Releases: unibas-gravis/scalismo-faces
0.9.0 - Modality Renderers
New Features
- depth-map, albedo, illumination and normal renderer added (#98)
- rendering of correspondence images (with backface culling) and tests. (#97)
- Draw a random coefficient sample of MoMoInstance (#94)
- Masking Models (#95)
- Pixel shader that creates Lambertian shading with a point light source. (#96)
- MoMoRenderer: Speed up the rendering of the same instance under different poses and illuminations by caching the instance() according to the MoMoCoefficients and not the full RenderParameter. (#91)
- Keep the original domain when creating a posterior PancakeDLRGP. (#93)
Bug fixes
0.8.0
This release brings face manipulation, robust illumination estimation, segmentation and other new functionality to our framework.
This is the version used for the semantic morphable model tutorial on http://gravis.dmi.unibas.ch/PMM/.
Detailed changes (since 0.5.0)
Features
- Robust illumination estimation (#75)
- Adding Segmentation by Loopy Belief Propagation (#76)
- added segmentation proposal
- evaluator for labeled images added
- adding mirroring capability to PixelImageOperations (#72)
- adds labeled independent pixel evaluator (#77)
- Morphological filters (#62)
- adding push-pull interpolation (#63)
- added Poisson solvers with tets (#66)
- adds poisson inpainting: seamless cloning
- WarpExtrapolator: find values for undefined warp regions (#68)
- Face manipulation (#69)
- updated to scalismo 0.16.0 (solves issues with random numbers in threads)
Fixes
- resolves #70 - fix HSV RGB conversion and added tests
- fix number format not writing points depending on local setting of the OS
- some test changes
- updated PLY and GravisArrayIO writer to keep points for numbers
- spherical harmonic in band 4 subband 1 is now scaled correctly.
- fix for Scala 2.11
- added error message if landmarks file corrupted (#74)
- constructor for MoMoCoefficients without expression (#61)
- Maximum Number of Levels of Gaussian and Laplacian pyramid now as expected. (#56)
0.5.0
This release brings Scala 2.12.1
This is the version used for the summer school about probabilistic shape modeling and fitting held 2017 in Basel.
Detailed changes
- Fixed missing Random in
GaussianTranslationProposal, #59 ImageCenteredProposal, keep image position of face fixed, #53- Deterministic Tests, #51
- Fixed inconsistent handling of logs in detection maps, #45
GUIBlockrenaming for easier imports, #44- SH light from breeze vector, #46
- Close
Scannerafter reading - more PLY data types, #41
- Scala 2.12, #39
- Cleaner
PancakeDLRGP, #37
0.4.1
This is an important bugfix update to 0.4.0 and solves
- gamma correction issue #25 and #26 (thereby removing type
SRGB), fix #28 - reading of V3 rps fixed, #30 with #31
- landmarks evaluator: added missing parts to get a working version #27
- remove left-over debris of unfinished segmentation parts, #32
0.4.1 should be preferred over 0.4.0 for all applications.
Remark: Some fixes required removing non-functioning interfaces (e.g. SRGB removed)
0.4 - Rendering & Fitting
Warning: consider issue #25 when using this release - best wait for the fix (should be available very soon)
First full open source version with
- Images
- Meshes
- Rendering
- Statistical models
- Fitting methods: Proposals and Evaluators for use with the sampling library already in scalismo
It is now possible to create images and handle the Statistical face models, e.g. the Basel Face Model
v0.3.0
Respects sRGB and linear RGB color space defaults
Internally, all our operations typically rely on the linear RGB colorspace. However, images are stored using sRGB color space (if not explicitly stated differently). From v0.3.0 on onwards, images are loaded using the SRGB type which encodes sRGB color. Conversion is possible through SRGB.toRGB and RGB.toSRGB.
Expect images which are loaded from disk or displayed to behave slightly different.