Estimate illustration work from the number of accepted assets the book needs, then add explicit revision scenarios based on a representative test. Separate concept exploration, factual corrections, layout adaptations, and final review. Use current provider information for any usage estimate, and keep human editing time visible alongside generation charges.
Count accepted assets rather than empty slots
List the illustrations the finished book requires and describe the purpose of each one. A decorative chapter image, an instructional diagram, and a cover concept involve different acceptance questions. Counting them as identical requests hides the work needed to make them usable. Record which assets can be reused legitimately and which need a distinct composition or explanation.
Distinguish an image slot from an accepted asset. A slot describes a planned location, while acceptance means the chosen image meets the relevant brief and review criteria. One generated candidate may fill a simple slot quickly; a demanding diagram may require several attempts and manual correction. The estimate should reflect that uncertainty without pretending you can predict an exact number of attempts for every image.
Test the most consequential visual demands
Choose a small pilot that represents the book’s difficult requirements. Include an image with consistent objects across a sequence, a diagram with meaningful spatial relationships, or another demanding feature actually present in the plan. A successful decorative image provides little evidence about the effort required to produce accurate instructional artwork.
Record the brief, output, correction requests, editing work, and final review result. Note why candidates were rejected or revised. The purpose is to learn which requirements cause repeated work and whether a clearer brief or different production method can address them. Do not simply average the number of generated files without understanding what happened between the first attempt and the accepted result.
Separate different kinds of revision
Group expected revisions by cause. Concept changes alter the intended composition or style. Content corrections fix an inaccurate object, sequence, or relationship. Layout adaptations adjust an accepted image for its actual placement. These categories help reveal whether the project is still exploring its visual direction or is ready to produce a defined set of assets.
Avoid treating all rejected candidates as unavoidable generation variance. If every attempt fails because the brief asks for an unclear scene, refine the brief before increasing the allocation. If the task requires precision that the tested method does not reliably provide, consider an appropriate manual production path. A credible estimate includes the method capable of producing accepted work, not only the method that creates drafts fastest.
Apply current usage information to bounded scenarios
Use the actual provider’s current charging information and the features selected for the project. Record the source and date behind the estimate. Keep image generation separate from text generation, manual retouching, and other services when they are billed or scheduled differently. Do not assume that a subscription allowance or one pilot charge establishes a fixed cost for the entire illustrated book.
Build a straightforward scenario and a scenario with the specific additional corrections observed in the pilot. Explain the assumptions instead of presenting a mysterious contingency figure. If a service quote includes only a limited revision scope, record what happens when the work exceeds it. Unknown charges or unavailable terms should remain marked unknown until you can verify them before committing to a larger run.
Review actual progress against accepted output
Track how many required assets have passed review, how many remain candidates, and which are blocked by unresolved text or layout decisions. A folder containing the planned number of images can still represent an unfinished illustration set. Compare actual effort with the estimate using accepted assets as the meaningful output, while retaining the draft history needed to understand repeated problems.
Update the remaining plan when the manuscript or visual approach changes. A revised instructional sequence can invalidate several accepted images, and a new format may require adaptation work. Preserve the earlier estimate as a dated scenario rather than overwriting its assumptions. Before release, continue with the relevant format and content checks for the actual assets used. The estimate supports controlled production; it does not establish that the resulting images are automatically suitable for publication.
Frequently asked questions
Clear answers for this publishing decision.
Should I budget exactly one generation per illustration?
Only as an explicitly limited scenario, not as a reliable expectation. Use a representative pilot to identify likely correction and adaptation work, and keep uncertainty visible for more demanding assets.
Do rejected candidates always mean the budget was too small?
No. Repeated rejection can indicate an unclear brief, an unresolved design decision, or a production method poorly suited to the task. Diagnose the cause before committing more requests.
What should count as progress in the illustration budget?
Track accepted assets and remaining review work as well as usage. Generated files are intermediate outputs; the book needs images that satisfy their actual purpose and production requirements.
How wide does the full-wrap cover need to be, and how thick is the spine?
The cover width is back cover plus spine plus front cover plus bleed on both sides; the spine width is the final even page count multiplied by the paper-specific factor Amazon publishes, so the cover cannot be sized before the interior page count is final.
Prepare the cover in a book project