I'm curious how AiColorPalettes handles the mapping from natural language prompts (like "sunset" or "cyberpunk") to color palettes. Is it using a fine-tuned vis

I'm curious how AiColorPalettes handles the mapping from natural language prompts (like "sunset" or "cyberpunk") to color palettes. Is it using a fine-tuned vision-language model like CLIP to associate semantic concepts with colors? Or is there a custom embedding space built from annotated palette datasets?