Hello Everyone, For those who aren’t aware, there is a pending PR for a new open source JPEG 2000 driver, based on the Grok JPEG 2000 library
https://github.com/OSGeo/gdal/pull/3449 Notable Library Features: 1. support for reading TLM and PLT markers, for fast random access into large tiled or single-tiled images 2. support for memory mapped files, which helps speed up random access decode for large images 3. internal tile cache that allows efficient reading of multiple tiles from single Grok codec object 4. support for new Part 15 of the JPEG 2000 standard (HTJ2K) which promises up to **10x** speedup of encode and decode Benchmarks: I haven't done a thorough comparison between Kakadu, Grok and OpenjPEG, but here are a few results (more details in PR) 1. Compression: multi-tile Kakadu is 2X faster than Grok; Grok is 5X faster than OpenJPEG 2. Decompression: large single tile image with PLT markers Kakadu is 60% faster than Grok; Grok is 2.5x faster than OpenJPEG 3. Decompression: 4000x4000 sub-region of large single tile image with PLT markers Kakadu is 40% faster than Grok; Grok is 5X faster than OpenJPEG This year should see Grok reaching performance parity with Kakadu. There is an RFC for the new driver, which is complete, and tested on a few compression and decompression work flows. Any comments or feedback is very welcome. Cheers, Aaron
_______________________________________________ gdal-dev mailing list gdal-dev@lists.osgeo.org https://lists.osgeo.org/mailman/listinfo/gdal-dev