Prompt Enhancer

I tested several generations using the Gemini model instead of Auto mode, and the results were inconsistent. Without proper animations, the output was distorted. When the enhanced prompt option is enabled, it seems to modify what I originally intended to do.

I am not sure what is happening, but when I typed the prompt directly without enhancements, the result was accurate. I think the AI prompt enhancer needs further improvement and training not to modify real intended scanario. Btw i’m happy with the last result below