Hello Guest it is October 15, 2019, 03:19:01 AM

Author Topic: video probe, tool or toy?  (Read 15560 times)

0 Members and 1 Guest are viewing this topic.

video probe, tool or toy?
« on: August 23, 2009, 11:56:08 AM »
I think we'd all agree that a good working laser 3d probe would be an invaluable tool.  There are some commercially available units out there that seem to do a fine job but at a high price.

I applaud the efforts here towards getting a functioning tool.

My question is this. At this point in the development has anybody been able to actually scan an object and end up with a usable tool path?

Has anyone actually milled a duplicate of the item scanned with satisfactory results?

Again I think the laser scanning plug- ins are very cool but are they a useful too at this stage?
Re: video probe, tool or toy?
« Reply #1 on: August 23, 2009, 06:26:15 PM »
At this point in the development has anybody been able to actually scan an object and end up with a usable tool path?

Has anyone actually milled a duplicate of the item scanned with satisfactory results?

The purpose for the optical and the mechanical probes is to gather data on the target. The optical method is faster and collects a lot more data but in the end you just have a point cloud data set.

Because I have an optical background I have focused on the optical methods. First to make an inexpensive probe then to write some usable software while taking advantage of Mach's motion control.

I too would like to see where the data goes next. I am sure this part of the problem has already been solved for the folks that use touchprobes.

To me, though, it seems that the next step is to read the points into CAD software where a human operator can fit lines, arcs, and surfaces to sections of the data set. Then, as with any other finished drawing, create the Gcode.

Except for showing off at a trade show I doubt that you can just scan to create a point cloud then spit out Gcode to do each point. Point by point execution would be painfully slow.

Anyway, I think that probing the minds of the touchprobe users might yield some useful info.

BTW, there are examples in this Video Probing Forum of things made using data from an optical scan. One that comes to mind is a giant boot used as an attention getter in front of a camping or climbing store.

Tom Hubin
thubin@earthlink.net 
Re: video probe, tool or toy?
« Reply #2 on: August 24, 2009, 03:22:43 PM »
I am interested in this process and have gone as far as ordering a line laser and webcam that I belive will be suitable.... hopefully anyway :-)

From reading other posts on the subject, the laser scanning process to create point clouds seems to do reasonably well. It is not tremendously difficult to go from point cloud to and STL file, which can be loaded in to quite a few CAD/CAM applications and machine code genereated from this.

I have used a lot pre-designed 3d models in STL format to machine 3d objects. I have also taken a few point clouds (from touch probing a 3D surface) and turned them into STL objects. So the remaining piece of the puzzle for me is to see how well this laser scanning process can create an accurate point cloud. For my purposes, if I can get +/- .005" (five thousandths of an inch) I will be overjoyed, but even if it's only within +/- .050" I will be extremely happy with this essentially free plugin, and even in contributing to the project in any way I can. From what I understand this goal is very attainable.

I don't know much how well this whole process will work in my case, but for a true production environment, I think that maybe an estabilshed commercial product would be more reliable, but at a very large cost. If this plugin works, then those semi-professionals like myself with limited budgets can have a foothold into 3D scanning which otherwise may not be possible.

For any who may be interested, my use for this 3D scanning process is in recreating asthetic components such as firearm stocks that will tolerate quite a bit of variation in specs because final finishing would be by hand anyway. The 3D Scan will hopefully allow me to quickly rough out a stock which would be finished by hand.
Re: video probe, tool or toy?
« Reply #3 on: September 07, 2009, 09:59:52 PM »
I too am looking to collect accurate date as far as shape is concerned. I'm a sculpter and work in the human form. Presently I take a cast from a live body with plaster and other materials and produce a usable mold much the way it was done thousands of years ago.


I have a couple of three axis mills I've built over the years and I've build a pretty good touch probe to go with it.

My life would get a lot less complicated if instead of smearing a model with plaster and such for more then an hour or two at a time I could sweep then with a laser and essentially machine a body-double from foam in a very short time. An hour modeling session could yield a dozen scans.  An added plus to the model is that no one has to touch them or get messy stuff stuck to their boobies. (pretty clear advantages here, from the models perspective anyway)

I see a lot of OK scans of kick knacks and such but seriously it wouldn't be too hard to make a mold from a plastic statue.

Like they say of kittens and cats, Kittens are one thing, cats are a whole different ballgame! Such is the difference between scanning a miniature garden troll and a living human being.

Lawrence
« Last Edit: September 07, 2009, 10:04:21 PM by Tristar500 »
Re: video probe, tool or toy?
« Reply #4 on: September 08, 2009, 02:45:21 AM »

I have a couple of three axis mills I've built over the years and I've build a pretty good touch probe to go with it.

My life would get a lot less complicated if instead of smearing a model with plaster and such for more then an hour or two at a time I could sweep then with a laser and essentially machine a body-double from foam in a very short time. An hour modeling session could yield a dozen scans.  An added plus to the model is that no one has to touch them or get messy stuff stuck to their boobies. (pretty clear advantages here, from the models perspective anyway)

I see a lot of OK scans of kick knacks and such but seriously it wouldn't be too hard to make a mold from a plastic statue.

Lawrence
>

Hello Lawrence,

I don't see any technical problems with scanning a human body.

The method that I use is to scan using something like a 3 axis mill with the spindle replaced by the scan device (camera and laser and optics). So, for scanning a human body, you need a three axis stage that is large enough to scan a human body. Let's say a 600mm wide by 2000mm length by 300mm depth.

To make specific recommendations I would need to know the accuracy requirements and maybe the timing requirements since a live person must be relatively still during the scan.

