Elysia Tools
Navigation
Workflow Playbook
Rescue a damaged PDF, strip what should not travel with it, and cut its weight — repair broken headers and cross-reference tables with diagnostics, export annotations before erasing them, remove metadata and form fields, then downsample and recompress to hit your size target.
Hubs
A PDF that will not open has usually not lost its content — it has lost its map. Broken headers and missing cross-reference tables account for most dead files, and the standard repair rebuilds them quickly. When that fails, the diagnostic repair earns its keep — detailed findings, a before-and-after comparison, and a log that says what was recovered and what was not, which turns a rescue from folklore into a record. One rule guards the whole step — back the damaged original up first. Repair is a write operation, the damaged file is the only source of everything the repair might recover, and a repair attempted directly on the only copy is a rescue that risks becoming a burial.
The cleanup that gives this document its hygiene is a deletion pass, and deletions deserve a receipt. Annotations hold the review's conclusions — highlights marking what mattered, comments recording the questions, stamps declaring the verdicts — and once cleanup runs, they are gone without appeal. The export takes seconds, lands the annotations in a file of their own, and costs nothing; skipping it costs the review's memory. The habit generalizes — before any destructive pass in this workflow, ask what the pass deletes and whether anyone will want it back.
What travels inside a PDF goes beyond its pages. Metadata names authors, tools, and dates; annotations carry the review's inner monologue; bookmarks and form fields expose structure that was never meant for the recipient. Cleanup strips this furniture from the document so the deliverable carries content and nothing else — and the acceptance check is blunt — the properties panel must come up empty, and the annotation panel with it. The boundary with redaction matters here — cleanup removes the document's attachments, while removing words and images from the page itself is redaction, a content-level operation that belongs to the signing and distribution workflow.
Slimming has one irreversible move in it, and everything else is patience. Stream recompression and redundant-object cleanup shrink bytes at almost no visible cost. Image downsampling is the real trade — resolution spent is resolution gone, so it belongs in screen copies where 150 dots per inch reads perfectly, and stays out of print masters and archival copies that may be enlarged. Let the destination set the target — an upload limit is a number, and the diet is finished when the number is met, not when the file is as small as vanity can make it. When the first optimizer pass misses the target, the advanced pass adds structure cleanup and stronger recompression before any thought of deeper downsampling.
This page heals, cleans, and slims an existing document — it does not change what the document says or how its pages are shaped. Restructuring pages — rotating, cropping, deleting, imposing — belongs to the page assembly workflow. Sealing, stamping, and tracing the result belongs to the signing workflow, and notice that its order still rules — cleanup and slimming alter every byte, so they finish before any signature is applied. Long-term archival to PDF/A with accessibility checks belongs to the archival workflow, which is also where damaged scans get straightened and denoised. Send each job to its own desk, and this one stays what it is — the clinic a tired file visits before it travels.
Workflow playbook
Attempt the standard repair of damaged headers and broken or missing cross-reference tables, and escalate to the diagnostic repair with before-and-after comparison and logging when the file still refuses to cooperate.
Pull highlights, comments, stamps, and markup out of the document while they still exist, so review conclusions survive the cleanup that follows.
Remove metadata, annotations, bookmarks, and form fields so the deliverable carries its content and nothing that identifies its authors, its internal systems, or its review history.
Downsample images, clean redundant objects, and recompress streams with the smart optimizer, and turn to the advanced pass — structure cleanup plus recompression — when the first round misses the size target.
Run the standard repair first — it fixes broken headers and missing cross-reference tables quickly; escalate to the diagnostic repair only when the file still fails, because it brings detailed diagnostics, before-and-after comparison, and a repair log that documents exactly what was recovered and what was not.
Removing metadata alone protects privacy while leaving the document's working parts intact; erasing annotations, bookmarks, and form fields too produces the bare deliverable — decide by audience, and export the annotations first whenever they carry review conclusions.
Stream recompression shrinks bytes with little visible cost, while image downsampling permanently trades resolution for size — screen copies may downsample boldly, print and archive copies should not, and the upload limit, not ambition, sets the target.