OpenPrompt: Image Editing

Prompt works only—without exception—when you’re working with a starting image. You can always tell the AI to “regenerate / recreate / revise” as a keyword and then reference the image (Img_0 = first image). If you’ve loaded more than one image (Img_1 = second image, etc.). After that, you can apply all the commands typically used in image editing—such as adjusting sharpness, contrast, color spectrum, etc.—as well as style commands (e.g., “Generate an art style based on image Img_0 as a pencil drawing”). This allows you to perform both simple and highly complex image edits without using any editing software. Note: On newer models, the processing of commands is significantly more precise and also allows for specific adjustments. For example, making the character’s nose smaller is easily possible on these models (GPT 2.X for example)