update on my ai waifu app, she is a pirate. can use web search, react to images even her own image. using qwen 3 VL for the llm and vision model, the tts is using omnivoice tts api that i made the api myself .
testing PuruPuruPNGtuber using see-through-webui to convert the image into layers and use codex to add the character. this is actually good for making ai character more alive.