Open hexabits opened 10 years ago
Including original NIF, original collision, modified collision, and template (which is just original NIF in this case). No output NIF, since ChunkMerge fails.
These had to be done by hand.
In case there is any confusion:
NIF | Description |
---|---|
scastlewall04 | Original NIF |
scastlewall04_coll+ORIGINAL | Original collision, from ChunkExtract |
scastlewall04_coll+EXPORT01 | Modified collision, exported from Max |
scastlewall04+TMPL | Modified NIF in order to avoid ChunkMerge error |
scastlewall04+INPUT.nif | Pre-ChunkMerge input NIF |
scastlewall04+OUTPUT.nif | Post-ChunkMerge output NIF |
scastlewall04+FINAL.nif | Original NIF with bhkCO from +OUTPUT |
Importing scastlewall04.nif and scastlewall04_coll+ORIGINAL.nif into 3ds Max will give you two meshes in different locations.
The collision is "incorrectly" at the origin, whereas the visible part of the mesh is offset by its transform values, so to work on the collision at all, you need to move the visible mesh. This isn't a huge ordeal, but it can be very tricky. If you move the collision, you need to reset the transform before export, or manually zero out the transform in NifSkope.
I am using 1.2.2.978. When ChunkExtracting a collision from a NIF which has bhkCollisionObjects in sub-nodes i.e. not the parent BSFadeNode, the resulting NIF has transforms that are all zeroed even if the parent NiNode(s) had transform offsets.
Example: meshes\architecture\solitude\scastlewall04.nif
Structure:
The NIF produced by ChunkExtract does not offset the collision into the correct position. It should either keep the same depth, and however deep replicate the parent NiNodes with their transforms intact, or it should condense all these transforms into one transform for one parent NiNode.
This makes importing the collision into 3ds Max a pain since the collision does not show up where it should, since the visible mesh and the collision no longer have the same origin/transform. Secondarily, there is a compounding issue with ChunkMerge and bhkCollisionObjects where sub-nodes aren't recognized by Templates, and also the bhkCO is unable to be inserted back into the mesh at the proper depth. (Issue #58)