Using scanners to capture images of the foot for orthoses construction

http://www.asicsamerica.com/asicstech/windlass_mechanics.htm
http://www.podiatry-arena.com/podiatry-forum/tags/index.php?tag=/windlass-mechanism/

 

I suspect that Joe is probably your man here. But I'd be interested in getting involved too, for the European end.

 

Jeff, I agree with your comments above, but I guess if we were to use a first ray cut out this may help in reducing dorsiflexory forces on the first met from the orthosis shell, also a 2-4 valgus forefoot extension may help. Dave, how do you get around this using the AMFIT scanner?

 

Jeff

This is a problem of resolution and range and in principle I agree with you, this is one reason why high flanges and deep heel cups are not really possible with the Amfit system. Since the pins have a limited vertical displacement they cannot capture data that requires greater displacement than 30mm. The next limitation related to range is that the blanks are only 30mm thick so even if you could scan a vertical displacement greater than 30mm you couldn't mill it anyway.

Fortunately, in reality the resolution is not such a problem since the curve of a heel or arch or whatever is fairly predictable in terms of interpolation and is unlikely to change shape by very much within the difference in displacement of two pins. Even if it does the soft tissue can attenuate this discrepancy in the charactreisation of the curve.

So you could look at the displacement of shape as the magnitude of signal and the rate of change in shape as the frequency of signal and the pins or data points as the sample rate. As long as you can estimate the maximum frequency you can be fairly sure of the sampling rate estimate giving a reasonable characterisation of the change in shape. How often will the real frequency be far outside the estimated one? Not often by my experience.

The Amfit pin array has 8mm between pin centres and I would guess that Amfit have estimated that this is a usefull resolution and that thinner pins closer together may be more prone to damage and to cause trauma to the skin.
I have seen that another manufacturer of a simmilar scanner does have a greater pin resolution (can't think of the Name at present) So I guess its a case of designer discretion. Eight pins / squ centimetre will give a better characterisation of the foot shape than 4 pins / squ centimetre but will it be any more useful.

The problem is you can't predict where the most data points will be required when you design a machine so you would have to settle for your estimation of the optimum / most useful number. More data usually also means more headaches in the collection and processing and expense, in both finance and time. And at the end of the day is it any more useful?

That's a good point about orientation and reference however the Amfit has the advantage of a built in reference system ie the box you stand on.

Quite

All the best Dave

 

So why aren't the vast majority of scanners using multiple cameras? I don't want to keep going back to the past, but the scanners I worked on "back in the day" used 4 cameras to capture the whole foot, if memory serves- it may have been two- I really can't remember. Is it so difficult to write the software to integrate data from multiple cameras? Clearly the hardware cost isn't the limiting factor...

 

Love the way the bottom link says "ass-mechanism"

Bottom link... get it? Whatever... Simple things please simple minds.

"Please stay a child somewhere in your heart"- original of the species- U2

It's U2tastic at moment- sorry!

 

One system I worked on used 2 cameras, but not for any other reason than stereovision, meaning that the cameras were the eyes. For a LPi Camera/laser plane based design, only one camera is necessary, as the other eye is the laser itself, effectively. As to other surfaces, I see it as possible to capture a volumetric scan of the foot, simply by projecting the laser off of mirrors, creating a complementary plane and anterior surface scan, but it is significantly more complicated.

v-mirrors- v
/ foot \
|
|
laser
camera


The drawback to this concept is that the scanner height would become an issue, the projected plane relative to the camera would be more complicated, but accuracy and performance probably would not suffer much as a result. Also calculating an optimal reflection angle would need to allow for latitude when it comes to non-weight bearing implementations, and would probably increase overall cost and maintence issues.

Using multiple cameras means extra time for scanning from multiple sources and additional potential failures, and practicioners don't want a device that is slow. I've seen one scanner that used 8 stationary cameras, and to me that was an attempt to create a low profile device and suffered from ambient lighting issues. That unit was over $20K in cost.

