
A PDF caption can look right while its characters cannot be extracted correctly. Keep visual rendering and text extraction as separate checks. Adobe explains that Acrobat needs enough font information to map characters to Unicode for reading and text-export purposes. If that mapping fails, increasing image resolution or sharpening the page does not supply the missing character information.
Key Takeaways
- Correct-looking letters do not prove that copying or reading them will work.
- Real PDF text, text converted into shapes and a photograph of text require different diagnoses.
- Repair the supported source workflow where possible, then inspect the actual new export.
Compare the same short caption
Choose a known passage containing the characters that fail, such as an approved name with an accent. Keep its correct wording from the source document. Select the passage in the delivery PDF, copy it into a plain text destination and compare the actual characters. Record the PDF and receiving application rather than using a screenshot as evidence of extracted text.
If the lettering is visibly wrong too, use the font-substitution and glyph guide. That task starts with changed letter shapes or missing boxes. Here the useful symptom is that a visually acceptable caption becomes different text when extracted.
Do not assume every failed copy operation proves a font-mapping defect. Establish whether the words are real text and whether document permissions interfere with the attempted operation. A page containing only an image of lettering has no ordinary text characters to select in the first place.
Identify the relevant source structure
Adobe's PDF accessibility explanation distinguishes searchable text from scanned images and explains font-to-Unicode extraction. A visual proof alone cannot distinguish every underlying structure. Inspect the authoring file and export settings with the person who prepared them.
| Observation | Question for the author | Evidence to compare |
|---|---|---|
| Letters look right but paste incorrectly | Can the exported font map to the intended characters? | Known wording and extracted wording |
| Words are part of a scanned page | Is searchable text present? | Image content and any recognized text |
| Words were converted into shapes | Is an approved live-text source available? | Source structure and delivery requirement |
| Only a protected copy refuses extraction | Are permissions affecting the operation? | Authorized security settings and recipient route |
These are diagnostic branches, not confirmed causes in a tested document. Preserve the original before preparing a candidate.
Recreate a supported text export
Adobe's character-encoding guidance says some issues are not repairable inside Acrobat. It lists using an appropriate different font, preferably OpenType, in the original document and recreating the PDF among its options. A font substitution still needs visual proofreading; it can change spacing and approved typography.
For a scan, use an appropriate OCR workflow and proof the recognized wording. OCR output is not evidence that an unclear historical name has been recovered correctly. Keep uncertain text identified rather than accepting a plausible replacement automatically.
Avoid flattening all words into pixels as a cure for an extraction requirement. It can preserve appearance while abandoning the searchable text the recipient needs. Preserve supported tagged export and review heading roles separately.
Accept both appearance and text
Reopen the exact new delivery PDF. Compare the caption visually, repeat the extraction and check the intended reading or search operation. Record which characters and software were checked. Accept only when both the approved appearance and the required text behavior are supported by that saved file; one successful caption does not certify the entire document.
Related tasks
- Why Does My Print PDF Show the Wrong Font or Empty Character Boxes?
- Why Did My Word or PowerPoint PDF Lose Alt Text and Document Structure?
- Why Does My Large InDesign Heading Export as an Ordinary Paragraph in an Accessible PDF?
Sources and scope
- Adobe Acrobat: Accessibility features in PDFs: Fonts must support extraction to Unicode for text-reading uses; correct visual display does not establish that mapping. Image-only text requires recognition before searchable text is available.
- Adobe Acrobat: Create and verify PDF accessibility: Some character-encoding problems cannot be repaired inside Acrobat; documented options include suitable fonts in the source and regenerating the PDF. OCR is a separate route for image-only pages.
Based on supplied complete Adobe accessibility and character-encoding guidance, reviewed October 29, 2026. No file was repaired or text extracted for this article. The caption comparison is a proposed diagnostic procedure; no universal font repair or OCR accuracy is promised.
Prepared by Imagild Editorial, the company editorial team responsible for these guides. AI assisted research and drafting. This is documentation-based guidance with proposed examples, not a claim of hands-on testing or an independent product review. Report a correction.
Cover: AI-generated conceptual editorial illustration; not a real application screenshot, observed output or before-and-after test.