FBX to GLM converter

[Pande]

I think the real issue is the vertex lighting in the renderer - it should be detecting vertices that are in the exact same position and merging the normals over the seam for lighting. 

 

I agree, this is one of the worst problem about character models.

[Xycaleth]

The problem is and always has been the exporter. I'll look at fixing this problem in my exporter. I have a few ideas about how to go about fixing the problem.

[Tempust85]

If you import a GLM, the splits are there as they are ingame so it's not the renderers fault. The problem is either the dot xsi 3.0/3.5 format or carcass itself.

[Archangel35757]

I talked with Xycaleth on IRC and he might have an idea of what to do, but it's pointless until we can get animated models + animations running through the converter. After everything is working, I'd like to have XSI 3.0/3.5 file processing so we can still work with the Raven source files otherwise it will be a BIG job to have to convert them all to FBX.

I just saw this... I also have the XSI Converter Toolkit Software, that along with the XSIFTK, should also allow for making a standalone XSI-to-Ghoul2 converter application. I'll email it to Xycaleth.

[Archangel35757]

If you import a GLM, the splits are there as they are ingame so it's not the renderers fault. The problem is either the dot xsi 3.0/3.5 format or carcass itself.

Export a 3 light setup on the model and do a text edit on an original dotXSI 3.0 to add the lights (@@minilogoguy18 and I did this back when I was working on the 3ds max exporter) and use the XSI Viewer to see if the UVW seams are visible. My hunch it's the conversion to Ghoul2.

[Tempust85]

I was just thinking. Would it be possible to add bone scale support to the ghoul2 format?

 

For example, you set a bone's scale via key frames as you do with position & rotation.

 

While on the topic of adding things, are jiggle bones a possibility?

[Xycaleth]

The GLA file already states how much to scale the model, but this applies to the whole model. Are you asking to add support for scaling individual bones? Jiggle bones won't be possible without changes to the renderer so that's a no go - at least for an exporter.

[Tempust85]

Yeah individual bones, but that would require a ghoul2 version bump right?

[Archangel35757]

Why would you want to scale individual bones?

[Tempust85]

It would be a nice feature for future modders and there are many things you could do animation-wise that would look pretty cool. You could for example have a mod where you can make your character thin, heavyset or somewhere inbetween.

[Archangel35757]

Seems like that would lead to UVW stretching when the mesh stretches.

[Tempust85]

Yeah it would. Still a neat feature to have methinks, but I have no idea how hard it would be to add support.

[Pande]

Would rather blend between two identical-vertex-order models the way SWG does, provides more control for that sorta thing. 

[Xycaleth]

I've updated the program. Same link as always

 

https://dl.dropboxusercontent.com/u/874909/fbx2ghoul2.zip

 

So now it can convert weighted models from FBX to GLM, as long as the model is using an existing skeleton (such as _humanoid.gla).

[Tempust85]

@@Xycaleth

 

I'm only getting "model_root" bone used for my stormie, the same file I exported for you to test with. It's ignoring every other bone.

[Xycaleth]

Did you read the read me file?

[Tempust85]

Lol yes, it won't compile unless there is a GLA to point to.

 

Tried this:

fbx2ghoul2.exe -anim models/players/_humanoid/_humanoid.gla models/players/dt_stormtrooper/stormtrooper.fbx

And This:

fbx2ghoul2.exe -anim models/players/_humanoid/_humanoid.gla stormtrooper.fbx

In a bat file, no errors come up and it appears to compile fine.

[Xycaleth]

Is that where the GLA and FBX files are actually located, relative to your bat file? (So you're running it from your base/ folder?)

[Tempust85]

Yeah, the converter & FBX are in my base folder and the GLA is in models/players/_humanoid/_humanoid.gla. The converter is setting each mesh object to be weighted to model_root. I've tried different FBX versions and tried ASCII as well and no difference.

[Xycaleth]

Oh, it looks like I forgot to add an error message if the GLA file couldn't be read. If you run the converter on the FBX file without specifying a GLA file then that's the output you should expect. I'm guessing the converter doesn't output anything apart from "Converting <modelname> to GLM" when you specify the GLA file. Have you extracted the GLA file from the PK3? If so, try putting the .GLA file in the same directory as the converter and running:

 

 

fbx2ghoul2.exe -anim _humanoid.gla stormtrooper.fbx

[Tempust85]

Yeah and still only model_root. I had a look at the GLM and the GLA looks to be mentioned, just no bones.

[Archangel35757]

@@Xycaleth -- Are you recursing down the bone tree from model root? Maybe something wrong there...

[Xycaleth]

Interesting.... @@DT85, can you send me the model and GLA file you're using? Maybe I should add some logging in to this to make debugging a bit easier...

[Tempust85]

I'm using a fresh GLA from base JKA, and exporting the exact same FBX file as I sent you for testing. If I specify an invalid path to the GLA then nothing will compile so it must be getting the correct path, just not doing anything with the GLA.

[Xycaleth]

Odd. I'm using the same GLA and FBX file then... I'll double check things through I guess