 

Too much karate movies or slam dancing in your youth? ;-)

 

I have no problem sharing ideas with other companies, even competitors. While it isn't often reciprocated I see my participation as win win. Building a 3D scanner is easy, especially with bar lasers. If one understands a ccd pixel as a 3D line, and the laser as a plane, the rest is simple identification of camera characteristics, laser plane relative to camera, and about 5 lines of code, coupled with assorted filters. The real trick is implementation, of which, I'm now in my 3rd revision, ergonomics. There is a tradeoff with a single camera implementation in terms of accuracy of the profile, and making it user friendly for the doctor. The next version of my device will be 12 inches high, and probably slightly uncomfortable for a doctor to have it between him/her and the patient, but a scanner that has an extremely low profile is most likely a simple page scanner, and not the direction I'm thinking appropriate. Most likely, the doctor or technician will still have to be off to one side. Most important I see of all of this now that some level of maturity in my product is acheived, is to make the system as friendly as possible for the user, making available every possible casting method a technician could use, naturally of course, that is biomechanically correct for casting. Leaving the business end out of this forum, I don't see why more people can't be involved, discussing what sorts of implementations they would have on their wish list, and hopefully even competitors would be willing to participate also.

While I already have developed a complete fabrication system in place for manufacturing of orthotic devices, I'd rather share concepts and get the slackers up to speed on what needs to be in place to produce orthotics than watch our overall product quality degrade. Even now, one competitor made it into my network on at least one customer site, busted up a jig, and isn't anywhere close to ready to produce prescription devices, conventionally speaking, and yet at least one lab in the states is using them for prescription devices. I've already created one thread on cast corrections, focusing on the rearfoot, and as it develops, hope to cover all aspects of orthotics manufacturing, including any and all concepts related to providing a quality product to the end user. It would be nice to see professionals from both camps participating (Labs and podiatrists).

 

Jeff, Simon

Yes a NWb POP cast can capture a forefoot deformity and commonly that is a forefoot valgus. This can of course then be balanced intrinsically and the resultant orthotic may assist the function of the 1st ray/MTP

But taking a POP cast takes time and cast modifications take time. The question is do you need to capture the deformity be able to add the valgus post? Semi-weight bearing casts eg foam are oft criticised for the obliteration of a flexible forefoot valgus but does this really matter? As long as you note it during assesment and add posting (extrinsic) accordingly is there then any difference?

I would think it the same on the AMFIT

 

Lawrence,

In short, the answer may well be no. There are many ways to skin cat's. However, do those methods really result in the same outcome? If the interaction between the foot and orthosis is significant, and if the mechanics of that interaction are, in part, determined by the mechanical behaviour of the orthoses under load, then I believe the differing methods would result in different mechanical properties of the resultant orthoses. The real question is, how do these differences influence outcomes? I don't know.

 

If you take a nw cast of a foot with an everted forefoot condition (ie ff valgus or plantarflexed 1st ray) and place the positive model with the heel vertical, you will see that the entire lateral column back to the cc joint is elevated off of the supporting surface. If you cast the same foot in biofoam, the medial column will be dorsiflexed by vertical grf and the lateral column will not be elevated off of the supporting surface to the same degree as it would in a nw cast. A forefoot valgus extension would not replicate the contour of the orthotic shell, since it is placed on the distal aspect of the orthosis and on the top cover's distal extension. To get the same influence as the everted forefoot captured in the cast, a valgus extension would need to be placed on top of the orthotic shell starting proximally at about the cc joint. It would need to have a valgus angle as well be wedged in the sagital plane with an anterior angle of inclination. We sometimes place wedges like these on top of existing orthoses to increase the valgus support.

