Just wild / simple thinking so far, if I'm not mistaken:
- The sections of script text spans in 2 areas, with each sub section marked with a "PSI3" close to its start. These markers are in 0135b2fc-0135b8bc and also in the main area spanning 01720a91-018c7d71. Assuming everything is filled wit script there, it means there are more than but about 1.7 to 1.8 MB of script (once turned into the game's format, with LZ77 already applied).
- It is all encoded in SJIS-like, so 2 bytes per character, which can be reduced to 1 if we work on the font. If I read patch v6 correctly, even though I don't understand why it seems to replace some Japanese with more Japanese, it seems to stick to 2-bytes encoding; So the tools from Richburn might be further developed to handle this?
- Richburn mentioned he was able to compress things better than the initial game - I'd be curious to get numbers on that.
- The free space at the end of the ROM is about 282kB.
- English is vastly larger than Japanese, I cannot remember the math but I'm afraid it is about a factor of 5?
So if my math adds up, we need roughly 1.8/2*5 = 4.5 MB, out of which we have 2MB. But then there's the Richburn compression upgrade ration, which we have to check.
Since the ROM is already 32 MB, which I read in various places is the max size, we need to find room (or compress like crazy).
So an interesting action would be to look for what we can get rid of / bypass / erase. Chances are, with a near-full 32 MB ROM, there is dead code, but there could also be unused media.
Has anyone tried to map out the ROM? (where are graphics vs sounds vs...)
Random corruption to try and find unused stuff comes to mind, but then the ROM needs to be played a lot to see if it is still working. I think I saw on the GBA thread that you had save states - do they span enough throughout the game that this could be tried?
(Enough thinking for one post.