PDFs, scans and comics
Upload a PDF, a picture of a page or a comic archive, and choose before the upload how it comes back: as text, written onto your original pages, or on pages rebuilt in the original’s style.
On this page
Which files can I upload?
- PDFs, digital or scanned, including handwritten pages.
- Pictures of pages: PNG, JPG, WebP or TIFF, such as a photo of a letter or a scanned page. Each picture becomes one page.
- Comic archives (.cbz): each image becomes a page, in the order of the file names.
A file can be up to 300 MB; a large one uploads in parts. Transept recognises a comic or a manga from a few of its pages and reads it as one.
Text only, Same pages or Rebuilt pages: which should I pick?
When you upload a PDF or a picture, the upload dialog asks What do you need? and shows each choice’s price and time before anything is charged.
| Choice | What you get | Download as | Cost |
|---|---|---|---|
| Text only | The text, translated | Word, Markdown or PDF · Text only | No layout fee |
| Same pages | The translation written onto your original pages | PDF · Same pages | The layout fee |
| Rebuilt pages | Each page rebuilt in the original’s style, so longer text fits; Same pages comes with it | PDF · Rebuilt pages or PDF · Same pages | 1.5 times the layout fee |
Pick Same pages when the translation should sit where the original text was, such as a form, a brochure or a comic. Pick Rebuilt pages when the translation runs longer than the original and needs room, as German often does. Pick Text only when you only need the words.
You can change your mind once. A Text only document offers Same pages and Rebuilt pages in Vision until you translate it; after the first translation it stays Text only. A document on Same pages offers Switch to Rebuilt pages for the difference in price.
What does it cost?
Laying out the pages has its own fee, charged once, when AI prepares the pages after the upload. Translation is charged per source word on top, as for any document.
| Fee | Credits |
|---|---|
| Each document | 200 |
| Each digital page | 80 |
| Each scanned page | 200 |
| Each comic page | 150 |
| Each handwritten page | 700 |
| Rebuilt pages | 1.5 times the above |
| Check layout, each page, each run | 60 |
| AI redraw of text on the art, each image of up to nine texts | 800 |
A 20-page digital PDF on Same pages costs 200 + 20 × 80 = 1,800 credits to lay out. If your balance doesn’t cover the pages, the document opens with the text and a Prepare now button for when it does.
How do I fix the translation on the page?
In Vision, click a translated block on the translated page and type. The page sets your words as you go, in the original’s font and size, and a bar above the block says whether they fit:
- Fits: the text sits in the original’s space.
- Shrunk to 90% to fit: Transept made the type a little smaller to keep it in.
- Too long: about 15% doesn’t fit: shorten the text, or switch to Rebuilt pages, where the page makes room. A block that doesn’t fit also has a red outline.
Bold and Italic work on the selection (Cmd/Ctrl+B, Cmd/Ctrl+I). Press Enter or Done to save, Shift+Enter for a new line, Esc to cancel. Regenerate asks AI for a new translation of the block, with a note on what to change, and the … menu holds Translate, proofread and comment.
A text drawn into a picture has no letters you can click into, so its text opens in the bar instead, with the note “Its letters are drawn into the picture, so its text is edited here.”
How do I check the layout of a whole document?
Check layout in the Vision header compares each translated page with the original, fixes what it can and leaves a comment on the rest, such as a heading too long for its line. It costs 60 credits a page and needs a translation to compare.
The original text wasn’t all read. How do I add what’s missing?
Open Text areas, then drag on a page around the text that was missed. Click a text area to change it: drag it to move it, drag its corner to resize it, or press Delete to leave that text as it is. A text area you set is marked “Set by you: the AI leaves it alone”, so later runs keep your choice. Each text area also takes a note for the translator, such as what a label refers to.
How is the text drawn into a comic translated?
Speech balloons and captions are lettered like any page. Text drawn into the art itself, such as sound effects, signs and words with no balloon, gets a choice in the Translate dialog, under Text on the art:
| Choice | What happens | Time and cost |
|---|---|---|
| Lettered | The original is painted out and the translation lettered in its place | No extra time, no extra fee |
| AI redraw | An image model redraws each one in the original’s style, after the pages are done | About 25 seconds more; 800 credits an image of up to nine texts |
| Gloss | The original stays as drawn, with a small translation beside it | No extra time, no extra fee |
While the art is redrawn, the run’s card says “Redrawing 4 texts on the art with AI” and the pages update when they land. Check the spelling of a redrawn word: an image model can get a letter wrong.
To change one text, open Text areas and click it. Under How it’s drawn you can letter it in the balloon, on the art or in a display face, have AI redraw it, keep the original with the translation beside it, or leave it as drawn. A redrawn text you don’t like has Redraw again.
With a gloss, Where the translation sits puts the small translation on the side you pick: Auto, Below, Left, Right or Above. Auto takes the first side clear of other text. While the text area is open, you can also drag the small translation to any spot on the translated page. To move the original’s text area itself, click it in Text areas, then drag it, or drag its corner to resize it.
How do I download the translated PDF?
In Vision, click Share & export (the arrow button in the header). Under Export:
- PDF · Same pages: your original pages with the translation written in.
- PDF · Rebuilt pages: each page rebuilt in the original’s style. Shown for documents on Rebuilt pages.
- CBZ · Translated pages: a comic’s pages as a .cbz, one picture a page, at your pages’ own resolution, for comic reader apps.
- PDF · Text only: the translated text alone, simply formatted, without the pages.
Blocks you haven’t translated yet keep their original text on the page. Exporting is free. The same menu also exports the text as Word or Markdown.
A comic downloads either way, as a PDF or as a .cbz, whichever you uploaded.
Why did my PDF come back as plain text?
It was uploaded as Text only. Until you translate it, Vision offers Same pages and Rebuilt pages above the translate button. Once it’s translated it stays Text only, so upload the file again to get the pages.
Why does a scanned page cost more than a digital one?
A digital PDF carries its text, so AI only works out the layout. On a scan or a photo there is no text to read, so AI reads every word from the image first.
Can I get my comic back as a .cbz?
Yes. CBZ · Translated pages in the export menu gives one picture a page, at your pages’ own resolution.
Is there a visual proofread for PDFs?
Check layout is the PDF’s visual check: it compares each translated page with the original. For the language itself, run Smart Proofread as on any document.
Still stuck? Ask Literess in the app, or write to support@transept.ai.