A traditional valgus extension is typically used to enhance the valgus support at the distal aspect of the orthotic device, not more proximally as described above. A functional orthosis is a complex triplane support, not just a valgus wedge. This is why I'm a big believer in the value of non-weightbearing casting. As Simon said, there's more than one way to skin a cat. We might be talking about the difference between cats and dogs because these devices might be different animals.

 

This thread has become rather interesting and Jeff's explanation here is excellent regarding POP casting, modifications and FF valgus deformities. Early on when I was learning about orthoses it seemed that every orthosis that made its way into my office dispensed by another chiropractor had two common characteristic (okay well three because many were also Foot Levelers inserts and crap if you ask me ):

1. High MLA
2. Extrinsic forefoot varus post

I pondered why I felt more of these patients had FF valgus issues not varus and many complained of 1st MTP pain and MLA and/or lateral column discomfort. Most were cast in biofoam. This perplexed me until I realized that there were more people with FF valgus foot types out there and that recognizing them was not something that we were taught to do.

Am I wrong or at a certain level of training does everyone see supinatus due to correctable muscle involvement as a FF varus? Owning a lab I bet Jeff can answer this question.

:good:

Regards,

 

Jeff, Simon and David

Definately David the mechanism and orthotic failure with foam cast devices that you have observed is a real phenomenon. In my mind the capture of the flexible forefoot valgus with a NWb POP cast or creation of a (relative) supinatus with a foam cast is the crux.

And I would agree with you Jeff that adding a forefoot 2-5 extension of say 3mm flexible material or valgus wedge under the met heads is not the same as a valgus forefoot post in the shell for the reasons you suggest.

However I have found, with foam cast Rx's, if you use a shank dependent material and extrinsically grind it you can incorporate the degree of valgus posting you want and in doing so have a post that extends more proximally in the manner you so clearly describe.

I have tended recently to make these sulcus length and ground through under the 1st met head. I also add some degree of medial heel skive more often than not. (if your interested!). Shank interdependent materials and foam casts are not particulary consistent and if that material is wanted POP NWb impression taking is probably far more consistent. I find this to be due to pitch/elevation problems

This might be a good time for David (Smith) or Bruce Williams to chime in as they both use the Amfit system and prescribe shank dependent device.

 

Lawrence

I generally use a semi weight bearing scan for most of my orthoses. This allows capture of the foot shape as you might see it non weight bearing but also gives enough expansion of soft tissues to approximate the weight bearing shape.

When i want to enhance a certain feature eg a forefoot valgus or a high lateral arch etc I can increase the pin pressure and for instance manually hold down the first ray while dorsiflexing the hallux. The contours captured are readily visible in the 3D characterisation or the 2D contour map of the foot scan. Using the software tools these shapes can then be proportionally accentuated or minimised as required. Of course infinitely variable wedges and posts, positive or negative, can be added to increase the bias of a certain area as required.

Dave

 

Jeff:

So good to have you contributing regularly here on Podiatry Arena where you can share your knowledge with many others. Your insights here are extremely valuable to us.

I totally agree with Jeff about the benefit of non-weightbearing casting. Being able to see the dorsal, plantar, medial and lateral aspects of the foot during the casting procedure, I believe, is one of the most positive benefits of using non-weightbearing casting. I will also modify the usual non-weightbearing "neutral suspension casting" procedure to either plantarflex the medial column/first ray, dorsiflex the medial column or position the subtalar joint more pronated than normal. In other words, non-weightbeairing casting gives me the option to either increase or decrease the medial longitudinal arch of the resultant negative cast as I see fit for the patient's biomechanical and therapeutic needs. Then, once the negative cast is off the foot, I can visually inspect the three-dimensional contours of both the interior and exterior of the negative cast to make sure that I have captured the shape that I wanted.

Of course, much of what we do in medicine is a product of how we were trained. I was trained in the plaster neutral position suspenstion casting technique by the originators of this technique, which has probably helped me greatly in my practice career. I have been train podiatric students and podiatric surgical residents on negative casting for nearly a quarter century and still think the non-weightbearing method is the best way to cast a foot for orthoses.

