.dds works in NifSkope but not Oblivion

[LHammonds]

I received the NIF and DDS files.

 

My initial concern when I saw the archive (and required that I take a double-look) was that the NIF of the sword was 2,433 KB (2.5 MB). That is fairly huge.

 

When I opened it up with NifSkope, I immediately noticed a "NiPixelData" node in the root that I have not seen before. I can only guess it is due to vertex coloring or image embedding. It was also lacking the NiStringExtraData node which normally signifies either "BackWeapon" for claymores or "SideWeapon" for longswords.

 

The NiTriStripsData showed that the initial model contained 11,194 verticies and 5,825 triangles. I was scratching my head on this one because the amount of triangles is usually greater than the amount of vertices by about 2 to 3 times as much.

 

But it seems that this model is acting like a Direct X (.x) model in that every triangle has it's own vertex...therefore, where a normal vertex exists as a single vertex, the vertex in this mesh is actually 4 vertices if there are 4 faces connecting to it.

 

If I export the model as is, it exports to a 629 kb NIF file but still has those 10,000 verts in there. However, if I use Blenders "Remove Doubles" feature, it tends to merge verticies that are just simply close to each other rather than exactly on top of each other. If I can figure out how to fine-tune the merge doubles feature, I can clean this mesh up and reduce the size quite considerably and increase performance by a great amount.

 

That may explain why the faces had problem figuring out what to do. Tangent space cannot be calculated among the entire object because no single face was actually connected to another.

 

I'll get back to this when I find a merge solution.

 

EDIT #1: I found the threshold setting which is on the Edit panel --> Mesh Tools. Trying to find optimum value...

 

EDIT #2: I've managed to remove the duplicate verts but this has caused some issues with the UV map which means it needs to be tweaked or simply re-created. I'll let you decide which and just hand you the Blender file. The current NIF file is 250 kb with 4,271 verts and 7,045 triangles and I have every reason to believe it will look the same in Oblivion as it does in NifSkope...although there are some faces that have UV map issues (very few)

 

EDIT #3: MediaFire link sent to your inbox.

 

LHammonds

[Genzel]

Do you know what caused the issues? And are all the lines a type of collision box?

[LHammonds]

I really do not know "how" your mesh got into the condition it was in.

 

Oh, and yes, the lines are a custom convex collision mesh that is tailored to the shape of your weapon. I actually created a very low-poly box (had about 64 vertices) version of your mesh and used the Hull script on it to create the collision model which is then used for your sword. The hull script would not have been a wise choice to use on a mesh that had thousands of vertices. :thumbsup: Oh, and if you are wondering why I did not use standard cylinders, it is due to the NIF Scripts export not being able to create perfect-working collision models for primatives such as boxes and cylinders. You would actually have to use NifSkope to copy/paste primatives into your NIF file and the do a lot of NifSkopery to clean it up and re-size and position.

 

Layer 1 = Your model

Layer 2 = Collision model

 

LHammonds

[Genzel]

Thanks for all the help! I finished another mesh and its texturing today, and I guess I'll be seeing if it has the same problems soon :sweat: . Now I have to finish studying for my exams. :unsure: