'Synthetic 3D' of a single SEM image

[Rene]

Heribert showed a method to generate 3D on basis of a single image. Assuming a deeper layer in the subject is darker, the depth map can be generated on basis of luminance. That is a promising method for SEM with an inlens (upper SE) detector. An inlens detector is very sensitive and gives good grey levels, but the topology can be difficult to interpret, compared with the standard chamber SE detector.

Heribert showed it working on an image of Entomoneis (diatom). The pores were reasonably naturally shown. If however the pores are small like in this image of Actinoptychus, they go quircky. I assume the system thinks the dark pores are in a deeper layer, and places them in a different place in the image.

First image: original image. First double image without tilt, the second with default 3D settings (viewing angle 3) and the effect on the pores in a crop. Last image is a red-blue anaglyph. Watch especially the piece of debris and the slit-like rimportula sticking outat 4 o'clock in the image!

Best wishes, René

 

[Heribert Cypionka]

Dear René,

Thank you for showing your results!

To me it doesn't seem that bad. Looking at your original image I first thought it would not work at all, but it does!

Of course there are problems with the little dark holes, which are surrounded by above-level white rings. The method will set any black spots to the very background (while the white rings go to the top) and lets the spots look quirky in this case...

Also the background of the diatom shell doesn't look flat due to the shadowing.

So please be kind in your assessment - I still find it worth trying!

Regards,

Heribert