Separate main text, notes, and names
Capture scale
Clear regional crops that keep character groups, readings, and nearby labels visible.
Region relationship
Character context helps separate names, meanings, and sound-based readings.
Variant check
Preserve official names and verify rare characters with a knowledgeable reader.
Dense characters need close evidence
Capture dense text closely and keep annotations, names, and script conversion distinct from translation.
Give dense characters enough detail
Move closer until internal strokes remain separate at full image size. Photograph small packaging text one panel at a time and avoid strong JPEG compression. A distant high-resolution scene can still leave each character too small.
Use even focus across the line. A shallow focus plane on a curved package may make the middle sharp and both ends unreadable, so rotate the object and make overlapping captures.
Separate main text from annotation
Japanese material may mix kanji, hiragana, katakana, Latin letters, and small pronunciation guides. Chinese pages may contain vertical captions or traditional and simplified forms. Korean text may mix Hangul with numbers and Latin product names.
Crop annotations with the word they explain but label them as annotations in the extracted result. Otherwise a reading aid can be inserted as if it were a second word in the sentence.
When annotations use more than one direction or script, segment them with the mixed-script reading-order workflow.
Check names and character variants
Personal and place names often have readings that context alone cannot guarantee. Preserve the source characters and compare with an official website, ticket, card, or directory when an exact Latin form matters.
Do not treat script conversion as translation. Converting traditional and simplified Chinese forms, or romanizing Japanese or Korean, changes representation but may not explain meaning. State which output the reader needs.
Improve image translation for Chinese, Japanese, and Korean text, including dense characters, vertical writing, annotations, and mixed scripts. Continue with the languages and scripts guide collection for adjacent workflows.