Translate the
document.
Keep the design.
Upload an InDesign IDML or PDF. Get it back in any of 40+ languages with every font, column, table and page break exactly where you left it.
Source vs. output, identical layout
.idml → .zh.idmlIdentical layout · same page count · same styles · same master pages
We read the structure, not the pixels.
InDesign stores documents as a tree of frames, styles and links. Pagebirdy parses that tree directly — not a flattened export. Every text frame, master page and style sheet survives intact, with its properties, position and relationships preserved.
Every node read, translated, rebuilt in place
Under the hood.
Six stages. Every one automated.
Full page mirror, not just text direction.
Right to left. First class.
Arabic and Hebrew get the full treatment — margins flip, gutters swap, bullets point the right way, and the whole page mirrors inside the original frame. Not just text direction. The whole layout.
No Japanese in your typeface? We find one that fits.
When translated text requires a character set your document's font doesn't cover, Pagebirdy substitutes from its bundled Noto library — matching cap height, x-height and baseline so column widths and line counts stay identical. No manual font work after download.
How it works
Upload your file
Drop an IDML or PDF. We read the full layout tree — text frames, linked graphics, master pages — not a flattened text dump.
Pick language + glossary
Choose from 40+ targets. Import a TBX or CSV to lock terms — product names, legal phrases, brand terms.
Download the result
Same file extension, same styles, same page count. Open it in InDesign or Acrobat and keep editing.
Everything the file carried, carried across.
08 capabilitiesLayout preservation
Every font, column, table, footnote and page break lands exactly where it was. No manual cleanup, no reflowed text, no missing elements.
RTL mirroring
Arabic and Hebrew get the full treatment — margins flip, gutters swap, bullets point the right way, and the whole page mirrors inside the original frame.
OCR for scanned PDFs
Scanned documents get a full OCR pass. The translated text layer rebuilds in position over the original scan — searchable, copyable, exact.
Font substitution
No Japanese cut in your typeface? We pick a metrically compatible face from our bundled Noto library — not a generic fallback that breaks your line lengths.
Multi-engine consensus
OpenAI and DeepL run in parallel. When they disagree beyond a threshold, the segment is flagged for human review — before it ships, not after.
QA scoring
Every job gets a layout-integrity score: reflow rate, overflow count, font coverage. Catch problems in the report before the client does.
Side-by-side review
Reviewers edit the translation against the source page, paragraph by paragraph. Approved edits feed directly back into your glossary.
40+ languages
From Afrikaans to Vietnamese — including Arabic, Hebrew, Japanese, Chinese, Korean, Thai, and every major European language.