XPS to FBX converter — XNALara model in, one FBX out
VTubeMe XPS to FBX Converter reads your XNALara / XPS model — generic_item.mesh, xps.xps or the .mesh.ascii text version — and writes a binary FBX: skeleton, weights and textures, in one file. No Blender, no add-ons — and you only pay for successful conversions.
What comes across, and what cannot
The model keeps its bone names, hierarchy and rest pose. What XNALara's own renderer adds on top of the geometry and the main texture has no place in FBX.
Comes across
- Every bone, with its name and parent
- Skin weights, so the mesh deforms with the rig
- UV layers and the model's own normals
- The diffuse texture of every part (DDS, TGA, PNG), embedded
- Vertex colours
- One mesh per XPS part, named after it
Does not
- Bump, specular and lightmap textures
- XNALara's shader settings (render groups)
- Saved poses (.pose files)
- Morphs — XPS models have none to begin with
Where people take the FBX
- Blender, Maya, 3ds Max — pose, re-texture or edit the model without an XPS add-on.
- Unity and Unreal — a skinned mesh with its skeleton. For Unity's Humanoid, check the bone mapping in the Avatar configuration: XPS names vary by source.
- VTuber apps (VSeeFace, Warudo, VNyan) — they read VRM, not FBX. The 3D model to VRM converter takes the same XPS upload and maps the humanoid for you.
Common questions
What do I upload?
The model folder as a .zip (or .7z / .rar): generic_item.mesh (or xps.xps, or generic_item.mesh.ascii) together with its textures. The model file only points at its images by name, so on its own it converts untextured. When a folder holds both generic_item.mesh and generic_item.mesh.ascii, they are the same model; the binary one is read, and the text one only if the binary cannot be.
Which XPS files work?
The three containers XNALara and XPS write: the original binary generic_item.mesh, the newer xps.xps (and .mesh saved by XPS), and the text generic_item.mesh.ascii. Bone names and the skeleton are kept exactly as the model has them.
Are the textures kept?
The main (diffuse) texture of every part is — including .dds and .tga — embedded in the FBX. XNALara's bump, specular and lightmap slots, and its shader settings, have no standard FBX equivalent and are not converted; the materials come out as standard ones you can re-light in your tool.
What about poses and facial expressions?
The model arrives in its rest pose; saved XPS poses (.pose files) are not applied. XPS models carry no morphs (blendshapes), so there are none to convert.
Can I use the model in a VTuber app?
VTuber apps read VRM, not FBX. The FBX to VRM converter takes XPS models directly too: it maps the humanoid skeleton and makes a VRM that walks and follows your tracking — only without facial expressions, which XPS models do not have.
What does it cost?
Your first conversion is free. After that a single credit is $2.99, or packs: 3 for $7.99, 8 for $14.99, shared with every paid tool on the site. A credit is spent only on a conversion that produced a file; failures refund automatically, and a result you have not downloaded can be refunded from the result panel.
Something came out wrong
Refund it before downloading from the result panel. If the file matters, a person can do the conversion by hand ($7.99) — the option is in the result panel. Otherwise, support@vtubeme.com.
May I convert a model I downloaded?
The converter cannot check that for you. Many XPS models are ports from games or other people's work, and their readme often limits where they may be used. Read the model's terms first, and keep the result to what they allow.
No model of your own yet?
Turn a selfie into a rigged, face-tracked VRM avatar — free to make and preview, $7.99 to download.
Create my avatar