 

Dave,

Can you differentially alter the pin pressure, that is can you set some pins stiffer than others?

 

:good: Agreed!

The puzzle to me is that they all may "work". The question then becomes which is the "best" for each given patient?

 

If and until one can compare the cost and medical benefit of these similar but different methologies, that is the key question.

 

This spurs more questions. Any scanner would need to meet specifications that the end user requires to provide a method the practitioner is most comfortable, and the technology used needs to have credibility. To simplify this, let's assume that credibility issues need not be discussed, but rather, that each practitioner would want to document how they cast a patient, according to foot position relative to grown level, supine or prone, technician position and patient position, with enough information to test the method, and then document that they support it or not. For example, many doctors cast in the prone position, with the patient laying on their belly, and the foot locked into a subtalar neutral position, and that they used plaster previously. Obviously, the scanner either supports this method, or doesn't. If the practitioner has to change their practice to meet their needs, I would have to say that the technology was inferior to the in office practice. On this note, I would have to concede that although this is my own favorite method, my product is inferior, but this is only a temporary issue and one that will be resolved quickly.

In layman's terms, if I want that practitioner as a customer, I would have to meet their needs. I also think my reputation would suffer if I claimed to support a method and in reality, either didn't or the implementation of their method was poor.

 

Simon

No, the increase is distributed to all the pins equally. However to get a similar reaction you can stiffen the ray or rays of interest by applying a manual force to them ie hold them down. So for instance you could capture a flexible 5th ray without elevating the 1st ray if you simultaneously hold down the 1st ray.

Dave

 

A couple of things to look for in a potential scanner:

1) The scanner is not proprietary to one lab. Ideally, we need a "universal" system which provides data that can be utilized by any lab willing to accept the data.

It would be interesting to send the same data captured by such a scanner to several labs and observe the amount of variability in the finished products.

2) The scanner can capture a 3 dimensional image of the foot, non-weight bearing while the clinician manipulates the foot to the desired position (eg. subtalar neutral, pronate the midtarsal joint, plantarflex the first metatarsal or any other position that a clinician deems appropriate for a patient.)

The goal of such scanners is to offer a more precise tool for image capture; hopefully not a means to further "dumb down" the process of procuring orthotic devices so that clinicians with minimal training can provide such devices.

One scanner I am looking at is the "Tomcat." Any thoughts or experience with this?
http://www.tom-catsolutions.com/index.html

Ed Davis, DPM

 

My own system is slightly less than that ideal as all agreements in terms of scanners are locked into my own software exclusively, which actually is free, except for royalties (or click charges if you will) that are charged for devices made using the system. When a client contacts me directly, they have their choice of now 13 labs to engage in business with, internationally, which is up 62% from just a few months ago.. When a new client is referred by a client/server site, assumed is exclusivity with that site unless they agree otherwise or amicably terminate their agreement with a specific lab.

Will my software work with any Cad/Cam based lab? Yes, but an initial visit is required to configure a client system there that will make the conversion, usually done in 1 day but allowing for 3. On a positive note, while I have no plans to provide compatibility with other CAD/CAM orthotic manufacturing systems, the working data from my own system can peacefully coexist with any G-Code based 3 or more axis machines with or without ATC (Automatic Tool Changer) configured to produce orthotic devices, so the end result is the same.

I would be more concerned about the care a vendor takes in installing their software to work on a lab's machines and level of responsibility a vendor accepts when something goes wrong than anything else. My last install took less than half of a day and didn't leave any christening marks on any hardware. A competitor wasn't so fortunate with the same machine, and ultimately destroyed a $1300 jig (A workpiece holder) after 3 days of trying to get it working, and from what I am told, have offered nothing in compensation for their mistakes. Thankfully, my customer has other machines to produce with, or it would have put him out of business for quite some time, or would have caused severe quality problems if he continued to run the machine in its current condition.

