Question 1

How does layout-preserving PDF translation work?

Accepted Answer

Academic Translate uses deep learning computer vision (MinerU Layout Parser) to isolate text bounding boxes while keeping mathematical equations ($...$), vector diagrams, and two-column structures in place, re-injecting translated text at the exact original coordinates.

Question 2

Why do Google Translate and DeepL break academic PDF layouts?

Accepted Answer

General machine translation engines treat PDFs as unstructured text streams, causing two-column papers to merge into single columns and corrupting mathematical notations into unreadable text. Academic Translate is built specifically to prevent layout collapse.

Question 3

Can I export the translated document to Overleaf / LaTeX .tex?

Accepted Answer

Yes. Every translation produces a complete Overleaf .zip bundle containing ready-to-compile .tex source files and all cropped figures.