[general]
name=GeoKlein Raster Optimiser
qgisMinimumVersion=3.34
# Do not remove: with no qgisMaximumVersion set, QGIS infers <major>.99 from
# qgisMinimumVersion, so 3.34 alone becomes an effective cap of "3.34 - 3.99"
# and QGIS 4.2 reports this plugin as incompatible. Confirmed on a clean 4.2
# profile install. Raise this only when a real incompatibility with a newer
# QGIS major version is found, not preemptively.
qgisMaximumVersion=4.99
# Deliberately near-identical to the GitHub repository description, not
# identical: that one adds "A QGIS Processing plugin." as a second
# sentence, which is redundant here since the Plugin Manager listing is
# already a QGIS plugin context. Keep the first sentence in sync with
# GitHub; the trailing sentence is GitHub-only on purpose.
description=Converts slow, unoptimised orthomosaics, elevation and multispectral rasters into fast-panning Cloud Optimized GeoTIFFs (COGs).
version=1.1.2
changelog=1.1.2 - Corrected the help panel's description of the
    Float32 option to match how 1.1 works. No change to processing.
    1.1.1 - Changed one error-handling pattern flagged by the
    QGIS plugin repository's security scan. No change in behaviour.
    1.1.0 - Adds Float64 elevation support. A Float64 DSM, DTM
    or CHM is now detected as elevation and written losslessly at
    Float64, with no data type change - previously these were refused
    with "Data type Float64 isn't supported in v1", which rejected the
    default output of PDAL's writers.gdal and so most LiDAR DSMs. The
    same lossless ZSTD settings and floating-point predictor as Float32
    are used, confirmed on a 16000x16000 LiDAR DSM whose output band
    statistics match the source exactly. A new log line reports, for
    information only and with no change to the data, that the source is
    Float64 and what storing it as Float32 would save and cost in
    rounding for that specific file. Also fixes two reporting bugs
    found alongside it: running core/converter.py from the command line
    without --quiet crashed with a TypeError on the log line after a
    successful conversion, and detection warnings about a raster whose
    bands are not all the same data type were never shown in the
    Processing log at all. Adds an opt-in "Store Float64 elevation as
    Float32" checkbox under Advanced parameters, off by default: ticked,
    a single-band Float64 DSM, DTM or CHM is stored as Float32 instead,
    which measured about a third of the finished file size on a
    16000x16000 LiDAR DSM, for a height change the run states exactly
    and then verifies against every valid pixel of the output. The file
    is kept at Float64, with the reason logged, if the rounding would
    exceed a millimetre or if its NoData value would not survive the
    conversion, and the box is ignored with an explanation on anything
    that is not single-band Float64 elevation. With the box clear -
    the default - every value is still preserved exactly, and the
    output is byte-for-byte what it would have been with no such option
    in the build.
    1.0.1 - Adds the plugins.qgis.org listing URL to the file
    metadata this tool writes, alongside the existing GitHub URL. Also
    fixes two static-analysis findings: a deprecated Qt enum access and
    a missing import in a type annotation. No other behaviour change.
    1.0.0 - Initial release.
author=GeoKlein Ltd
email=mail@geoklein.com
about=Large orthomosaics and elevation rasters from drone photogrammetry or LiDAR are technically correct but practically unusable: no tiling, no pyramids, often no compression, so QGIS or QField crawl trying to pan around them. GeoKlein Raster Optimiser detects what a raster actually is and applies the correct tiling, pyramid and compression settings automatically, with no GDAL flags to memorise: 8-bit RGB imagery gets a real choice between lossless and lossy, and everything else it trusts (Float32 and Float64 elevation, 16-bit imagery, multispectral) is processed losslessly. Float64 is kept at Float64, and the log notes what re-exporting as Float32 would save and what it would cost. It refuses cleanly, with a clear reason, on classified/categorical rasters rather than risk silently corrupting them. A file that is already a valid Cloud Optimized GeoTIFF at the target compression is left as it is; one that is only part-way there, tiled and pyramided but not a COG, is rebuilt.
icon=icons/icon_64.png
tracker=https://github.com/GeoKlein-Ltd/raster-optimiser/issues
repository=https://github.com/GeoKlein-Ltd/raster-optimiser
hasProcessingProvider=yes
tags=raster,orthomosaic,ortho,optimise,optimize,optimiser,optimizer,slow,slow raster,large,large file,reduce file size,shrink,pyramids,overviews,compress,compression,tiling,tiled,tiles,geotiff,cog,cloud optimised geotiff,cloud optimized geotiff,jpeg,zstd,performance,speed,speed up,basemap,drone,photogrammetry,lidar,dsm,dtm,chm,elevation,multispectral,16-bit,qfield
category=Raster
experimental=False
deprecated=False
homepage=https://github.com/GeoKlein-Ltd/raster-optimiser
