Re: [rp-ml] algorithm&implementation for STL slicing-SplineScan

From: Charles Overy <>
Date: Mon Aug 28 2006 - 20:18:56 EEST

Sebastien, Markus

Thanks for doing the "google" for me! Oops. I had just never seen one
of these things but had seen various hacks and the parts etc.
The links do seem to indicate that, as one would expect, it is not
trivial to retrofit using the existing systems.

Re: Markus's email:Did you consider how the bearings and actuators of
your CNC machine feel if you let the table run full left - full right
half a million times?

Yes, that is why I assumed any practical approach would use a grid/array
or other way of sampling a relatively large area of the surface at a
time and would then merge those patches. I had assumed it would be a
Move, stop, scan, move, stop, scan type of approach so that the motion
of the mill could be reduced or eliminated. The scans would be aligned
using common features from overlapping mesh surfaces. From
Sebastien's links it appears that what most current applications do is
to use the mill to scan a stripe over the surface of the object in a
continuous fashion. It would appear that the weakness here is that the
scanner must control the mill fairly carefully.


Sebastien Bailard wrote:
> They are relatively common. From a quick search on "laser scanning head cnc
> mill":
> Many links:
> -Sebastien
> On Friday 25 August 2006 13:51, you wrote:
>> I keep wondering why someone does not adapt a project like this to make
>> a non contact scanning head for CNC machines. Off the cuff it would
>> seem that you could create a hardware head that would sit in a collet
>> holder and software that would output generic G-code to scan an
>> arbitrary sized area. Is it more difficult to patch meshes where the
>> scanner is moving rather than the part?
>> Charles
>> Sebastien Bailard wrote:
>>> On Thursday 17 August 2006 19:01, you wrote:
>>>> At 4:25 PM -0400 8/17/06, Sebastien Bailard wrote:
>>>>> For raw 3D structured light scan to stl , there is some open source
>>>>> software called . (This is the opposite of
>>>>> what you are asking for, as far as I can tell.) It is GPL'ed python
>>>>> code.
>>>> Hi Sebastien,
>>>> Thanks for all the info! Splinescan looks interesting and fun - I may
>>>> have to give that a try.
>>> Do. If you need support, you'll have to get it through the splinescan
>>> mailing list. The lead developer's working on development, not
>>> documentation just now, but someone should answer. I have no idea when
>>> they're going to publish their next release.
>>> They've got some interesting tweaks in the pipeline - they're going to
>>> publish a hardware package that you can RP, so all you have to do is
>>> assemble it, bolt on a camera, laser pointer diode, cylindrical lens
>>> (glass rod), and you've got the scanning unit. For the turntable, just
>>> go scrounge a turntable and have your laser printer print out a paper
>>> encoding strip*, paste it on, and you're good to go. The camera/python
>>> code figures out where the turntable is by looking for the encoder strip.
>>> I've been daydreaming that all you need is a sheet of paper printed with
>>> artoolkit tags and a camera, and you could put an object on the sheet,
>>> take loads of photos, and then dump them into gpl'ed code that figures
>>> out all the computational geometry and spits out your point
>>> cloud/simplical surface/.stl file. It would take some keen programming
>>> and computational geometry, but it would be so useful.
>>> If you happen to be interested in the computational geometry background
>>> for all this, look at "Surface Reconstruction from Unorganized Points" by
>>> H. Hoppe 903 citations, and fairly readable,
>>> +unorganized+points&btnG=Search
>>>>> Roy - what materials are you working in? You may be able to use our
>>>>> extrusion head along with the control software:
>>>>> sio nHead
>>>> I've been aware of Reprap but didn't realize or look closely at the
>>>> software part of the project. have to take a closer look. My materials?
>>>> Well... this is a 'for art' project so I have been looking at various
>>>> materials which flow and then harden - slurries of clay, etc. I was
>>>> thinking about wax but wanted to avoid the complexity of a heating
>>>> element.
>>> I would go with kaolin slurry or store bought "earth clay" slurry or slip
>>> as well in that case, or cncmill/RP a reprap extruder head and use
>>> thermoplastic. It'd be a lot easier to set up.
>>> Waxes can be nasty - one tuft of cat hair in your feedstock, it clogs,
>>> and unless you've got a thermostat or thermal-overload detector and hard
>>> shutdown circuit, it all goes up like a candle and you kiss your studio
>>> goodbye. So many artists die just because they forgot to use a double
>>> boiler, crockpot, or soup warmer to heat their wax, so they just put it
>>> directly in a pot on the heating element, maybe go and answer the phone,
>>> the bottom layer liquefies and starts to boil even though when the top's
>>> solid, then you get a big cloud of vaporized wax and poof! Wax is the
>>> main reason art foundries and studios burn down, so you'd want to trouble
>>> shoot the hell out of any autonomous 24/7 mechanism you're building that
>>> uses wax. A slurry printer screws up, and all you've got to clean up is
>>> a puddle.
>>> One thing to realize is whatever earth clay you go with is possibly going
>>> lose all its plasticity if you run it through a feed system - it depends
>>> how happy the bacterial colony is in the clay, after it's been extruded.
>>> (Bacteria makes the gel that lets the clay platelets slip past each
>>> other.) But a couple spritzes with dilute vinegar could perk it up again/
>>> If you're not going to work the clay after extruding it, this doesn't
>>> matter.
>>> You might be able to use a good oilclay (aka plasticine, modeling clay)
>>> instead of wax. Oilclays liquefy when heated, but I don't think they can
>>> burn.
>>> One problem with oilclay is you can't sell it as a finished piece without
>>> going through the casting/moldmaking steps. Wax or thermoplastic you can
>>> do lost wax casting with, slurries you can fire in a kiln.
>>>> For a 'print head' I have been tinkering with building a peristaltic
>>>> pump - the idea being that with that I can deposit a measured amount of
>>>> material that is fairly viscous and it seemed a simple way to do it.
>>>> It's all in the testing/failing stages now.
>>> Sounds fun. Did you see
>>> Think you can clip one onto your 'Drip Painter' machine?
>>> Apropos of nothing, we're using wood filler for a test material. It's
>>> pretty much wood glue (polyvinylacetate) and marble dust.
>>>> I'm assuming (and hoping) that this will be a functional "expressive" 3
>>>> printer - along the lines of Giacometti type forms. I think I can get a
>>>> fair degree of precision in placing the material- and the material will
>>>> slump and distort in unpredictable ways. The art/sculpture "part" of
>>>> this is both the machine as it's working and the forms it creates.
>>>> --Roy
>>> It's an interesting approach. Sort of like manipulating the hell out of
>>> the negative in the darkroom, where I'm stuck on photorealism/toolmaking
>>> - I want to image an oilclay model or broken machine part, work up the
>>> virtual image, print it, and have a perfect duplicate.
>>> How are you planning on creating your input code - c code that bit bangs
>>> the printer port and the motors do their thing, traditional cad cam -> g
>>> code -> emc (linux machining controller - ), or something
>>> esoteric whipped up in processing? (algorithmic art / data analysis tool
>>> with thriving community - ,
>>> ep/ )
>>> Regards,
>>> -Sebastien Bailard
Received on Mon Aug 28 19:48:24 2006

This archive was generated by hypermail 2.1.8 : Tue Jul 21 2009 - 10:27:52 EEST