Hi HN! I built DCP after struggling with PDF corruption in a document workflow.
Instead of editing PDF internals (which breaks things), DCP treats the PDF as an immutable substrate and applies JSON-defined visual deltas on top.
Think of it as "Git for document surfaces" – you get:
- Deterministic rendering (same input = same pixels)
- Full audit trail (delta.json is human-readable)
- Zero risk of corrupting the original
Would love feedback on the protocol design (SPEC.md).
I built this as an experiment after realizing that “editing PDFs” is fundamentally broken.
Instead of modifying PDF internals, DCP treats PDFs as immutable surfaces and applies deterministic visual deltas (whiteout + overlay) on top. The original PDF is never modified. All edits live in a JSON delta ledger.
Think: Git for 2D surfaces.
Hi HN! I built DCP after struggling with PDF corruption in a document workflow.
Instead of editing PDF internals (which breaks things), DCP treats the PDF as an immutable substrate and applies JSON-defined visual deltas on top.
Think of it as "Git for document surfaces" – you get: - Deterministic rendering (same input = same pixels) - Full audit trail (delta.json is human-readable) - Zero risk of corrupting the original
Would love feedback on the protocol design (SPEC.md).
I built this as an experiment after realizing that “editing PDFs” is fundamentally broken. Instead of modifying PDF internals, DCP treats PDFs as immutable surfaces and applies deterministic visual deltas (whiteout + overlay) on top. The original PDF is never modified. All edits live in a JSON delta ledger. Think: Git for 2D surfaces.
Editing PDFs is not broken.
Just because you don't understand something complex, does not mean it is broken.