The results would vary by as many clients as I have, as each lab has different styles and designs and as there are more than 300 variables available with which to work with for any prescription design. Newer clients, however, having prexisting connections to my network often just scan and fax in a prescription leaving all of the design work to the lab to do, but all have the design capacity of their server sites so that as they learn more, they too can have their own style of device as well, dependent entirely on their willingness to learn.

I do regret to say that my software and scanner will produce devices using the MASS style of casting. I prefer to work with sites who use conventional casting methods. MASS casting seems to twist the midtarsal joints from what I'm used to seeing on a longitudinal basis and while I know it's a function of supinating the foot, I'm personally not confortable with this concept.

How the device is casted/scanned will dictate the results. The skill of the podiatric professional is essential to providing a good casting of a patient's foot, electronically or otherwise. Simple to use would imply some level of knowledge of how to cast a patient using conventional casting methods, and intuition about how a patient should be scanned based on those same conventional methods. As a vendor of scanners, my job first with any lab is to make my product match closely to methods they are already using with plaster. I haven't met this objective completely yet, but it wont be long and as the process of manufacturing is automated, customizations go from 3D design to reality quickly, allowing me to show the customer the new design in 3D prior to making it.

One of my customers uses their data as it is a scanner that my software supports conversion from ".raw" format to my own software's format. I'm not personally satisfied with the results, but my customer is able to produce a functional device from their product. It has the look and feel of a page scanner from the videos on their site, but perhaps they have overcome the hurdles of unstructured light scanning. With an obvious bias toward specific technologies, and given the size of their scanners, I'm inclined to think we are using different methods to produce similar results. Also, and as we are competitors, my own bias towards my own products need to be considered, and it would be best if you considered other views as having more merit than my own unless there are supporting statements from individuals with little or no professional bias. In terms of ergonomics, they win hands down and this is an area of weakness that is being addressed as much as is possible given the physical size of my own product.

As for the videos I've seen on their site, they seem quite slanted, and in particular, this one seems heavily so. I haven't heard of lead times from lab to practice exceeding 2 to 4 days typically,and never have I seen a practice act that way in reality unless someone was intentionally trying to act like an idiot. The only real time benefit I see to a scanner is to be able to transmit in minutes what would normally take 1 or more days, and elimination of the cure time of plaster, which is negated with biofoam anyway for those that cast using that method.

Another caviet to watch for is maintenence costs. For my own system, this is modest in comparison but all us who provide hardware and software in the CAD/CAM industry have a profit margin somewhere. For any sale of a scanner, I make a modest profit on the sale, but my real profit is that of royalties, which some competitors call 'click charges'. The unit I make costs about $460.00 to manufacture counting labor, and sells for $800.00 excluding the computer cost. I charge my client/server sites, or optionally the client site directly, $2.00/pair for hard shell devices, and $1.00/pair for soft accomodative, positives and prefab designs, which is all inclusive, covering warranties and tech support, both onsite and off. There is one company that has similar charges, $3800.00 purchase cost (Including computer?) and ranging from $5.00/pair to $12.00/pair dependent on whether the hardware is purchased or leased. Your business model should include initial costs and the maintenence costs. In terms of considerations, ranging from highest priority to low, Accuracy, ergonomics, cost of purchase, maintenence costs, are all factors that should be kept in mind.

 

I asked the admin to either pull much out of the above post because I didn't like how it came out, too arrogant and too much stinking of marketing. While there's much accuracy I felt that it detracts from the topic and was way too self-gratifying and selfservingand just didn't feel right, but it was too late to edit. In advance, if the edit isn't made or the post removed, please accept my appologies for that.

 

I'd forgotten about this one, I'll post some pictures tomorrow of some scans achieved for under £100 using the green laser and a clockwork kitchen timer to drive the laser (£2.65).

