I want to build a FreeLSS-based scanner

Quote


QAs a video professional. I have yet to find a
ray-tracer program that's fast enough to he
usable in my business (I produce video logos and
graphics for television advertising). Could you tell
me whether there is a way to scan in objects in 3D
to the machine without drawing it all by hand? I
would he grateful for any help.

A There is no commercially available package to
do this, but after some thought it's theoretically
possible. |i does however involve quite a bit of extra
hardware and specialist 3D programming (it's a bit like
ray-tracing but in reverse). For those who fancy a chal-
lenge, here goes:

You have a robotic arm linked to your Amiga
through the serial port and controlled by the computer
- nothing outrageously special about this really. In the
arm's hand you mount a black and white video camera,
linked to the Amiga via a digitizer. You need a fairly
flexible arm. because the camera might have to look
around odd corners or see under overhangs.

Program the robot arm to move the camera around
the object, and take a series of black and white images,
Supposing that the object is illuminated well and is sit-
ting on a black background, the computer could workout
what the object looks like in 3D from the shades of light
and dark, if it knows the position and angle of the light
source. Then it would write that information into a file
suitable for a ray-tracer.

Although this approach would be quite cheap, it
would be slow and inaccurate. You would probably
need to tidy up the 3D object in the computer after-
wards with ray-tracing software. However, it would
certainly remove a lot of the leg work involved with
ray-tracers, as standard digitizers and flatbed scanners
are useless for 3D work.

A slightly different approach currently being used
in Virtual Reality research is to put the object on a
turntable and spin it slowly. A laser beam bounces off
the object, and by measuring the return time the depth of
all surfaces can be worked out and the overall shape
calculated. Despite being more expensive, it's very
accurate and takes less computer power, because the
mathematics are less complicated.

Quote
av8r1
I'm back to the familiar conclusion that if I'm going to get somewhere, I'm going to have to go outside, dig a hole, mine some germanium, make my own transistors, re-invent digital computing from scratch, and write my own software just to make sure the failure of humanity doesn't creep in.

Quote
av8r1
I mean seriously folks, has anyone gotten one of these damn things to work?

Quote
DragonFire
I must admit, LSS kit looks easier to assemble (no soldering required). The Hat for the Fabscan Pi has to have the through hole components soldered, and not all of the wires are pre-tinned. That's with the Watterott kits.

The Fabscan Pi is more complex to build, but in theory, being enclosed and better camera, I should get better accuracy. It is also maybe easier to modify the design if you want a bigger scanning volume. I like it a lot, and am trying to contribute to it. But it is an ongoing release.

Whether you need the extra possible accuracy that for the job you are attempting is a different story. "Good enough for the job" is a consideration. You only get one red line laser with the basic Fabscan Pi. And one servo to move it with. Another green line laser and servo help improve accuracy, in theory.

What I've found really annoying about this sort of project is sourcing suitable line lasers. They're kind of treated as an "adult" item in the UK, disclaimers everywhere about unsafe usage, danger, etc etc. "Would sir like to collect that in a plain brown bag, or anonymously posted to be signed for? No, we can't use regular mail, sorry." Bah. Makes me sick.

Anyway, still trying to calibrate and measure accuracy. I think I'll try doing a known object (scalpel handle with metric measurements on it) to check that.