Currently entity definition is done in a very crude way, based on the name, age and gender of character (and those are actually guessed, they are not reliable)
same for non-character entities (eg. locations
While we offer to users the possibility to edit the entity settings, this is a slow process when you have multiple characters, and some users only want to focus on the fun part of creating and visualizing a story, not doing the data entry (it is a movie assistant job)
Solution
Use a LLM to infer the characteristics, by using the scene and/or synopsis and/or the full screenplay as a context.
Context
Currently entity definition is done in a very crude way, based on the name, age and gender of character (and those are actually guessed, they are not reliable)
same for non-character entities (eg. locations
While we offer to users the possibility to edit the entity settings, this is a slow process when you have multiple characters, and some users only want to focus on the fun part of creating and visualizing a story, not doing the data entry (it is a movie assistant job)
Solution
Use a LLM to infer the characteristics, by using the scene and/or synopsis and/or the full screenplay as a context.