FBX import new options definitions?

Hello everyone!

Could anyone define those import options? There is no any definitions in docs :frowning:


Hi AlFlakky,

I’ve responded on your post on the forums as well: FBX Import Options update - Feedback & Requests - Epic Developer Community Forums

Build Adjacency Buffers

Tooltip/Source comment: Required for PNT Tessellation but can be slow. Recommend disabling for larger meshes.

This is really helpful when importing large meshe. Building the adjacency buffer so that a mesh can be used with tessellation takes a long time for these and if you disable this option it can help speed up this process.

Build Reversed Index Buffer

Tooltip/Source Comment: Reduce the number of GPU state changes enabled. This is required to optimize a mesh in mirrored transform. Double index buffer size.

Transform Vertex to Absolute

Tooltip/Source Comment: If this option is true, the node absolute transform (transform, offset, pivot) will be applied to the mesh vertices.

The mesh will be set to the world origin, retain its tranform and any offset of the orginal DCC. Disabling this means that the Pivot, Transform, and offset will use what the mesh had in your modeling program.

Bake Pivot in Vertex

Tooltip/Source comment: Experimental If this option is true the inverse node rotation pivot will be applied to the mesh vertices. The pivot from the DCC will then be the origin of the mesh. Note:“TransformVertextToAbsolute” must be false.

This options, if there is any confusion, means that if you have a mesh that has a rotated Pivot and this is enabled, the mesh will rotate to align with the pivot.

Some of this is included with the Release Notes for the major versions. Some of this is taken directly from comments in the source files or the tool tips.

We currently have UEDOC-948 for this task. I’ve assigned myself this so I can get this updated in the near future.

Tim, could you clarify one thing - the two buffer options, they both increase the mesh size in memory? (GPU memory?)

Most of my meshes aren’t going to use either tessellation or mirrored transform - I’m guessing I should disable those options on import each time? (or are they automatically cooked out if not used?)

They can be disabled on import or from the Static Mesh’s Build Settings options if they are already in the engine.

I’m not 100% sure if they are disabled if not used or reduce the mesh size for memory. I’ve not done any tests with that particularly or looked to see. I know the Build adjacency buffer saves the time by not building it for use with tessellation, but I’m not sure if it’s disabled. My best guess would be that it is disabled though. Same for the Mirror Transform since it shouldn’t be doubling the index buffer if that’s disabled. Again, my best guess given the tool tip.