Objective
This procedure describes the steps to edit an image using artificial intelligence accurately, avoiding unwanted changes to other elements of the image.
Key step
1. Understanding JSON
- JSON is a JavaScript-based data interchange format.
- It serves as a support for artificial intelligence to generate accurate changes.
2. Prepare the image and prompt
- Choose an image to edit (e.g. an image of Rome).
- Create a prompt to generate a JSON file that describes the image in detail.
3. Generate the JSON file
- Include in the description:
- General overview of the scene.
- Present subjects (people, animals, objects).
- Visual details and relevant features.
4. Use an advanced language model
- Use the most advanced language model available (e.g. Gemini Pro).
- Make sure you have output in JSON format without any additional text.
5. Make specific changes
- Give specific commands to modify the image (e.g. remove clouds, add objects).
- Make sure your changes are accurate and don’t affect anything else.
6. Check the accuracy of your changes
- Check that the requested changes have been made correctly.
- Make sure the final image meets your initial requirements.
7. Repeat the process with human faces
- Use an image of a human face as a reference.
- Repeat the process of generating the JSON and making changes.
8. Make sure the face remains unchanged
- Make sure the changes do not alter the appearance of the person’s face.
Notes
- Make sure you don’t overload the AI with overly complex requests.
- Always check the final output for any errors or unwanted changes.
Suggestions
- Use clear and detailed descriptions in your JSON for better results.
- Test different requests to see how the AI responds to specific changes.
SEO & GEO specialist.
