pouët.net

With a tripod and a reflex you can take a lot of photos of a scene with different focus positions. By picking the most focused points from all the photos, you can generate a depth map.

I've find very few information about this in internet, just one paper, but it looks it works good. I think I'm going to try today. I will share results and/or dataset if you want.

cool idea. keep us posted!

hmm... might work, but I suspect it will be hard to get very accurate..

I've tried with that before, but to get half decent results you'll need a very shallow DOF. f/1.2? :)
I've only tried it with a compacta camera, with an obviously wide dof. I should try with a friends Nikon 50mm f/1.2.
I've used something called jimage or imagej, with a plugin.

yep please show

I've started every sentence with «I've», so I'll just continue.
I've found the plugin: http://bigwww.epfl.ch/demo/edf/index.html (not really sure it was that)

My very first dataset so we can play: http://www.romancortes.com/ficheros/dataset1.rar

I will try to do another later, with less shacking of the camera, a more interesting scene and more shots

Texel, can provide the ratio of f between the first and last image in the sequence? (I will assume a linear change in between.)

I have no idea Optanes, but I tried to do it linear, of course.

Have you tried looking at the EXIF?

Can it be put in a sequence of linked tables for C64? ;)

http://www.raytrix.de/index.php/home.181.html