AI Hints
Home›Generative AI Tutorial›How AI Creates Images
Generative AI · Lesson 45 of 77

Image Prompting and Control

Getting exactly what you imagine.

⏱ 2 min read▶ Video at 1:43:23🎓 Free
Watch this lesson in the Generative AI courseWatch on YouTube ↗Full Generative AI course

Image models reward good descriptions. A few habits will lift your pictures from random, to remarkable. And beyond prompts, there are tools that give you real control, over pose, layout, and edits. In this lesson, you will learn both.

Image Prompting and Control
From the video: Image Prompting and Control
  • Describe like a photographer
  • Tools for real control
  • Edit, do not just generate

A formula for image prompts

A reliable formula has five parts. The subject, what is in the picture. The setting, where it is. The style, such as watercolour, photograph, or cartoon. The lighting and mood, like golden hour, or soft studio light. And the camera, or composition, like close up, or wide angle. Compare, a dog, with, a golden retriever puppy, running on a beach, at sunset, warm light, wide angle photograph.

  1. Subjecta golden retriever puppy
  2. Settingrunning on a beach
  3. Stylea realistic photograph
  4. Light and moodwarm golden-hour light
  5. Camerawide-angle composition

Negative prompts and seeds

Two more controls help. A negative prompt lists what you do not want, such as blurry, extra fingers, or text. And the seed, is the number that picks the starting noise. Keep the seed the same, and change one word, to see exactly what that word does. Change the seed, to explore fresh variations of the same idea.

Negative promptwhat to avoid: blurry, extra fingers
Seedpicks the starting noise
Same seedcompare changes fairly
New seedfresh variations

Beyond prompts: real control

For real control, there are powerful tools. Image to image, starts from your own sketch, or photo. Inpainting, repaints only a masked area, to remove or replace an object. Outpainting, extends the picture beyond its edges. And tools like ControlNet follow an outline, a pose, or a depth map, so the layout is exactly what you drew. Creativity, with precision.

Image to imagestart from your own picture
Inpaintingrepaint just one area
Outpaintingextend beyond the edges
ControlNetfollow a pose or an outline

Using image AI responsibly

A few responsible habits. Respect the rights of artists, and check the licence of any tool you use. Do not create misleading images of real people. Label A I generated images, where honesty matters, like news and advertising. And remember that models can reflect biases in their training data, so review results with care.

Do

  • Check the tool's licence
  • Label A I images where it matters
  • Review for bias and errors

Avoid

  • Misleading images of real people
  • Copying a living artist's style to deceive
  • Passing generated images off as photos

Summary

Wonderful. Describe subject, setting, style, light, and camera. Use negative prompts and seeds, and use image to image, inpainting, and ControlNet for control. And use image A I responsibly. Next, we move to sound, and to video, where the same ideas return.

Key takeaways
  • Subject, setting, style, light, camera
  • Negative prompts and seeds
  • Inpainting and ControlNet for control
Next up: Speech and Voice — how AI speaks and listens