Posted July 27, 2020 by EsoDev
Well, this week turned out to be weird. I focused on the nonverbals, but it led me to interesting places.
Maybe let's start with the technical. A short while ago I described a kind of grammatical AI approach, this does 100% work for non-verbal communication and allows for very interesting features to be implemented, but it has the issue of requiring a model to keep the data in order. Here's where some of the interesting stuff happened. I realised that I can use the consistency flag system I developed a while ago to support narrative consistency for some modules (specifically kissing and headpats) to achieve this, but considering how it would work I came to the conclusion that the system as it currently exists has issues. So I re-wrote it. Which included an optimisation to the system (not needed, really) which I instantly squandered by giving it a semantic wrapper. In any case, it's adequate for what I need at the moment. And the semantic implementation means I can basically re-write the whole system anytime I want without any problems since the semantic calls remain unchanged. So when I get this rolling, I'm going to be building a kind of nondeterministic finite-state automata, and I am both a bit scared and excited to get to that.
Now for the real rabbit hole of this last week. I decided to maybe, possibly talk with a few of the man-creatures to get some information from them about male nonverbal cues. Firstly this completely shook my world about male nonverbal cues in general, making me realise that I might have a tendency to give male characters feminine body language. Not specifically in the game, just in general. However, I think I came out at the end of it with a far better understanding of how men emote and what I should focus on or should avoid in the future. There's definitely a bit of a weird feeling to be confronted with this issue, but I can't say I really paid much attention to men in general.
Moving into this week, I've gone over the existing nonverbal cues that could be found in the text already existing in-game. A lot of the writing is salvageable, and most definitely some of it can be moved towards being variable in the future. I now have a base of texts that I can work with, as well as some insight into stacking them. That is to say, composing multiple nonverbal cues together into more complicated expressions. It's definitely an interesting experiment, but it goes very deep into what I wanted to be doing from the very start.