open_iA

Binaries

A zip file with prebuilt windows binaries is attached below; a linux version is available in our flatpak repository.

A few outdated modules were removed in contrast to last version (see below). If you need them, use a previous release. In addition the binary releases do not include the ONNX model runner (AI module) included in previous releases; it is still included in source code, but expected to either be removed or to receive a major overhaul in future releases.

Changes

This release contains many improvements and fixes: - General: - Faster histogram computation - Proper dependency handling for filter parameters in GUI (the inputs for parameters currently unavailable / not applicable because of other parameter values are now disabled dynamically) - Option to hide slicer/renderer title bar - Mesh renderer: Point/wireframe/surface option - ARM target architecture compatibility (through github actions, see below) - Slicer: - Optional lines indicating other slicer positions - Alternate color scheme for slicer/axis colors - Fix crash of slice export after dataset 0 removal - Option to hide ruler/scalar bar - Compression level option for slice .png export - (Arbitrary ->) Free slice plane: show plane parameters - Fix 0.5 spacing shift of linked slicers - Annotations: - Also shown in VR with new VRAnnotations module - Automatic positioning of new annotations in renderer/slicer - Surface/Mesh filters: - Fix crash on flying edge surface extraction - Split up combined surface extraction + simplification into separate filters - add triangles/normal computation filters - FeatureScout: - Fix class loading - Stop tool if dataset removed - Remove outdated/unmaintained modules: - Astra reconstruction - Foam Characterization - Uncertainty - ... and many more minor improvements and fixes

Full changelog: https://github.com/3dct/open_iA/compare/2024.9...2025.6

Test Data

There is no new test data available, you can still use the one from the 2020.09 release.

A few small test projects are now included in the source code, these are not included in the binary releases.

Reproducibility

The windows release was built using github action; see the build workflow. It can also be built with the same set of libraries using the open_iA superbuild, the library versions used are the defaults as in the 2025.6 tag. The linux flatpak is built using this manifest: at.zfp.openia.json.

SHA512 checksum for open_iA-win64-2025.6.zip: 829DF4571C8463E78B793887646C3B4E3C0A48B5C970FA88D9381AAE528FEFA1495C04E40733951ACE687508EBFF92B738A00303739F3F82A3A262E10E586D7C

Testing

While all of open_iA is basis for research prototype and therefore the occasional bug is to be expected, we explicitly want to point out that the following modules are not regularly used, and untested: - 4DCT - AdaptiveThresholding - BoneThickness - DreamCaster - DynamicVolumeLines - FeatureAnalyzer - FeatureAnalyzerComputation - FiAKEr - FuzzyFeatureTracking - GEMSe - MultiModalTF

As always, if you encounter any problems, please reach out in the issue tracker.