The guys at the "David Scanner" site and forum are brilliant http://www.david-laserscanner.com/. As I like to play with little projects like this, I'm currently working on a stepper motor controlled scanner to work with the David software, I think this will be achieved for about another £100- so a 3D foot / cast scanner for a couple of hundred quid!

 

I'm not sure what happened to my response, but here's a good site for a microstepper controller. One caviet, however, is that if the inrush diodes are not installed, they need to be. Here's an example picture of what's needed, noting the stripes on the diodes as they conduct electricity in one direction only, clamping feedback voltage to a range of -0.7vdc~v+0.7vdc. The manufacturer recently assured me that these diodes are again being installed, but it cost one new customer and several hundred dollars worth of failures as a result.

 

Not sure yet, Arduino http://www.arduino.cc/ + shield seems a good option http://www.david-laserscanner.com/wiki/the_open_hardware_project, or this approach: http://www.david-laserscanner.com/forum/viewtopic.php?f=6&t=1124. Joe, you should contribute to the David forum.

 

Well, whatever the method, I've found linear travel to be much more reliable and repeatable than to rotate the laser across an object. RadScan 1 had almost no moving parts, but error is magnified over a long distance, making it difficult to have a known laser angle. My first design bounced a laser off of a moving mirror mounted to a stepper motor, and had my home position sensor been closer to the object to be scanned, it would have likely worked better.



These days, however, having an inhouse CNC designed for machining pretty much makes it unnecessary to consider cheaper and inferior methods. It's pretty obvious how they're identifying the plane of the laser, using the walls of the corner, and very intuitive and innovative.

 

I guess you didn't read through it, scanning without the background can now be done with the David system; the laser angle is measured as the scan progresses. Some people have built linear tracks with David, but the guru's there all seem to use rotation- something about illuminating more pixels? Cost is easy to measure, how do you measure inferiority of methods when it comes to scanning a foot or a cast, if the system works and provides an accurate representation of the target object, how is this inferior? Proof of the pudding is in the orthoses produced- right?

To be honest, I've seen examples of your scans, I have a commercial scanner and I've worked with other commercial scanners in the past, IMHO, I don't think the David scanner is inferior to any of these; it is however, considerably cheaper. But being open source, it is IMHO a potential threat to those selling commercial scanners and click to use systems, particularly when we view this in connection with other open source projects such as meshlab http://meshlab.sourceforge.net/ and reprap http://reprap.org/bin/view/Main/WebHome. Time will tell.

 

By itself, it isn't a threat, but it's really only one component of many required for automation. The math behind finding the laser plane, as I said before is pretty intuitive, using the walls behind the object to be scanned. It is also dependent on knowing the location of the walls relative to the camera, and implies a stationary camera position. This means the camera must be far enough from the object to be able to view the entire object. Much like my RadScan 1 design as shown in the photograph, this means an increase in error due to pixel resolution. The human element also leaves much to be desired, which is why I suggested a source for microstepper controllers. As the camera travels with the laser in my own design, it can't help but be a little more accurate, simply because it is closer to the object being scanned.

 

As I said before, once calibrated you don't need the background planes.

True, the human element is overcome using micro-stepper controllers such as Arduino etc. and are already being used by those using David. Please explain the error due to pixel resolution. i don't think the static camera argument cuts any mustard if the scan is still accurate.

Please explain how proximity makes for increased accuracy. Moreover, please explain how this ultimately provides a more efficacious orthosis?

 

I'm assuming, then, that the calculation of the laser plane is prone to error. It would be intuitive if there was a known combination of surfaces at a known position in real space, but the pitch, yaw, and roll being unknowns could present problems.

Imagine a pool cue, held at one end. Then, pitch it 1 degree, and measure the displacement at the end, and compare it to the midsection. You will find that the further away from the end that is being held, the displacement will be dramatically different. This is why I suggested that I would have had improved results in my first design, had I moved the sensor closer to the object being scanned. Still, even with micro-stepping capability, resolution is much poorer than a linear travel mechanism, for similar reasons.

