A sorta DIY 3D laser scanner. You do need your own camera and laser but the software is free for personal use albeit only outputting low res models. Unfortunately Windows only (and obviously not open source).
i tried 'david' with a line-laser and a video-webcam - it works, but the quality is highly dependant of the webcam and the ambient light.
When i'll get a highres USB-camera then i'll try anew with a green laserpointer and a line-filter or better an old DIA-beamer and the 'inverted' option, where a thin wire in a frame is projekted over the object.
I think the shadow of a string in the beamer is more accurate than the line-laser which has nearly 2mm width and didn't have the same illumination over the length ...
Well, paint my 100Euro mouse and than dip it in water to remove the paint ...
let me think
Most parts I like to have scanned are not paintable for various reasons,
so that is no option in most cases.
I scanned some putto with very ugly results and large holes in the resulting cloud, although I scanned 8 different angles the best I could.
Okay, putti are hard to scan because of numerous undercuts, but that was the only white'ish non glossy object I had laying around (sitting that is )
scanning small parts is almost impossible using a shadowline Viktor.
To have a sharp lineshadow you need to have the pointiest lightsource you can find, regarding the aperture of the camera the thread will have to be too far away to draw a line smaller than 1mm I guess.
Even the backlight of some tft (which is in fact a very nice lightsource to use for photography for example) draws diffuse shadows. sharp enough for galanty shows, but not sufficient for this task I guess.
Nevertheless, give it a try, I wish you luck.
Using a beamer sounds like a good idea, but remember they're made to make sharp images on a FLAT surface...
... i'm tinkering around with the idea to use fibercoupled lasers as light-source - then i'll have a 'spot-source' of 100 microns in red/5mW or 60 microns in NIR/1000mW (but here i should remove any IR-filters on the camera-chip).
With inserting my green laserpointer in a glas-fibre i'll have a green spot with 125 microns and something like 0,3 mW.
With a selfmade collimator of 40mm diameter i'll have a 'beamer' that could project a 0,3mm wire as a shadow of maybe 0,5mm i hope ...
But if you want to be cheap, then I would recommend doing this: Take lots of pictures from the same angle, on a rotating frame. Maybe even put a cross on the platform. Then you open google Sketchup, and create points on the pictures you took. Its also something you can do.
(now registered, but brandnew so please pardon)
Just some notes on DAVID:
- Since version 2.0 the scanning step result of DAVID can be saved only at a camres of 320x240 in the free version (every pixel hitting the object gives one vertex). So to get a higher resoluted mesh you
Edited 3 time(s). Last edit at 01/12/2009 12:56PM by MagWeb.
Every colour is a separate scan, that is aligned and fused to an average point-cloud.
I'm testing with some other concepts and optical modifications, so don't await such qualities from me in the next days ... but this should be the quality you can expect with normal/common SL-scanning ...
For better controlling the angular positions I'm using a pan/tilt-disc, as common for macro-stereo-microscopy use ...
Edited 1 time(s). Last edit at 09/28/2012 05:12AM by VDX. Viktor
This is very much on my to do list. I'm trying to wait until the holidays to see if they run a promotional price again. On the other hand, I have a couple minor projects I need to model parts for that don't lend themselves to measuring with calipers.
Your results are inspiring. Compared to RepRap, is it more or less difficult to get good results ?
... I've tested with reducing the image size for smaller objects and higher accuracy.
Without a reducing lens the closest distance is 650mm and the smallest image size is 250x190mm:
With different lenses I received different setups - the smallest usefull for normal camera-scanning give a closest distance of 50mm and smallest image size of 20x15mm.
But here I have to reduce the intensity of the light drastically to avoid overpowering the camera:
With my high-DOF microscope cams the smallest image sizes are around 0.8x0.6mm and maybe 1mm DOF - could be really interesting to test, if the beamer image can be reduced to a corresponding ratio and low enough intensity to SL-3D-scan with microscopic parts ...
The funny thing is that it's basically a MendelMax with the top chopped off and a few optics. We control the stepper motors with the RAMPS/Mega/Pololu stack (a bit overkill) and we position the object using the TrinityLabs monorail and GT2 pulleys and belts.
We took an off-the-shelf window lift motor (mass-produced and cheap worm gearbox), hacked off the DC motor and attached a stepper motor using an aluminum motor coupler and a 3D printed adapter. This give us enough torque to rotate very heavy, unbalanced objects with ease.