Allow users to edit image outputs without having to start from scratch.

Two current ideas.
Allow the user to train the AI by rating their satisfaction with the image output from 0 to 10. The system could create tags (variable identifiers) for each adjustment within that session. Once a style or character is like the system can save and name/identify the variables. These tags could be added to the prompt and saved for later use. Tags can be created for the image style, character, clothing, or objects within the image.

When users select ArtSmart generated images from "my gallery" as a reference image an additional box will appear offering the option to keep or change. The user could then provide text prompts for the specific variables they would like to adjust within that image.
example keep: background, eye colors, clothing, and hairstyle
example change: skin tone to F6E311, adjust pose to standing, character is smiling