...

I have now compressed subfont images for 32 bitts Kanji font, and I can see
those by tweak, and which was my test to compare the 16 and 24 bitts fonts
between yours and mine.

Now, I'm going into next step, ie. to make Kanji subfonts non-multiple of eight,
such as 14 dotts, 12 dotts and even 8 dotts.  To achieve those, we need an
algorism (of course, we can do it only by strong force, too :-)) to pack redundant
bitts included in BDF BITMAP lines.  For an example, each bitmap of k14.bdf file
includes 14 lines as below lines:  (those are expressed as hexadecimal number)

x000 0000 0000 000x
x000 0000 0000 000x
         ....
x000 0000 0000 000x

where 'x' bit are the redundant bit, and are always 0.

I think we have to eliminate all the redundant 'x' to make 14x14 bit subfont
image.

Kenji