Exports
.blend
Always written by the automatic pipeline and by a manual Load Data run, saved as <recording_name>.blend in the recording folder, the same convention as an app-driven session.
FBX and BVH
The Export 3D Model panel exposes: destination folder, format (fbx or bvh only), bones naming convention (default, metahuman, daz_g8.1), rest pose type (same three choices), restore-defaults-after-export (default on), and an FBX subsection (leaf bones, off by default, plus primary/secondary bone axis enums).
The export sequence in export_3d_model.py is deliberately reversible. It optionally renames the armature object to root (the manual operator passes rename_root_bone=True; the automatic pipeline does not), swaps in the selected rest pose, poses the armature by markers at frame 0, then renames bones per the naming-convention mapping tables before exporting. A code comment explains why: the FBX exporter treats whatever pose the armature holds at export time as its rest pose, so frame 0 is forced to the expected rest configuration first. FBX export selects the armature and its child meshes and calls export_scene.fbx with all-bones baked animation; BVH export calls export_anim.bvh across the scene frame range. With restore enabled it then undoes everything: inverse bone renaming, original rest pose, mesh realignment, removal of the temporary DAZ correction constraints and forearm twist bones, recreation of the Metahuman thumb.carpal bones, and restoration of the original armature name.
glTF is not available
A gltf branch exists inside export_3d_model(), but it sits directly under a TODO - Fix glTF export of animations - output appears broken comment, the default format list is ['fbx', 'bvh'] with 'gltf' commented out, and the UI format enum offers only FBX and BVH. Treat glTF output as not implemented.
Video rendering
The Export Video panel renders composited videos straight from the scene. Profiles live in video_config.py:
| Profile | Resolution | Cameras | Rendered elements |
|---|---|---|---|
debug | 1920x1080 | Front | center of mass, rigid body meshes, videos, Skelly mesh |
showcase | 540x960 portrait | Front | videos, Skelly mesh |
scientific | 1920x1080 | Front plus right and top inset views | center of mass, rigid body meshes, videos |
multiview | 1920x1080 | Four quadrant views (front, right, top, left), 960x540 each | center of mass, rigid body meshes |
custom | user-set | any of front/left/right/top, each with its own size and normalized position | chosen implicitly by scene visibility |
Each profile also carries an output name suffix (_debug, _showcase, and more) and a logo overlay placement. The custom profile is assembled from the panel's UI properties before rendering: overall resolution, per-camera enable/size/position, and an optional FreeMoCap logo with adjustable scale and position.
Rendering uses EEVEE Next with MPEG4/H.264 output through Blender's FFmpeg settings, transparent film, and a bundled background image; the pipeline places cameras and lights, rearranges background videos, prerenders each camera scene, composites the result into the recording folder, and resets scene defaults afterward. Before running, the operator clears leftover compositor nodes, movie clips, and previous Render_Camera scenes so repeated exports do not accumulate garbage.