Scanning a flat object

[inksigne]

I have a problem that after checking many ways I don't find a solution.
I try to scan a flat piece (0,5cm) wide but every time Zephyr makes the same, only render one side of the object, and it seems that when the camera arrives to the position where the objets is completelly frontal it missed the objet and only makes one side.
I had made the two sides prefectlly in two diferent projets but was imposible to make them in the same project.
I try diferent turntable bases, white, chess pattern, with control points, and not matter, also I mask ALL the photos, only some of them, make photos in on direcction, in the other, and not solution.
And it only makes this with this flat object, the other objetcs I scanned were perfect.

Also I have another problem with this, when I try to make one side, save as an object, after make the other side, and import the first one to complete the two sides the imported object comes completelly out of range and position.
And I try to impot the discarded cameras in the proyect with the sides and back objet, but never was succesful becuase software says that the camera points don't have coincidences for the fusion.

Any one have this problem? and a solution?

I have attached two photos, one of the iobject in a frontal photo, and anaother were the camera loose the tracking

[DanielMuirhead]

Two ideas:

1. Take more photos at the corners (at the join between front and back). In my image below, if your subject is "a", and your current capture trajectory is the cameras represented in the ring "b", then capture more photos at location "c", on both left and right side.



2. The subject's surface is uniform and featureless (except its geometry), i.e. it is plain grey plastic. Temporarily coat the surface with something that will make it easier to be detected by 3DF. E.g. water-based paints, daubs of different colors, to give the surface a unique randomised texture. (Surfaces with good random texture, e.g. stone masonry with its naturally occurring randomised grain, are much easier to process, compared to surfaces which are shiny, self-coloured, uniform.)

Just my two cents, all the best!

[cam3d]

Hi inksigne! -

DanielMuirhead Made some great points, and to add some additional cents: if you put a feature-rich surface below the subject, you'll be able to use that as an aide for better camera orientation. By feature-rich, I mean something like Perlin noise (attached) but as DanielMuirhead suggested, any unique randomized texture (with lots of detail) will help! Wood, concrete, whatever, as long as it has lots of surface detail, isn't transparent or translucent or shiny

Furthermore, coating the surface of your subject with a very light dusting of talcum powder will be beneficial to both camera orientation and surface reconstruction. Have a go with that, see how it resolves, and if you're still stumped, I'd be more than happy to check out your data and give you detailed feedback on improving your image acquisition

[inksigne]

Thanks for your help.
I will try all these tips and give your feedback of my results.

Thanks again

[cam3d]

No worries!