Write a visual brief, not a shopping list of pictures.
A consistent ebook starts with a few declared choices: what images must help the reader understand, which visual language belongs to the subject, and which details must remain stable across chapters. “Add ten illustrations” describes a quantity. It does not describe a job.
A practical guide for first-time consultants
Purpose: make abstract offer-design decisions visible. Interior system: simple two-color diagrams and annotated worksheets, never decorative stock scenes. Consistency: navy lines, coral decision points, off-white background, same label style. Do not generate: fake dashboards, client logos, realistic people, or numbers presented as real results.
A cover attracts. An interior visual explains.
| Image | Its job | Quality question | Common failure |
|---|---|---|---|
| Cover | Signal subject, tone, and audience at thumbnail size | Is the title readable in a small storefront preview? | Detailed art overwhelms weak typography |
| Diagram | Show a relationship, sequence, or decision | Can the reader explain the connection after viewing it? | Decoration is mistaken for information |
| Screenshot | Demonstrate a real interface state or step | Is the relevant control obvious and current? | Private data, tiny labels, or stale UI |
| Scene or illustration | Set context or support a narrative moment | Does it preserve characters, objects, and setting? | Continuity changes from image to image |
Review the image in context, not in the generation gallery.
Mark the need
Place an image note beside the paragraph it must support. State the reader question the visual will answer.
Specify invariants
Record the palette, composition, recurring subject details, labels, aspect ratio, and elements that must not appear.
Generate or source
Choose the safest method for the job. Real evidence often needs a real screenshot or licensed photograph, not synthetic approximation.
Inspect the facts
Check anatomy, objects, numbers, maps, labels, interfaces, and every detail that could be understood as a factual claim.
Add context
Write the caption, credit, and text alternative according to what the image contributes where it appears.
Test the export
Open the actual EPUB or PDF, resize it, zoom it, and inspect reading order, cropping, contrast, and legibility.
Every visual should pass this eight-point check.
- Purpose: the image does a job that the surrounding prose does not already do better.
- Accuracy: factual, anatomical, typographic, and spatial details were inspected by a person.
- Continuity: recurring subjects, visual language, color, and labels remain stable.
- Rights: source, generation tool, relevant terms, permissions, and edits are recorded.
- Privacy: screenshots and photos contain no unintended personal or confidential information.
- Accessibility: meaning is available in context without relying on sight or color alone.
- Small-screen behavior: labels and important details remain usable on a phone-sized viewport.
- Export behavior: aspect ratio, position, file size, caption, and reading order survive the final format.
For purpose-based text alternatives and complex-image guidance, use the W3C Web Accessibility Initiative image tutorials. For a deeper production workflow, see images in ebooks.
AI ebook images FAQ
Can AI Ebook Generator add images to a book today?
Not in the current core text studio. It handles the book brief, outline, chapter drafts, editing, autosave, and a reading preview. This page is a production-planning guide for authors who will create or place the cover and interior images in a separate visual or formatting tool.
How many images should an ebook contain?
There is no useful universal number. Add an image when it explains, demonstrates, or sets essential context better than prose. A practical guide may need screenshots and diagrams; a short argument may need none.
Should text be placed inside an ebook image?
Avoid putting essential instructions or long passages inside an image. The text may become unreadable on small screens and unavailable to assistive technology. Keep necessary information in live book text and use the image as support.
Do AI-generated images need alt text?
The need depends on purpose, not how the image was made. Informative images need a concise equivalent in context; decorative images should not create noise. Complex diagrams often need a nearby explanation as well as a short text alternative.
Can I use any generated image commercially?
Do not assume that generation alone clears every right. Review the image provider's current terms, your publishing platform's rules, trademarks and recognizable people, and any reference material used. Keep a record of the tool, prompt, date, edits, and license terms.