Accuracy on the order of 1% of the camera field of view is reasonable to expect. So if you want 1mm accuracy then you are pretty much limited to a 100mm field of view. Since a human body is probably 6 times this width you would need to make six passes, moving the scanner over 100mm for each successive pass.

Each lengthwise pass could be 1mm steps for 2000mm length.

If you need finer detail then you could use a scanner intended for finer accuracy and make more passes to cover the area.

Keep in mind that the Z field of view is also limited. So multiple Z axis scans would be necessary if the model is well endowed.

To get all sides of the model you would have to role the model over and try again. The multiple scans would have to be assembled with CAD software that can work with point cloud data.

If you are serious about making this convenient then consider a scanner mounted to travel around a standing model or a model reclining on something transparent. This would use x, y, z and theta for axes.

Say more about how you would want to do this and I will try to come up with more specific suggestions.

Tom Hubin
thubin@earthlink.net 
Re: video probe, tool or toy?
« Reply #5 on: September 09, 2009, 12:26:25 AM »
H Tom,

First I just want to make sure everyone understands I am totally behind the develpment of technology on a grassroots level as you have so well pioneered. I hope the title of this subject wasn't taken to belittle yours or anyone else s efforts. 

OK, that said,

When making a scan, cast, etc or a human form it is important that the model be posed in the same fashion as the final piece will be displayed.  This it true with the human face and perhaps more so with the female torso.  Gravity effects a body and the subtle changes in the geometry will be appartent in the finished sculpture if it was not cast in the same postition.

I'm thinking of constructing a lightweight machine  probably belt driven to scan the human subject. The model would be in most cases standing so the machine would resemble a mill on its end.

Does this make sense?

Best. Lawrence

Keep up the great work Tom!

Re: video probe, tool or toy?
« Reply #6 on: September 09, 2009, 05:08:31 PM »
Hi Laurence,

You mean scanning a live human body or a cast made out of it?

I've been quite successfully scanning forms musical instruments - violins, violas and cellos. The resulting point cloud is processed to a g-code using software like VX Cad Cam or similar. I'm not reproducing my scanned instrument parts exactly as I prefer to do the final shaping myself, but I use my router as an "assistant" doing the rough work for me. The accuracy would be enough to make quite a finished product if that is what one desires... I've been using both Art's and Tom's video probe plug-in. With good adjustments and patience you can make a reasonably accurate scan with either of them, but after finding the proper setup Tom's probe is much more practical.
In the last month I haven't had time to go on with my experiments, but the last scans were already very good thanks to Tom's help.
Scanning a live human body would take quite a while. A cello back arching - an area of c. 450m x 760mm - with Y step 1mm and X step 60mm takes roughly an hour. And luckily for the accuracy, the cello back is a dead, immobile object...

For your purpose, I think this could be interesting: http://www.david-laserscanner.com/

Good luck,
Eero

Re: video probe, tool or toy?
« Reply #7 on: September 09, 2009, 08:05:06 PM »
Hi Eero,

Yes, scanning a live person. not a cast.  I'm looking to machine a "mannequin" from the scan of a warm body.  I had no idea Art and Tom's scanner plug's would require that much time.

I've looked at the David system and am undecided..  I need more convincing that I'd have a working instrument before I'd spend the money on the software and hardware.  Most all of the examples I've seen scanned are miniature statues etc...

Good luck with your musical instruments, sounds very intriguing!

Best, Lawrence
Re: video probe, tool or toy?
« Reply #8 on: September 10, 2009, 03:35:30 AM »
Hello,

Timing depends on design resolution and motion speed. Also a little on computer speed but there are ways around that if necessary. Axes are arbitrary and can be setup any way that you like.

So lets say that you want to scan 600mm wide by 2000mm length by 300mm depth and you want 1mm resolution.

I would choose the optics for 100mm x 100 mm field of view since I expect about 1% of that for resolution. So a single frame would scan a line about 100mm long in the y direction and produce over 100 points, probably closer to 500 points . Then move the optics 1mm in the x direction and collect the next frame. Do this 2000 times to complete a 100mm x 2000mm path.

Then move y 100mm and go back to the start of x. Scan the next 100mm x 2000mm path. Do this six times to complete the 600mm wide by 2000mm long area.

Then move z 100mm and repeat all of the above. Do this 3 times to include the 300mm depth.

So we have stopped and grabbed a frame 2000 x 6 x 3 = 36000 times. How fast can you do that? Once per second is 10 hours. Ten frames per second is 1 hour. That would require that the motion ramp up for 0.5mm then ramp down for 0.5mm then grab and process a frame in 0.1 second.

This, of course, only gets the view from one side. If you want the view from the sides and the rear then you need to do this three more times.

Alternatively, you can have multiple scanners working in parallel. That is more complicated hardware but the live model is not so tired.

As for which system you use, I think the math for time vs resolution needs to be done for each system that you consider. For example, if you only need 2mm accuracy then half as many frames are needed in each axis. That can be 1/8 the time if the motion system can keep up.

What resolution do you think you need?

How fast do you need to complete the scan from all sides?

Tom Hubin
thubin@earthlink.net

Re: video probe, tool or toy?
« Reply #9 on: September 10, 2009, 06:58:18 AM »
Hi Tom,

I've only got three axis on both my mill and cnc router so for now anyway I'm happy with a 2.5 D scan. Any undercuts and the finished product won't come out of the mold. (I'm using simple one piece molds)

I don't think I need really high resolution for most of the work I'm doing.   I mm would be more the enough I would think. Basically If I can scan a living person and get a life-like 2.5D scan that I can machine into a mannequin mold I'm there!

I use the negative mold to produce pieces like this..