#71 Sload Mesh CTD's Oblivion after being PYFFI'd... bug report

[Koniption]

I used:
Nifskope 1.1.0-rc5 (for Windows)
PyFFI 2.1.11 cefd181
Blender 2.49b
Blender Nifscripts 2.5.8
Python 2.6.6

Niftools team,

I have been working again on revamping the sload race for Blackmarsh. The skeleton works fine, and is not the object that is crashing my game...it's after I PYFFI the sload's body mesh, that the game CTD's when I look in the direction of the sload.

The nif called “SloadMesh_OLD_WORKS”, is the body mesh that does work in the game. But I need to used the PyFFI'd “SloadMesh_NEW_CTD” nif, because I already PYFFI'd it in hope of fixing the gaps & tears that appear on the sload's body, and because I reattached this body mesh to the skeleton in Blender, after I edited the two “hand01 & hand02” bones on each side of the body, to be directly parented under the regular left and right hand bones.

See, the “SloadMesh_OLD_WORKS” nif, was exported for a skeleton where those two hand offset bones where directly under the forearm, instead. However, this mesh does work with the new skeleton, it just has the gaps/tears in the body mesh when the sload ragdolls, and also the ends of the hands look messed up.

I need to be able to use the “SloadMesh_NEW_CTD” nif, since it matches the skeleton...but like said, it CTD's the game.

If you take “SloadMesh_OLD_WORKS” and PYFFI it...it will crash the game, too. But if you don't PYFFI it, it will work in the game – no CTDs.

I have included meshes, textures, esp, and Blender scene - sent via PM to Amorilia and Corwin. Esp was saved using Oblivion with Shivering Isles. Sload is outside the intro sewer exit, up the hill to your immediate right once you exit, towards the Imperial City.

Thanks,
Koniption

[Amorilia]

Sorry for the too long delay!

I am now looking at this issue properly. Just by looking at the nif and comparing the version which works and the version which does not work - could you

1. try regenerating the skin partitions in nifksope after optimizing with pyffi, and see if it still crashes the game?
2. if the nif produced in 1. still crashes the game, could you regenerate skin partitions on the original working nif in nifskope, and see if that causes the nif to crash?

My suspicion is a bug with the way skin partition bone indices are currently being set up in pyffi, nifskope, or both.

[Koniption]

No problem on the delay!
I will try out your steps soon. I'll report back once I've gone through them.

[Koniption]

Okay, I just got back after following your steps:

1. Yes, both the "CTD" & "WORKS" nifs, still crash the game, when I Regen skin ptns after PYFFY'ing them.
2. Yes, it crashed the game after regen skin ptns, whereas prior to regen, it did not. (I regen'd skin ptns on "SloadMesh_OLD_WORKS").

Seems to me, that both Nifskope and PYFFI are buggy, from what I can now gather.
Btw, since this initial bug report was made so long ago, I am now using a newer version of NIfkskope, though not the latest...:
Nifskope 1.1.0-RC6

I can update to the latest Nifskope to test things, but I am unsure as to whether I'd need to update to a newer Python version, in which case, it'd make it harder for me to mod Oblivion. I know - I'm always lagging behind in the latest and greatest utlilties. :)

[Amorilia]

Awesome - thanks for the quick testing! This confirms my suspicion.

A technical explanation: your nif has skin partitions with just a single bone. That's highly unusual, and our skin partition algorithms apparently handle it currently in a way that is incompatible for the Gamebryo engine.

No point in trying with a newer nifskope - the skin partition code has not been touched for ages. I'll post back here when I think to have fixed the problem.