Use Case: Save-as-Copy Burn-in

Generate a baked PDF with annotations burned into the page contents — interoperable with any PDF reader.

When you need to deliver a PDF that anyone can open and see the annotations (not just users of DocumentViewer), use saveAsCopy(). It returns a fresh PDF binary with annotations baked into the page contents.

Pattern

const buffer = await viewer.value?.saveAsCopy()
const blob = new Blob([buffer], { type: 'application/pdf' })

// Option A: trigger a download
const url = URL.createObjectURL(blob)
const a = document.createElement('a')
a.href = url
a.download = 'reviewed-doc.pdf'
a.click()
URL.revokeObjectURL(url)

// Option B: POST to your backend for archival
await fetch('/api/documents/reviewed', {
  method: 'POST',
  body: blob,
  headers: { 'Content-Type': 'application/pdf' },
})

When to use this vs. `exportAnnotations`

Need	Use
Show the annotated doc to someone who doesn’t have `DocumentViewer`	`saveAsCopy()`
Backup the annotations so the user can later edit them in `DocumentViewer`	`exportAnnotations()` (preserves editability)
Both (e.g. archive + future re-edit)	Do both — store the JSON for editing, ship the PDF for sharing

saveAsCopy() burns annotations into the PDF contents in a way that’s spec-compliant for PDF 1.7 — most professional PDF readers (Adobe, Preview, Foxit) will show them. But the original “editable annotation” metadata may be lost, so subsequent rounds of editing should start from the original source + the exportAnnotations() JSON.

Compliance burn-in (Phase 2: redactions)

When features.redaction ships in Phase 2, saveAsCopy({ applyRedactions: true }) permanently removes the redacted content from the PDF (not just covers it with a black box). This is the correct compliance workflow:

// Phase 2 — once redaction lands
const buffer = await viewer.value?.saveAsCopy({ applyRedactions: true })
// the buffer has the redacted text completely removed from the page contents,
// not just hidden behind a rectangle.

Threads are not embedded

Threads (replies + resolved state) live in the parallel CommentThread[] overlay, not in the PDF binary. If you need to ship both the baked PDF and the thread data:

const [buffer, threads] = await Promise.all([
  viewer.value!.saveAsCopy(),
  fetchAllThreadsFromAppState(),
])

// Bundle them in a multipart upload, or a zip, or two separate endpoints

Performance

saveAsCopy() re-encodes the PDF — for very large documents (1000+ pages with many annotations), this can take several seconds. Show a loading state during the call.