Home
WIMF is an experimental image codec with a Python frontend and a portable C++17 backend. New files use the WIM2 container and the canonical .wimf extension.
WIM2 divides an image into independently decodable tiles and chooses Raw, Predictive, Palette, or Wavelet coding per tile. This supports exact lossless output, lossy compression, region-of-interest decoding, high bit depth, localized corruption detection, optional anti-rot recovery, and indexed history states.
- Installation — install wheels and verify native acceleration.
- Python-API — encode, decode, inspect, ROI, metadata, and history.
- Command-line tools — headless workflows and diagnostics.
- WIMF-Studio — desktop encoder, inspector, and corruption lab.
- WIM2-Format — container and tile-mode overview.
- Conformance-Vectors — portable decoder expectations and hashes.
- Native-Integration — embed the codec without Python.
- Corruption-and-Recovery — strict decoding, diagnostic previews, and anti-rot.
-
Legacy-Migration — move WIMF v1,
.wif, AWIF, andROT!content to WIM2. - Roadmap — interoperability and adoption work.
WIM2 is the only recommended authoring format. WIMF 2.2 warns when deprecated legacy writers are used. WIMF 3.0 will remove those writers while retaining read-only legacy compatibility.
WIMF is experimental. Please report reproducible bugs at https://github.com/benchware/WorstImageFormat/issues with your OS, Python version, wimf runtime --json output, and a minimal sample when possible.
WIMF compresses data; it does not encrypt it. Never store secrets in metadata.