ImageVis3D 2.0 is a major release, bringing many new features and a great number of bugfixes. Changes since 1.3 include:
New Features
- ImageVis3D now supports rendering of geometry data mixed with volume renderings.
- A '2D slicing' render method was added for improved GPU compatibility.
- New file formats: Analyze 7.5, Kitware 'MHD', IASS (contributed), 64 bit floating point data in raw format.
- Supported file types can now be dragged and dropped into ImageVis3D from standard file browsers, such as Windows Explorer, Nautilus, or Mac's Finder.
- Support for ImageVis3D Mobile 2.0
- Support for new methods of stereo projections
- RGBA and RGB data can now be rendered with the 1D transfer function
- A batch mode (command line) converter is now distributed as part of ImageVis3D.
Bugs fixed in this release
- Fixed issues reading some DICOM datasets.
- Fixed a problem when dynamically downsampling 16bit data to 8bit data.
- Fixed a long standing rendering bug that occurred in low memory conditions.
- Fixed long standing problems associated with downsampling large histograms.
- Fixed error detection when reading improperly-formatted textual data.
- Fixed issues interpreting data types with the BOV reader.
Changes which are only of interest to developers
- 'Headless' rendering mode for external applications that want to use Tuvok for rendering.
- Abstractions were added to allow one to replace Tuvok's IO layer. See the 'Getting Data into ImageVis3D' manual for more information.
- Added example programs for using Tuvok without ImageVis3D.
Known issues with this release
- The 'Stack' ("stk"), single-file TIFF volume, and Fraunhofer EZRT ("rek") file formats can only be read, not written.
- The 'Brick Of Values' (BOV) reader does not support BOVs in which the data are split across multiple files.
- RGB[A] are the only kinds of multicomponent data supported.
- NVIDIA GPUs on Linux expose a driver issue which can hang the system for prolonged periods of time. To workaround this issue, we recommend sticking with the slice based volume renderer Linux for this release.
- Full-color datasets do not display a proper histogram in the transfer function editors.
- In any mode but MIP, the `Recorder' always uses the highest resolution data, regardless of any other settings.
- In the "Disable LOD" mode of MIP Recordings, progress information is not given. Progress is being made; a visibly-correct rendering is not required to create an on-disk image which is visibly correct.