[ FF7 ] ff7_redxiii_eye_fix

[Moe2912]

Hello, could it be that the variable "ff7_redxiii_eye_fix" in "ff7_init_hooks" should be dependent from the game's version? I have the german version, and the address 0x9079ec points to the string "\field\char" instead of any .tim file, which I would have expected.

[julianxhokaxhiu]

We do use relative addresses, despite the name the value for each var is relative to the language. So the real address is the value the variable contains during runtime. Are you having any particular issue?

[Moe2912]

But ff7_redxiii_eye_fix does not seem to be relative.

[julianxhokaxhiu]

ff7_externals.field_models_eye_blink_buffer is and that's the starting address we use ( which is relative to the language ) and then the offset movements are identical to other languages. Are you facing any particular issue?

[Moe2912]

Yes, ff7_externals.field_models_eye_blink_buffer is relative, but ff7_redxiii_eye_fix isn't. Let me try to explain:

You patch the bytes at address "ff7_externals.field_models_eye_blink_buffer + 0x58". For the german version this equals to 0x907990. At this location are 2 * 4 bytes containing 2 references to string constants. In ghidra for me it looks like this:

So, the patch now changes 0x000000005c7a9000 to 0xec79900000000000

So basically, you set the second string reference to null and overwrite the first string reference to a valid one. In the german version this points to a string containing "\field\char" and not to a tim file. I don't know, what the purpose of this patch is, but I'm guessing it should swap the 2 values? In this case the patched bytes for the german version should be "0x5c7a900000000000".

I'm not facing any issue here, I'm currently reverse engineering ff7 using FFNx (which is a great help btw.) and just thought this looked wrong.

btw: shouldn't it be "sizeof(ff7_redxiii_eye_fix)" instead of "sizeof(ff7_redxiii_eye_fix) - 1"? You're patching only 7 of the 8 bytes here.