Changes since 8.17.1:

• rank: fix an off-by-one error [larsmaxfield]
• popplerload, svgload: validate page size [Yang Luo]
• pdfiumload: allow both dpi and scale to be set [kleisauke]
• svgload: honor DPI when scaling elements with non-pixel units [kleisauke]
• disable redundant Highway AVX512 targets [kleisauke]
• openslideload_source: flag operation as "nocache" [kleisauke]
• remove vipsprofile.1 man page from install [kleisauke]
• tiffload: ensure processing halts for memory-related errors [lovell]
• tiffload: increase memory limit from 20MB to 50MB [lovell]
• fix vapi build [stydxm]
• tiffsave: include sample format when copying tiff data [manthey]

Windows binaries here:

https://github.com/libvips/build-win64-mxe/releases/tag/v8.17.2

Changes since test1:

• system: add "cache" argument
• add vips_image_get_tile_width(), vips_image_get_tile_height(): get tile cache geometry hints [jbaiter]

A test release to help nip4 betas.

A test release for nip4 development.

Changes since 8.17.0:

• add dcrawload, dcrawload_source, dcrawload_buffer: load raw camera files using libraw [lxsameer]
• add magickload_source: load from a source with imagemagick
• add vips__worker_exit(): enables fast threadpool shutdown
• larger mmap windows on 64-bit machines improve random access mode for many file formats
• pdfload: control region to be rendered via page_box [lovell]

Changes since 8.17.0:

• fix API docs build with meson < 0.60 [lovell]
• improve function checks in meson [kleisauke]
• tiff: use correct log domain in threadsafe warning handlers [kleisauke]
• shift 16-bit output down to 8-bit when unsupported by saver [kleisauke]
• text: prevent use of rgba subpixel anti-aliasing [lovell]
• tiffsave: always apply resolution unit conversion [kleisauke]
• cache: suppress invalidation errors in release builds [kleisauke]
• dzsave: IIIF: use named region of 'full' when no crop takes place [lovell]
• pdfload: fix potential crash with pdfium < 6633 [zhifengzhuang]

Windows binaries here:

https://github.com/libvips/build-win64-mxe/releases/tag/v8.17.1

libvips 8.17.0

Release notes here:

https://www.libvips.org/2025/06/05/What's-new-in-8.17.html

Changes since 8.16.1:

• gifsave: add keep_duplicate_frames flag [dloebl]
• add Magic Kernel support [akimon658]
• tiff: add threadsafe warning/error handlers (requires libtiff 4.5.0+) [lovell]
• tiffload: add support for fail_on flag [lovell]
• tiffload: add support for unlimited flag (requires libtiff 4.7.0+) [lovell]
• much more reliable operation caching
• colour: add support for auto-selecting the rendering intent [kleisauke]
• add matrixmultiply
• better handling of malformed size values [Simcha Kosman]
• improve performance of vips_shrink() [kleisauke]
• svgload: add support for custom CSS via stylesheet option [lovell]
• heifload: unlimited flag removes all limits (requires libheif 1.19.0+) [lovell]
• heifsave: improve alpha channel detection [lovell]
• convi: ensure double sum precision for floats [lovell]
• improve guard against corrupt ICC profiles with older lcms2 versions [kleisauke]
• improve vips_sink_screen() thumbnail rendering
• heifload: improve detection of seek beyond EOF [lovell]
• jp2kload: add oneshot flag [mbklein]
• share and reuse openslide connections
• drop support for openslide 3.3
• fix vips_quadratic()
• rename thumbnail's export/import profile options as input/output for
  consistency
• restore vips_remosaic(), it was not being linked
• deprecate VipsSaveable, add VipsForeignSaveable
• move vips_image_preeval(), vips_image_eval(), vips_image_posteval() into the
  public API
• jxlsave: add support for chunked save (requires libjxl 0.9.0+)
• increase minimum version of libjxl dependency to 0.7.0
• increase minimum version of libheif dependency to 1.7.0
• improve scaling of hough_line feature space [ecbypi]
• heifload: limit per-image memory usage to 2GB (requires libheif 1.20.0+)
• svgload: add support for scRGB output via high_bitdepth flag [kstanikviacbs]
• remove vipsprofile from default install
• add ppmload_buffer [kleisauke]

Windows binaries here:

https://github.com/libvips/build-win64-mxe/releases/tag/v8.17.0

Thank you again to all the volunteers who worked hard on this release!

First release candidate for 8.17.0.

Changes since 8.16.1:

