Improved compression is structure-aware exact re-entry.
This is not a generic codec claim. The next branch of T15 Orchestrator - Pedulli's lossless compression - is filetype-aware parsing, a Pedulli structure-pattern library, and an exact dual unparser. It uses SRD-derived logic together with standard codecs, then keeps the smallest verified output - so it is never larger than the best standard codec (worst case +1 byte). It wins outright on structured data, ties the best codec on already-optimal or random data, and always restores byte-for-byte.
Public verifier form
Input file Build a candidate artifact Verify byte-exact restoration Report measured size gain or passthrough diagnostic
The Orchestrator uses SRD-derived logic alongside xz, zstd and brotli alongside its SRD math transforms and keeps the smallest verified artifact, so it is never larger than the best standard codec. On structured data it wins outright; on already-optimal or random data it ties the best codec by selecting it. Either way the restore is byte-exact and the output never exceeds the input by more than one byte.
Three layers
- Filetype parser: C source, PDF, logs, database exports, MP4 metadata slices.
- Structure-pattern library: repetition, symmetry, pointer relations, table reconstruction, grammar closure.
- Dual unparser: exact return to the original bytes.
First target
C source and synthetic structured binary: controlled, testable, good for demonstrating the parser/unparser roundtrip.
Second target
PDF object graphs, xref tables, repeated dictionaries, and offsets.
Boundary
On random / high-entropy data the Orchestrator ties the best standard codec by selecting it - never larger, worst case +1 byte. No fake universal compression, no claim of passing any entropy bound.
Test it on a structured file.
Every public claim closes by benchmark or exact roundtrip artifact. Drop a file and verify the SHA-256 roundtrip yourself.