An illustration can look excellent on its own and feel out of place on the next page. A picture book needs characters readers recognize, a rhythm that fits the story and enough room for the words and binding. Work out how the pages will come together before settling on an image tool.
Start with Dreamina when the book has an original character and you need to see how they fit into several illustrated spreads. Choose Leonardo if you want another way to guide a difficult pose from a visual reference. Use Midjourney to decide how the book should look, ChatGPT to walk through a narrow image correction, or Firefly when the next edit will happen in an Adobe app. Once the illustrations are approved, Canva can hold a short editable dummy while you check how the pages read together.
- Choose a tool for the part holding up the book
- Dreamina: Develop illustrated spreads around an approved main character
- Midjourney: Find the palette and visual direction before producing the book
- ChatGPT: Talk through a narrow revision to one illustrated page
- Leonardo: Guide a difficult pose or layout with a second reference
- Adobe Firefly: Set up artwork for the next step in an Adobe editor
- Recraft: Build vector decorations you can reuse throughout the book
- Canva: Make an editable dummy from approved art
- Start with a small book dummy
- Generate representative pages in Dreamina
- Fix details in an editor before committing to the layout
- Keep text editable and check the physical spread
- Frequently asked questions
Choose a tool for the part holding up the book
Dreamina: Develop illustrated spreads around an approved main character
Establish the visual reference. In Dreamina, start with a single character image and a distinct list of details. For a fox who has lost a key, keep the red scarf, soot-colored tail tip and oversized brass button. Make a wide page with the fox under a window and a close-up with the fox looking under a chair. Compare the fixed details between the two before adding more scenes.
Build in room for typography. Ask for a quiet area when you brief each illustration. Put the words into the dummy as editable text. If a line crosses a busy branch or floor pattern, first try adjusting the page layout instead of remaking the illustration.
Credits for trial spreads: Daily free credits vary by region. Free accounts in the US receive 120 per day. The regular US web renewal for Basic is US$15 monthly with 1,575 monthly credits, before offers. Use the cost shown for the selected model to budget your retries. Dreamina pricing
Layout or closer reference control: Use Leonardo if you need to break a difficult illustration into separate style or composition references. Move approved pages into Canva for a quick book dummy.
Midjourney: Find the palette and visual direction before producing the book
Compare three views of the same moment. Try the scene with a cut-paper style, a restrained watercolor palette and solid ink silhouettes. Keep the character and page proportions steady so you are comparing the art direction. Use the strongest sample to set a line style, palette and level of texture for the rest of the book.
Set a reference you can keep using. Check the reference feature on your selected model before building all the pages around it. Keep one final reference and record which elements you want to repeat, such as rough paper grain or soft blue shadows.
Subscription and private generation: Basic is US$10 paid monthly. Midjourney lists Stealth Mode on Pro and Mega. Midjourney plan comparison
Reusing the chosen look: Use Dreamina when the look is settled and the next job is making a few distinct page compositions from approved visual references.
ChatGPT: Talk through a narrow revision to one illustrated page
Give the change a clear boundary. Upload a page you have permission to use and identify one change, such as shifting a lamp away from the left margin without altering the fox’s outfit. Compare the revision with the previous image at the same size. If the unaffected details shift, return to the accepted art and isolate the repair in Photoshop.
Keep a record of decisions. Save the illustration you approved, along with the accepted scene brief and the list of things that must stay fixed. A useful chat history can inform the next page, while the reference file gives you something concrete to check.
Chat image limits: Free accounts have limited image creation and uploads. ChatGPT Plus includes expanded image access for US$20 monthly. ChatGPT Plus
Changing only one area: Use Leonardo for a composition guided by a particular image, or Photoshop for a small repair that must leave the rest of the illustration in place.
Leonardo: Guide a difficult pose or layout with a second reference
Give each reference a purpose. Use your approved character image for the character and an original sketch for the gesture or arrangement. Tell the model exactly what to take from each. For example, ask for the fox’s silhouette and outfit from one image and the angle of the crouched pose from the other. The supported choices vary by model.
Compare one candidate against a known-good page. Check how big the character’s head and paws are, whether the scarf is still recognizable and how the character fits around the space for type. Leonardo image guidance
Tokens and private creations: The free plan lists 150 fast tokens per day, automatic model selection and public creations. Essential is listed at US$12 per month before tax and shows private generations. Leonardo plans
A new visual direction or manual repair: Use Dreamina when you want to go back to an original concept and try a different camera and setting. Use a drawing editor to fix a small detail on a finished page.
Adobe Firefly: Set up artwork for the next step in an Adobe editor
Choose the reference that fits. Use a style reference for a palette or finish. Use a composition reference if the picture needs to follow a sketch prepared for the text. Open the approved result in your art editor to adjust edges and place important elements on their own layers. Adobe Firefly image workflow For the same illustrated spread, Firefly is useful when the author has a page sketch for the composition and a separate sample for the finish. Switch to Leonardo if the difficult part is assigning the approved fox to character guidance while a different reference directs its pose.
Firefly credits: Firefly offers limited free generations. The US Standard plan lists US$9.99 per month. Check the credit cost for the model you use. Firefly plans
Vector ornaments and page layout: Use Recraft for a recurring ornament that must have clean editable paths. Use Canva if the illustrations are complete and the main job is assembling a short draft layout.
Recraft: Build vector decorations you can reuse throughout the book
Keep ornaments in a vector file. Generate a simple chapter emblem or a page-corner motif in a native vector mode. Open the SVG and select its main curves to check that each part is a path. Shrink a test to the smallest size it will appear on the page to see whether any lines need to be simplified.
