Probing/scanning the difficult/inaccessible, ideas & discussion

[henrik_o]

We have been approached by the head curator/conservationist of the local church co-ordination agency, and during a meeting today that was really about other things (display cabinets for bibles) he brought up some novel ideas.

Basically the deal is this: he oversees 14-15 local parish churches most of whom are quite old (we're talking AD 1200-1400) that have amassed considerable treasures in sculptures, epitafiums, and so on. These objects are very often damaged. They miss parts, especially delicate areas of "3D" carvings or sculptures, and in many instances a few objects in an arrangement (such as an altar table) have simply been stolen. There's apparently a big and thriving international black market for this kind of stuff.

For objects of national importance, a total conservation is normally done. This means 2-3 people spend several years restoring an object in absolute detail and using the old techniques: you can guess the price of that.

For parish churches, this is not possible, yet they often have invaluable objects that are damaged. So, the idea he threw out is if it would be possible to streamline the process. Missing parts would still need to be replicated as close as possible to the original, but not necessarily using manual techniques. In a sense, they would be "fillers", newly made objects for the conservator to use to repair damaged/missing parts as the whole object is restored -- today the shop he runs does this (mainly touching up pigments and gilding) without doing anything about the actual shapes and the damage on them.

The problem is that these parts/objects are typically highly inaccessible. Meaning that the original to be used for making a duplicate of a missing part is often itself an integral part of a sculpture or carving. While it sometimes can be moved from its display site, it's still part of a larger object and cannot be detached from it.

The following pictures illustrate the problem.

Here we have some shots of a larger sculptural scene depicting Mary and Jesus; Mary in the form of the enthroned Theotókos and Jesus crowned -- his circlet is missing several of the leaves.


8921

This scene can be physically moved, but the parts can not be disassembled.


8922

The form of the leaves is not very complicated, but overall it is a very delicate problem; a new 'filler' needs to fit the fixture as well as conform to the spacing of the existing leaves.


8923

Another shot (a lousy one) showing the backside of a leaf: a new 'filler' must correspond to these.

This could not be probed. It could, afaik, not be scanned using a stationary scanner. The existing parts are too fragile to make a cast. It might be possible to stage the missing parts with a grid table and photograph it from several angles, then reconstruct it in a 3D modeling package, but this is very time consuming. I don't know much about portable scanners.

The above photos depict objects that can be moved. Many objects can't, and they're not exactly easy to reach. Consider a hypothetical, let's say a broken acanthus on an altar table;


8924

Just getting to that piece would be a serious cost, since you need a crane. Once you're there, what you make of it is crucial because you don't want to haul a crane there twice.

I'm basically throwing this out here to start a discussion. I understand there's probably no immediate answers, but it would be interesting to hear what seasoned users think. We could always go low-tech, but I think the costs involved with that would make it a rare job.

[bill.young]

Hey Henrik,

I've been working off and on with a ShopBotting sculptor who's designing and building all kinds of non-contact 3d scanners, based on the ShopBot software, for scanning sculpture on-site. He probably knows as much about the options for that kind of stuff as anyone and might have some insights...I'll get him to get in touch with you.

Bill

[chiloquinruss]

Years ago when I had a software company I saw some software demonstrated that was developed by a student at MIT. The concept was simple, the code was awesome! I don't remember ever seeing if it made it to a commercial product but that might be a source to start. Anyhow what you did was take several photos from all kinds of different angles. You would then load the photos into the software and then go picture by picture and click on the screen noting that "this point" on pix 1 is the same "point" on pix 2, etc,. You cycle through all of the pix setting point 1, you then cycle back through all the pix picking point 2 and so forth. When you have done this for many points (the more points the more accurate the result) you click on 'model' and the software would go away for awhile and the next thing you saw was a 3d model built from the photos! It did all kinds of very clever things like you've seen done in photoshop like edge trace, and also did raster to vector to build a wire frame, and all of this from photos! If this product ever made it to distribution that would certainly be something to use for this impossible type of work. Good thread! Russ

[Brady Watson]

Henrik,
Digitizing 'full 3D' objects like the ones you show, is no easy feat...any way you slice it, per se. The main issue is that a probe (regardless of touch or non-contact) can only register what it can get access to. This becomes a challenge as there are many undercuts & blind spots that you simply cannot access. There are few non-contact solutions out there under $100k USD that will allow you to 'paint' the surface with a digitizing laser while keeping track of it's orientation. A Polhemus et al will do it, but it is really not practical if there is ANY type of metal anywhere near the subject to be scanned.

