Compare AI-generated image options: evaluate at the size users will see

A worked example: where the task becomes difficult

A fictional learning site has four generated hero images. One is striking at full size but noisy behind text, one contains a confusing pseudo-logo and another looks plain yet works well on mobile. Put them into the intended layout rather than judging isolated files. A choice should reflect the visitor task and page constraints, not only the most dramatic thumbnail.

Decisions to make before implementation

Set must-pass conditions before aesthetic scoring: truthful representation, no unintended branding, readable overlays and suitable crops. Consider file weight and dimensions. Review consistency with other assets without requiring identical imagery everywhere. If a candidate needs extensive cleanup, include that effort in the decision rather than comparing only initial appeal.

A practical sequence for the work

Use the sequence below as a task boundary, not as a claim that the example has been executed. Work with approved inputs and the project’s actual architecture. If a required integration or permission is unavailable, keep that stage visibly incomplete rather than generating a plausible substitute result.

• Write the visual’s purpose and must-pass constraints.
• Place each candidate into representative desktop and phone layouts.
• Inspect artifacts and factual implications at final display size.
• Select a candidate and record why alternatives failed or need revision.

A detailed brief you can adapt for your agent

Replace the illustrative context with your approved facts and controlled inputs. Keep the stated boundaries when adapting the brief. The expected deliverable matters more than a particular tool name: ask for an explanation grounded in the inspected material and evidence for the requested outcome.

Evaluate these permitted image candidates in the supplied page layout. Use must-pass checks for truthful context, unintended branding, text space and phone crops. Compare message clarity, artifacts and optimization effort. Do not claim conversion performance from preference. Return a selected candidate with concrete revision notes.

Failure modes that an attractive preview can hide

A numeric score can disguise arbitrary taste. Define each criterion and include notes. Avoid inventing conversion results from visual preference alone. Generated images may include accidental readable or near-readable branding; inspect rather than assuming the prompt prevented it. Preserve source and usage records for assets supplied from outside the project.

Technical references: W3C WAI: accessibility tutorials

Acceptance checks and the evidence to retain

Keep the scorecard definitions, layout previews and selection rationale. A future editor should understand the functional choice, not only the preferred style.

Specific answers

Common questions