GenAI is rising in 2025+. How do these generative AI models work ? Most of the time every design crossed through the nature or it's behavior. it was affected to GenAI also. In early stages GenAI or upgraded LLMs works similar way. It was text mining and understood the text relations those like NI/Natural intelligence. It's about how do we think ? Most of the time our brain used to think in words. As an example if we want to watch tv. In minimerlistic way we matched these word relations like we used to turn on the TV in remote. (tv-remote + switch + turn-on). Also the same way early stages in genAI uses words + words relations (text mining) to generate image. If we want to generate face before it divides parts of the face in words in simplest way which like eyes ,nose, ears,cheeks etc. Looks like we imagine a face in NI(Natural Intelligence). These are the LLM models solutions. Thats why dynamic prompts give variable outputs. After that it uses to generate image from generator and discriminator cycles. Think it likes software engineer/artist team and qa/errors-check team. First team generate image or sequences from it's imagination using pre-trained images after it's tested/checked by discriminator(qa team). This is parallel process utill get some satisfied output/outputs.This is the brief introduction of how does the logic behind GenAI think/work in image/video generation?
0 Comments