You have to ask yourself how much resolution you need & how much hand carving you want to do...or how many hours you want to spend in front of the computer trying to get lower cost solutions to do what the expensive ones will do automatically...It's a trade off either way and points that you should consider. The technology really doesn't matter too much. A touch probe, laser and even photogrammetry require that all aspects of the part be visible from a certain viewpoint.

'Full 3D' is not easy to do, even with very good equipment...but if you have good hand carving and software skills, you can make a lot of things work. So far nobody makes the silver bullet of digitizers...ones that do it all, do it well with high resolution, all the while being affordable.

-B

[radavis]

Henrik,
first let me saw it’s a wonderful thing that you are doing,
I work on old car, some time we run in to part we need to cast or cut with complex shapes, I also work for Boeing (yes the aircraft company defense side) so we try to keep up with the most new and best way to do some thing, we have so lager part on some of our produce that we can not take off or cast so we us this Digitizing probe arm, there are some folk out there that will make you a 3D model of what ever you want using this arm, we just had a project that was so big that we couldn't get it in the high bay, the guy took two week and we had a great model in about two more weeks later. here is a company that make one type of the arm. hope this help.
R
http://www.faro.com/quantum/en/index.html

[cknerr]

There is a laser scanning technique used in industry to make as-built drawings after a project is completed. It also gets used when older as-built drawings are no longer reliable. The scans will show up details down to fastners in a large room. The scanner is moved only once to get a reliable picture and "look around" corners that where shadowed in the previous scan. It takes mere seconds to scan any room.

There is an older type where many pictures are taken from differant angles and the same spot is picked in each photo and lableled. This is repeated ad naseum to build a point cloud. This way is very labor intensive and time consuming. This type of software is the cheapest. If you can get volunteers to help pick points, then the cost drops like a stone.

With the laser scanner, commercial people will drop their prices for the PR they can get from advertising helping a church. If they are in the area of the church needing the scan, then travel is at a minimum. You might get some leads from a company in Florida, G&T Conveyor, that makes conveyors for airport baggage handling systems. They are in Tavares Florida. DON'T mention my name. It won't go well if they remember me.

They can deliver a 3D model or at least a point cloud that you can work with. They are in a very small town and are helpful to the comunity.

The file they will give you will be very detailed since they can get close to it. 50 feet is considered being on top of something btw!

Best of luck,
Chris

[hespj]

"Anyhow what you did was take several photos from all kinds of different angles. You would then load the photos into the software and then..........would go away for awhile and the next thing you saw was a 3d model built from the photos!".......Russ

A bit like this?:

http://www.photomodeler.com/applications/architecture_and_preservation.htm

"PhotoModeler software is widely used as a measurement and modeling tool in architectural, preservation, conservation and cultural resource management applications for:
Documenting and measuring older buildings and structures for conservation and preservation
Generating 3D models for visualization and view studies"

[henrik_o]

Bill Y,

quote:Hey Henrik,

I've been working off and on with a ShopBotting sculptor who's designing and building all kinds of non-contact 3d scanners, based on the ShopBot software, for scanning sculpture on-site. He probably knows as much about the options for that kind of stuff as anyone and might have some insights...I'll get him to get in touch with you.
That sounds awesome and a very good fit with this particular problem. I've talked to an 'old school' carver/conservator but while interesting (the problem as such isn't exactly new and pre-dates hi-tech solutions) it didn't give much on how to 'streamline' the process, which would be necessary given budget constraints.

[chiloquinruss]

John - You are quite right, that certainly looks like the software I saw years ago (must be, they are on version 6)!
Thanks for the post on this software - I have need of that type of software for a different set of projects I'm working on. BTW MIT develops all kinds of experimental software projects and then licenses the patents to companies like EOS (and mine) and the royalties fund the cool stuff they do at MIT. If you ever have concerns about the youth of America take a tour of MIT and be prepared to meet some really smart kids!
Russ

[hespj]

Russ, let us know how you get on if you do use it. It's been one of those things which I've been wondering about for years.

[Brady Watson]

PhotoModeler is good for a lot of non-detailed parts. It is horrible at trying to create a relief from a photo...in fact there isn't software out there that will create 3D from a photo in real 3D. Some will do low-level conversions similar to a texture, but not true 3D without a LOT of user input - which is the case with PM. You have to click on each part of the photo (only ones taken with a tripod or properly set up will yield good results) - and stitch them together. You can do the same thing in Rhino by tracing out perspective views and create 3D from the cross sections...PM is a neat program, but make sure you know what it can & can't do - and it isn't cheap either. I believe around $3,000 or so.

-B

[hespj]

Ouch!! Thanks Brady, I can cross that off the list.

