Voodoo with Blender

hi all.
Is there a technique for lining up voodoo camera tracker imported data into blender? Or is it simply a guessing game of eyeballing everything?
So far I just line shots up by eye. Is there a more straight forward/easier way to do it rather than the hit and miss approach?
Advice, tips or a link to a practical tutorial on lining up voodoo data with blender model would be great!


I do that in maya but it should work in blender.
first draw a plane that should be the floor from your data, or something parallel to the floor. It don t need to be a square.
Then create a plane at 0 0 0 in rotation and translation.then align the plane with the floor (maybe there is some automatic tool to align in blender), then put all the data and camera of vodoo under the plane (may be group in blender) then reset rotation and translation of the plane to 000 000 and all the data wil be align. this is just a workflow. may be it could be adapted in blender.

I’m not sure that I understand well your question, but the first thing to do seems to rotate all the scene of 90°, because the axis system used by Voodoo is different.
This done, you have to locate a tracker (a dot in your cloud) which is on the floor in the real world and select it in edit mode. If your ground has a slope, it is better to select 3 dots and make a face from them.
Then, separate this face from the cloud mesh and place the center of this new mesh in the center of the selection. Resize the mesh, and you will have a ground plane.
In your background image, it is better to have an object or a space between two object with a known dimension. This dimension will be used as a reference for rescaling your 3D models when you will import them in the scene. Linking the two dots by a edge will help. You will be able to measure the distance between them and display it in your 3D window.
It is almost sure that you will get some weird results, mainly for the objects close to the camera, because the focus length calcuated by Voodoo is never the same as the real focus the camera which was used for the shot, so weird perspective could be expected.
I noticed that objects in the foreground looks always smaller than they should.

Thanks for the replys, Phillippe and Dagobert!
Extracting vertices from the cloud to form a ground plane is a great idea.
But it still involves an awful lot of guess-work and trial and error.
Gonna try out your advice now and see how it goes :slight_smile: