Random Conversations: Difference between revisions
Springheel (talk | contribs) No edit summary |
Springheel (talk | contribs) No edit summary |
||
Line 1: | Line 1: | ||
''Written by Springheel'' | ''Written by Springheel'' | ||
When writing the newer vocal scripts, I came up with the idea of including a random sequence of questions and responses that can be used to create | When writing the newer vocal scripts, I came up with the idea of including a random sequence of questions and responses that can be used to create believable conversations between AI. The following vocal sets support this (older ones do not): | ||
Revision as of 21:30, 4 July 2018
Written by Springheel
When writing the newer vocal scripts, I came up with the idea of including a random sequence of questions and responses that can be used to create believable conversations between AI. The following vocal sets support this (older ones do not):
1. vocal set_commander
2. vocal set_jack
3. vocal set_cynic
4. vocal set_drunk
5. vocal set_maiden
6. vocal set_moor
The sequence goes like this:
AI 1: "convo_question" vocal, ex, "So, how did the game go last night?"
AI 2: "convo_answer/response" vocal, ex, "Could have been better."
AI 1: "convo_close" vocal, ex, "Got it."
Each "convo_question" asks for a judgement response. "How was X? What did you think of X?" etc. Each vocal set has four or five questions/answers that are chosen from randomly. The overall effect is pretty good, I think, though some question/responses combinations make more sense than others.
The sitting AI in Mission 1: A New Job use this system. Basically, you create two conversations--one with AI 1 asking the questions, and the other with AI 2. Then use a trigger of some kind to start the conversations when you want them. A trigger_entityname could be used for patrolling AI. If the AI are sitting near each other, a trigger_sequencer could be used to have multiple conversations over time.