• gifsave: add keep_duplicate_frames flag [dloebl]
• add Magic Kernel support [akimon658]
• tiff: add threadsafe warning/error handlers (requires libtiff 4.5.0+) [lovell]
• tiffload: add support for fail_on flag [lovell]
• tiffload: add support for unlimited flag (requires libtiff 4.7.0+) [lovell]
• much more reliable operation caching
• colour: add support for auto-selecting the rendering intent [kleisauke]
• add matrixmultiply
• better handling of malformed size values [Simcha Kosman]
• improve performance of vips_shrink() [kleisauke]
• svgload: add support for custom CSS via stylesheet option [lovell]
• heifload: unlimited flag removes all limits (requires libheif 1.19.0+) [lovell]
• heifsave: improve alpha channel detection [lovell]
• convi: ensure double sum precision for floats [lovell]
• improve guard against corrupt ICC profiles with older lcms2 versions [kleisauke]
• improve vips_sink_screen() thumbnail rendering
• heifload: improve detection of seek beyond EOF [lovell]
• jp2kload: add oneshot flag [mbklein]
• share and reuse openslide connections
• drop support for openslide 3.3
• fix vips_quadratic()
• rename thumbnail's export/import profile options as input/output for
  consistency
• restore vips_remosaic(), it was not being linked
• deprecate VipsSaveable, add VipsForeignSaveable
• move vips_image_preeval(), vips_image_eval(), vips_image_posteval() into the
  public API
• jxlsave: add support for chunked save (requires libjxl 0.9.0+)
• increase minimum version of libjxl dependency to 0.7.0
• increase minimum version of libheif dependency to 1.7.0
• improve scaling of hough_line feature space [ecbypi]
• heifload: limit per-image memory usage to 2GB (requires libheif 1.20.0+)
• svgload: add support for scRGB output via high_bitdepth flag [kstanikviacbs]
• remove vipsprofile from default install
• add ppmload_buffer [kleisauke]

A test version of 8.17 to help nip4 development.

Changes since 8.16:

• add keep_duplicate_frames option to GIF save [dloebl]
• add Magic Kernel support [akimon658]
• tiff: add threadsafe warning/error handlers (requires libtiff 4.5.0+) [lovell]
• tiffload: add support for fail_on flag [lovell]
• tiffload: add support for unlimited flag (requires libtiff 4.7.0+) [lovell]
• much more reliable operation caching
• colour: add support for auto-selecting the rendering intent [kleisauke]
• add matrixmultiply
• improve performance of vips_shrink() [kleisauke]
• svgload: add support for custom CSS via stylesheet option [lovell]
• heifload: unlimited flag removes all limits (requires libheif 1.19.0+) [lovell]
• heifsave: improve alpha channel detection [lovell]
• convi: ensure double sum precision for floats [lovell]
• improve guard against corrupt ICC profiles with older lcms2 versions [kleisauke]
• heifload: improve detection of seek beyond EOF [lovell]
• add "oneshot" option to jp2kload [mbklein]
• improved scheduling for vips_sink_screen()

A test version of 8.17 to help nip4 development.

Changes since 8.16:

• add keep_duplicate_frames option to GIF save [dloebl]
• add Magic Kernel support [akimon658]
• tiff: add threadsafe warning/error handlers (requires libtiff 4.5.0+) [lovell]
• tiffload: add support for fail_on flag [lovell]
• tiffload: add support for unlimited flag (requires libtiff 4.7.0+) [lovell]
• much more reliable operation caching
• colour: add support for auto-selecting the rendering intent [kleisauke]
• add matrixmultiply
• improve performance of vips_shrink() [kleisauke]
• svgload: add support for custom CSS via stylesheet option [lovell]
• heifload: unlimited flag removes all limits (requires libheif 1.19.0+) [lovell]
• heifsave: improve alpha channel detection [lovell]
• convi: ensure double sum precision for floats [lovell]

A test version of 8.17 for nip4 development.

Changes since 8.16:

• add keep_duplicate_frames option to GIF save [dloebl]
• add Magic Kernel support [akimon658]
• tiff: add threadsafe warning/error handlers (requires libtiff 4.5.0+) [lovell]
• tiffload: add support for fail_on flag [lovell]
• tiffload: add support for unlimited flag (requires libtiff 4.7.0+) [lovell]
• much more reliable operation caching
• colour: add support for auto-selecting the rendering intent [kleisauke]
• add matrixmultiply
• improve performance of vips_shrink() [kleisauke]
• svgload: add support for custom CSS via stylesheet option [lovell]
• heifload: unlimited flag removes all limits (requires libheif 1.19.0+) [lovell]

Changes since 8.16.0:

• support multipage JXL
• fix PFM byte order on big-endian machines [agoode]
• morph: fix erode Highway path [kleisauke]
• morph: fix C-paths with masks containing zero [kleisauke]
• fix --vips-info CLI flag with GLib >= 2.80 [kleisauke]
• make subsample-mode=on and lossless=true mutually exclusive [kleisauke]
• fix SZI write with openslide4 [goran-hc]
• heifsave: prevent use of AV1 intra block copy feature [lovell]
• threadpool: improve cooperative downsizing [kleisauke]
• fix alpha shift during colourspace conversions [frederikrosenberg]
• heifsave: set image orientation using irot and imir transformations [lovell]
• XYZ2Yxy: guard against divide by zero
• fix MSVC compile error [na-trium-144]
• exif: ensure enumerated entries can be converted to string values [lovell]
• gifsave: add support for eval callback, ensure correct return code [lovell]
• tiffsave: honor disc threshold during pyramid save [kleisauke]
• fill_nearest: fix a leak
• colour: use suggested rendering intent as fallback [kleisauke]
• morph: fix Orc path with large masks [kleisauke]
• invertlut: fix final value in some cases
• matrixload: fix file format detect for some matrix types
• radload: improve sanity check of colour-related headers [lovell]
• heifsave: reject multiband images [lovell]
• heifload: prevent possible int overflow for large images [kleisauke]
• tiffload: add missing read loop [kleisauke]
• prevent possible use-after-free when debugging via --vips-leak flag [lovell]
• avoid possible overflow when multiplication result is cast up [lovell]

Windows binaries should be here shortly:

https://github.com/libvips/build-win64-mxe/releases/tag/v8.16.1