Marcel

I cannot replay my example of NI above anymore, but I've also upgraded from some very old version in the meantime. Anyway, I had another look at it: Even with...

No, I just did ```diff diff --git a/gemrb/unhardcoded/bg2/avatars.2da b/gemrb/unhardcoded/bg2/avatars.2da index aa05a310d..13029ff36 100644 --- a/gemrb/unhardcoded/bg2/avatars.2da +++ b/gemrb/unhardcoded/bg2/avatars.2da @@ -234,7 +234,7 @@ 0x7F38 MEYE MEYE MEYE MEYE 19 2 1 * 0x7F39...

I mean I don't know what these "base" bams even are supposed to do and I doubt the originals use them.

But then it's two issues, right? Fixing the palettes to recover the original look, and then to do what this mod above has achieved, i. e. putting two BAMs together...

Any idea how to continue best here? I have a bunch of umlaut display errors and console warnings so I'm tempted to give it some progress, yet I don't want...

I tried the obvious thing: ```diff --- a/gemrb/core/PathFinder.cpp +++ b/gemrb/core/PathFinder.cpp @@ -46,11 +46,11 @@ namespace GemRB { -constexpr size_t DEGREES_OF_FREEDOM = 4; +constexpr size_t DEGREES_OF_FREEDOM = 8; constexpr size_t RAND_DEGREES_OF_FREEDOM...

> I don't know ... considering how few cases require this, we could just patch the data instead and save many cycles that way. Somebody needs to check all search...

Sounds like a good compromise.

Sounds like a bit of overkill. Not sure if and how EE uses sqlite, but isn't that way more boilerplate over reading the settings into a copy, and write it...

Good to know, I've never looked at a single C++ ORM lib.