Tutorial · Stage 4 of 6
Select Content in ScanTailor Advanced: Get the Content Box Right
Select Content decides what on each page is worth keeping. The rectangle you approve here becomes the basis of the finished page’s size — too tight and you amputate page numbers, too loose and every page carries scanner-bed debris into the final book.
Why This Box Controls the Rest of the Pipeline
The stage determines the rectangular region of “useful” content on the page — shown shaded in the working area. It matters for two structural reasons, both spelled out in the project documentation:
- It sets the output page size. At Page Layout, your chosen margins are added around the content box, and the outer limit of content-plus-margins is the size of the finished page. An erratic content box means erratic page sizes.
- It keeps junk out. The fold line of the spine, black scanner edges, fingers, and neighbouring-page bleed all live outside the text block. If they fall inside the box, they are treated as content worth preserving; in most output modes the margin area is filled clean, but what is inside the box is kept.
Everything downstream — margins, dewarping, binarization — treats this rectangle as ground truth. Five minutes of checking here is the highest-value review in the whole pipeline.
Automatic Detection, Then a Human Pass
Run the stage’s batch button first: the detector finds the content region on every page. Then review. On ordinary book pages the detection is reliable; the failures cluster on predictable page types — sparse pages (chapter endings, title pages), pages where a dark illustration touches the text, and pages where the page number sits far from the body text.
Adjusting the box
- Drag the edges. Hover over an edge of the box, click, and drag it to where the content actually ends.
- Double-click to auto-adjust. ScanTailor Advanced can re-fit the box to the content at the click point: double-clicking on content expands or trims the nearest edge to match it. Holding Shift restricts the adjustment to the left/right edges, Ctrl to top/bottom, and Shift+Ctrl adjusts both nearest edges at once. The Advanced documentation gives the motivating example — a page number the automatic algorithm missed can be pulled into the box with one double-click below it, instead of laboriously dragging corners.
- Create or delete a box manually. When detection finds nothing (a nearly blank page that still holds a catchword or a small ornament) or invents content that is not there, right-click the page and add or remove the content region explicitly.
- Move the whole box. Page and content areas can be dragged together with Shift+left-mouse, and a corrected box can be applied to other pages with its position automatically adjusted to each page — the rescue for a run of identically laid-out pages where detection failed the same way.
What Belongs Inside the Box
Page numbers
The most commonly lost element. They sit outside the text block, often far from it, and automatic detection regularly misses them. If your edition’s pagination matters — for citation, for a table of contents, for sanity — verify the box reaches the page number on a sample of pages, not just one.
Illustrations and plates
Pictures that belong to the page belong inside the box. If the book will be output in Mixed mode, illustrations are handled as picture zones at the Output stage so they survive as grayscale/colour while the text binarizes — but they still have to be inside the content region first.
Footnotes and marginalia
Footnote blocks separate from the body by a gap and a rule line; detection may take the body only. Printed marginal notes in old editions are content too — box them deliberately, or decide edition-wide to exclude them, but decide once and apply it consistently.
Running heads
Header lines with the chapter title are a judgement call. Keeping them preserves the original page’s character; excluding them gives cleaner reflowed output. Either is defensible — inconsistency between pages is not.
The Classic Mistakes
- Cutting into the text. A box edge through a line of text means that line is treated as outside the page’s useful area — with margins and output sizing built around the smaller box, edge characters end up cramped against the trim or clipped. When in doubt, leave air: a slightly generous box costs a few pixels of margin; a tight box costs letters.
- Boxing the spine shadow. The dark gradient near the gutter is not content. If it is inside the box, dewarping and binarization both try to honour it, and the finished page keeps a dirty inner edge.
- Trusting the first page. Detection quality varies by page type. Check at least one page of each kind in the book: dense text, sparse text, illustrated, title/verso.
- Fixing sizes here that belong to Page Layout. If every box is correct but the pages still differ in size, that is the next stage’s job — soft margins exist precisely to equalise page sizes. Don’t distort content boxes to force uniformity.
Frequently Asked Questions
The automatic box missed the page number. How do I fix it?
Double-click on or near the page number: the auto-adjust feature expands the content area to the content at the click point. Holding Shift restricts adjustment to the left/right edges and Ctrl to top/bottom. This is much faster than dragging box corners, and the corrected box can be applied to similar pages with its position auto-adjusted.
Can I create or delete a content box by hand?
Yes. If detection finds content that isn’t there, or misses a nearly blank page’s small content, right-click the page in the working area and add or remove the content region from the context menu, then drag its edges into place.
Why does the content box affect the final page size?
Because Page Layout builds the finished page as content box plus the margins you set, and pages are then equalised against each other. An inconsistent box produces inconsistent page sizes, which is why this stage — not the margins stage — is where size problems are actually fixed.