Free SVG and commercial tiers: A compatible model can generate SVGs on Free. Free generations are public and lack a commercial license. Generation is metered against credits. For an illustration going into a book for sale, use a paid plan at generation and check the terms for your inputs. Recraft plans
Painterly page illustration: Use Firefly or Dreamina when your page illustration depends on textured paint, rather than shapes you expect to resize and recolor independently.
Canva: Make an editable dummy from approved art
Place each spread in the sequence readers will see. Import the art at the correct aspect ratio and put text in separate boxes. Print a few test pages, fold them in order and see whether the page turn lands before or after the moment you want to reveal. Confirm the printer’s trim and bleed before you commit to the layout.
Free layout and paid assets: Canva Free lets you place and arrange your own artwork free of charge. For premium assets and tools, Canva Pro lists a regular US price of US$18 monthly before introductory offers. Check the license for any premium graphics you add. Canva plans
Prepress or editable ornaments: Use Recraft for page ornaments that need editable paths. If the printer only wants the press PDF, use the layout application that can produce its specified trim, bleed and color settings. Ask separately which editable project the publisher needs for the next edition.
For a cover concept. Ideogram can help explore lettering as part of a cover’s visual style. Test a single title treatment and proof the spelling before redrawing or typesetting it. Keep the final title, byline and spine text editable in your layout.
When you want the story and its pages in one project. Neolemon combines scene creation with a page canvas and PDF export. Take a short passage with two characters through one page turn before moving the full manuscript. Can you replace the second illustration after a reader spots a costume change? Does a longer line of dialogue still fit when you print the spread at its final size? A usable proof makes the book-oriented workflow attractive for a short picture book. If the text needs tighter control or your printer asks for settings the PDF export cannot supply, keep the approved pictures and finish the book in your layout application.
Start with a small book dummy
Give every spread one job. Break the story into page turns. Write a line on what changes at each turn, such as “the child notices the small door” or “the fox realizes it lost the key.” Choose which moments need a wide spread and which benefit from a quiet page with more white space.
Save a reference sheet before making more pages. Show the character from the front and another useful angle. Name a few stable details, like an orange scarf, the shape of the ears and the number of pockets. Keep the visual style separate. The same character may appear at noon, after dark and in a close-up while keeping those fixed traits. Try the sheet against three moments from the story: the hero seen from behind, a close-up with a surprised expression, and a wide view beside another character. The scarf may disappear behind a shoulder in one view; write down the design rule instead of requiring the same number of visible stripes in every picture. Once the interior page count is fixed, build the eBook front and paperback wrap from the book’s final dimensions.
Generate representative pages in Dreamina
- 1
In Dreamina, choose a supported reference-led image workflow. Place the approved character art and style reference where appropriate.
2Write the drawing brief around one moment. Specify foreground, camera distance, action and the quiet area where text will sit.
3Approve one quiet page, one wide action page and one difficult close-up. Compare poses and props back to the original.
4Carry the approved setup into the next group. Keep the page plan and any detail corrections with the files. Dreamina picture book illustration guide
Fix details in an editor before committing to the layout
Make the smallest necessary edit. If the character has the right pose but the wrong sleeve, fix the sleeve on a separate layer or reuse an approved piece of artwork. Avoid redoing the whole spread when changing a narrow area would preserve the stronger parts. Use Photoshop for masked edits or Clip Studio Paint when redrawing lines and hand-coloring them suits your process.
Show the hard parts together. Line up the widest view and closest view. Look for changes in the main proportions, clothing and recurring objects. Keep a simple board of approved images, then check against it each time a scene changes lighting or camera angle.
Keep text editable and check the physical spread
Choose the layout before you flatten the image. With the same approved illustrations and print specifications, choose InDesign when the production editor needs to reopen an InDesign book project and relink its pictures. Choose Affinity Publisher when the publisher already maintains the series template there and will revise the story in that application. A press PDF lets the printer output either book; it does not give the next editor the native layout. A short layout in Canva can work when your printer's requirements can be reproduced there. Put story text in editable boxes and keep important faces and type away from the gutter.
Check the book at its final size. Apply the current printer template, including trim and bleed. Look at the spread order, type size and details that fall into the gutter. Order or print a proof if the book will be read on paper. Fix the source layout before regenerating artwork just to solve a cropping problem.
Frequently asked questions
Should I use one AI tool for the whole book?
Keep a consistent image and reference workflow when it works. Use a layout tool for editable text and printing, and an image editor for small repairs that need direct control.
Can I put the words inside the illustration?
Keep story text separate for accurate spelling, accessibility and revisions. A decorative element can sit inside the illustration if you can proof every letter and it will remain fixed.
How many pages should I approve before expanding?
Choose a few with different jobs, such as a close-up, a wide spread and a page with more text. Together they reveal whether the design and layout work across the story.
Does the cover follow the same layout rules?
The cover has its own requirements, especially for print editions with a spine and back. Prepare the interior and cover according to the printer's separate specifications.
What should I check before selling a book with generated illustrations?
Check the tool’s commercial terms for your plan at the time you generate the art. Save your own input files, receipts and the original editable layout. Confirm the rights to any reference images, typefaces and stock elements you include. Then review the publisher’s current submission rules before sending the files. Amazon KDP asks you to disclose AI-generated illustrations, including interior and cover artwork, when publishing a new book or republishing an edited one. It treats generated artwork as AI-generated even after substantial human edits. AI-assisted edits to art you created do not require KDP disclosure. Amazon KDP content guidelines