3D Markers to Mesh - position of mesh

I just tried this option on some tracked footage, and it created some vertices in a mesh.
But the position, size and orientation of the resulting points bears no obvious relation to the bundle markers.
With a lot of jiggering about I can eventually get it to line up with the bundles. But why doesn’t it create it in the “right” place to start with?

I’d like to know too. The Blender manual’s entry on motion tracking and reconstruction is really out of date. I bought the Track Match Blend training series off of the Blender store, but so far my internet connection has failed to download it! Gagh!

I’m having the same problems! I have a pretty good solve with 0.8 error but when I choose 3D markers to mesh the resulting points are totally off.
This feature is not explained anywhere, even a google search with “3D markers to mesh” provides very little information.

At the moment I have a very tedious workaround of first linking an empty to each tracker, then snapping the cursor to the empty and finally snapping a vertex to the cursor in the edit mode. I really don’t think I’m doing it in the way it was meant.

How do you guys attach meshes to walls or other objects in your 3D tracks?

Have any of you guys tried to do a manual/grid lens distortion calibration after or before the tracking. If your shot(s) doesn’t have any definitive lines that can be use as a guide, then film a grid calibration card with the same camera that was use for the original shot. Bring in that footage and do a grid calibration with K1, K2 or both. Those numbers could be use for your motion solve. Your 3d Markers to Mesh should be fine.

Now if you are experiencing some jitter in your solve or on a certain frame, that means a one or more markers are jumping around. After you have ran you motion solve, run “clean tracks - Select”. This will show you which track has a high outlier value. For me, anything above .6 or .7 could be cause alignment problems if the lens distortion is not calibrated for.

Thanks for the reply!

I have indeed a K1 lens distortion calculated in my solve.

But I don’t think it’s related to this issue, since the 3D markers to mesh produces points that are waaay off.

Does anyone have this feature producing expected results?

Hmmm… now that I think of it, could it be that I have to use this function before I setup my 3D-scene? Maybe the mesh points somehow align to the original orientation of the track (before the scene is re-orientated)?

That could be one reason. In better practice always solve, setup your scene orientation and don’t forget to select two features and apply a scale. See attached image. using version blender_2.67b-r57468.

Hmm damn I’m still struggling with this… I tried to do things in a different order, but the resulting points are still in a weird location.

Can anyone confirm if the “3D markers to mesh” feature is working properly with the latest version of Blender (2.67b)?

Does anyone have this feature working correctly?

I just tried this on a recent file (I usually don’t use the 3D Markers to Mesh feature.) The vertices all ended up wayyy below the ground plane and they were all oriented vertically. All the 3D markers were on the ground. It doesn’t have anything to do with setting Scale. Not sure what’s wrong.

ETA: I should point out that after I rotated them into the proper orientation, they lined up perfectly with the markers.

Steve S