Done, except you now have 30,000+ pool cues, one set being the camera, and one being the laser or bar laser. Conversely, too close to the object being measured, and accurate measurement of profile data (perimeter) with a single camera becomes impossible.

 

How do you calibrate your scanner?

But you haven't addressed the issue, if the system captures a scan that can be employed to manufacture an efficacious orthosis for the cost of a web cam and a line laser- who cares? Surely you need to demonstrate that the additional cost, provides additional symptom relief?

No I haven't been. Why not do the classic scanning a "penny test" using your system Joe and compare it to the example here:http://www.david-laserscanner.com/forum/download/file.php?id=1639&mode=view if you search the site you'll find other examples. Gallery here:http://www.david-laserscanner.com/?section=Gallery&page=u1

 

Very good question. Here's my opinion on the subject. If you are talking about reproducing a portion of a 3-d object such as the plantar aspect of the foot, then you should not look at the end result. For example, if you wanted to compare the contour of a model of a foot taken with plaster-of-paris, resin casting sock, impression foam, and one or more foot digitizers, how would you do it? The first goal should be to see how accurately these different methods can reproduce a rigid object.

The second issue is how might any of these methods contribute to variability when attempting to create consistent reproductions of a non-rigid object, such as the foot. This is where methodology, technology, ergonomics, etc come into play.

The final question then becomes how might this influence outcomes on average, be it positive, negative, or no significant difference? How close is close enough?

Respectfully,
Jeff Root
www.root-lab.com

 

First, I establish that the line that the laser projects is parallel to the window's edge. As most of the components are precision made, the only real consideration for calibration has to do with the threaded mounting hole for the laser rod. I simply scan a surface and verify the depth difference of the object from the highest point to the lowest. Lately, knowing the cause, I have been considerably more careful about the tapping of this hole (Cplate1 threaded corner holes in my drawings). Because little of the scanner is actually manufactured by hand, precision greatly minimizes calibration requirements, and additionally, Laser angle is derived, the remaining 2 axis x/y are automatically within calibration requirements, simply because of the relationship between the camera mount and laser and the math behind the technology. LPI (line plane intersection) calculations in a linear travel based system are relatively forgiving in terms of repeatability. unlike hand operated methods.

The end user might. If I remember right, a pound sterling is close to $2.00, and of my expenses that involve electronics, in total I probably spend close to $180.00/unit. (power supply, camera, laser, stepper motor and driver) Your 1 to 200 quid is about my manufacturing cost of a complete automated scanning system, at least for materials, as you posted in another message.

Webcams cost $30.00, and lasers cost $20.00, at least from current sources. A complete scanner costs me approximately $340.00 in materials, a couple hours of general labor running my CNC, and about 3 to 4 hours of assembly and testing with slightly higher skills. The difference appears to be more of ease of use and less of the dastardly human element to contend with. From a user interface and customer support perspective, the less the human is involved, the better. While a good majority of the human element can be automated away, even the most sophisticated system often fails the 12 o'clock flasher test. Over the course of the next year, I see the cost of my scanners to the end user dropping below $500.00, which will be typically less than the cost of the laptop to run it.

The penny test would obviously be an inappropriate comparison. A penny doesn't expose the weakness of having a camera too close to the device being scanned, because it has relatively sharp edges, compared to the human foot. Comparing a known surface to real measurements is one baseline test I've done on multiple occasions during the development of my product. The field of view of the cameras I work with seem very reliable in terms of apex height (Distance in pixels from the apex to the ccd array), which mitigates the need to calibrate for each camera of that type. The best camera I've found is the QuickCam STX, which works well in low lighting conditions.

 

Just to keep interested people up to date, the boys at David are now enabling the use of structured light scanning.

More here: http://www.david-laserscanner.com/forum/viewtopic.php?f=2&t=1145