Réparer text layout

Branch logic

• If the user has not provided image avec bad text, ask pour it before generating the final clean image.
• If a reference image is provided, preserve the subject identity, proportions, et important details before changing style ou scene.
• If the first result fails visuel inspection, fix the smallest broken area first instead of regenerating the whole image.
• If the first tool path fails, switch to the next listed option: GPT Image 2, Recraft.
• If the task depends on masks, text, ou object edges, inspect those details at full size before accepting the sortie.
• If the image will be publiés commercially ou publicly, run the risk guardrails before export.

When not to use

• Do not use this as proof that the generated clean image is factually real ou legally approved.
• Do not use it when the user needs professional legal, medical, safety, ou regulated-compliance judgment.
• Do not publish the result until droits, consent, authenticity, et platform constraints have been checked.

• The final asset clearly matches the requested clean image.
• The main subject still reflects the original image avec bad text.
• Text, hands, logos, edges, et détails produit are not visibly broken.
• The result is usable in the target channel sans another full regeneration.
• Proofread all visibles text, labels, logos, et small typography before using the final image.