Skip to content

NifSkope 2.0 Pre-Alpha 5

Pre-release
Pre-release
Compare
Choose a tag to compare
@hexabits hexabits released this 11 Apr 22:29
v2.0.dev5
0eb1bd1

Changelog

2016-04-20

  • Material files (BGSM/BGEM) will no longer crash NifSkope if they are in JSON format. Note: They will still not be read if in JSON format.
  • Fixed several bugs caused by the partial data model refactor.

2016-04-11

General

  • Vertex Selection mode
    • Due to implementation limitations, vertex selection works on a max of 255 trishapes. NIFs with more trishapes than that will not support vertex selection correctly.
      untitled-3
  • Partial loading refactor. Basically, all meshes will load more quickly.
    • To test this I ran XML Checker (in File menu) on entire folders. For the average mesh the speedup is about 2.0x. Full Benchmark
    • This also benefits copy/paste speed for branches.
    • Note: This does not necessarily confer a 2x increase on a single file opened as there are additional things that occur like 3D initialization, texture loading, etc.
  • Node rendering improvements.
  • [Fix] An issue with texture find operations was breaking the UV Editor as well as absolute texture paths.
  • [Fix] A lack of #version pragma in the vertex shaders was the likely cause of random AMD/Intel driver issues.

FO4-Specific

  • Connect point visualization (BSConnectPoint::Parents)
  • PreCombined visualization (BSPackedCombinedSharedGeomDataExtra)
  • Alpha rendering improvements. Also correctly ignores vertex alpha on trees.
  • [Fix] The appearance of wireframes, vertices, etc. was not consistently following the colors the user defines in the Settings.

Update Notes

Please note that there could be unforeseen issues with my partial refactor of the NIF loading. This is the first update where I've changed the underlying data/model classes.

The load refactoring mentioned above was actually done specifically to benefit FO4 NIFs. Using a proper XML definition for the vertex data, the previous release would take 18 minutes to scan meshes/architecture. After the refactor the proper vertex data definition only takes 17 seconds to scan. However, in the previous releases I did not include the proper XML definitions for this reason, and the XML actually consisted of several optimized variants. Even still, the new "proper" way is over 2x faster than the old "optimized" way. (See Benches)

Changes Prior to 2.0.dev5

For the previous changelog visit the 2.0.dev4 release page.

Assets 3