John

[chiloquinruss]

Ouch Indeed! I was thinking that if all it did was give you a good 2d that could then be manipulated in some other software, but . . . .
As good as the probe works how about making a mold in clay then casting the part in plaster and then probing the plaster part, would that work? Russ

[hespj]

I just got details of this by email (via Rhino).

Rhinophoto

Rhinophoto is a Photogrammetry plugin for Rhino and provide an automatic 3D digitizing from a set of photographs. You can now digitize in 3D with your Camera...

The price is not yet defined, but will be at the Rhino plugins price level. Anyway the price will not exceed 495US$.

http://www.rhinophoto3d.com/

Looks okay for simple shapes.

John

[henrik_o]

I just wanted to pass quickly by and say it's great to see this discussion take off -- I don't want it to be tied to my specific situation so an open exchange of ideas is great.

I've been around some churches today trying to get more cases on camera, to get a better feel for the range of objects involved. I've also started looking at the suggestions for technologies listed in this thread and it's a bit overwhelming.

In any case, this thread is a very good start, thanks everyone.

[hespj]

Rhinophoto to be released early Dec.

Price is to be about the $495 they promised .......... for a couple of weeks.

The regular 895 euros ($1150) puts it way out of my league.

John

[hespj]

I emailed my concern about the price to RhinoPhoto and they replied immediately telling me that the 895 euro software (offer price 395 euros) about to be released is a more sophisticated version. The 495 euro level 1 version will be available in March.

I should pay more attention.

Price list:

http://www.rhinophoto3d.com/index.php?option=com_content&view=article&id=82&It emid=75

Offer:

http://www.rhinophoto3d.com/index.php?option=com_virtuemart&page=shop.browse&c ategory_id=6&Itemid=78

John

[Brady Watson]

John,
Be careful...photogrammetry, for the most part, is only useful for large objects with relatively smooth surfaces. It is not at all in the same league as laser digitizing, nor is it 'as good' as the ShopBot digitizing probe for most surfaces. You will not be able to successfully digitize a 3D relief carving using that software...so unless you plan on digitizing very large objects, such as cars, boats & planes...save your hard-earned cash. It's a 'cute' program, but not very useful for most Botters. Also, bear in mind that there is a time factor involved with doing each photo, and more importantly, each target. You will need to stick targets all over the entire surface of the part to be digitized. The Rhino link shows actual pencil lines going down the hood, indicating that the targets need to be in a straght line, not just arbitrarily placed. This reminds me of the steps needed to run a digitizing arm, like a Faro or MicroScribe...which take coordination to use & are anything but automatic.

-B

[myxpykalix]

Henrik,
I have 2 different systems i use, and I am by no means an expert. I don't use this for tons of work but what i have done has turned out pretty decent.
I have an Inspeck and a Minolta non contact 3d digitizer. I have used them in the past to replicate missing parts for furniture and to replicate crown moldings in 100 year old homes.
If you are able to get a scan but don't have the means to manipulate the object to your liking there is a company that will do that work for you that i have worked with in the past:
www.directdimensions.com (http://www.directdimensions.com)

[woodworx]

what about taking molds from different angles of the subjects using plaster (or similar) that won't leave residue behind or damage the structures. You could cast the new models or probe them in separate pieces? Just an idea.

[hespj]

Brady, thanks for the warning, I would certainly have a good look at it before purchasing. In fact I wasn't thinking of using it with the ShopBot, I do a lot of CAD work for making ornamental metalwork, and this often means 15 hour days visiting London to measure steps or whatever. If I could just send somebody up with a camera, or even post a camera to somebody in London......

Obvioulsy it's not quite that simple, but I do like to keep my eyes open for things that will increase productivity.

[Brady Watson]

No problem, John. I've tested ALL of the technologies (and all of the branded hardware) listed in this thread, and each have their own 'snafu' depending on what your goals are. Photogrammetry may be a viable solution for you if your main focus is ornamental ironwork - especially since the scale tends to be large enough to get good results with that technology, and enough wiggle room in terms of overall accuracy. Since the thread started out talking about RLF carvings, I assumed that you wanted to do the same.

Jack,
DD is a good outfit, but my company beats the pants off them for scanning service prices on 3 & 4 axis decorative scanning. (The types of things that most SB owners need) I'll let those outfits handle the headaches associated with turbine blades and other high accuracy mechanical parts...Send me the pretty stuff!


-B

[jim_stadtlander]

Brady-------
I have tried emailing you a couple of times at Ibuild with no success. The email does not come back to me (so I assume the email is right), but have not received any response back. I am interested in the possibility of having items scanned once in awhile. Send me an email if you are interested.

Jim

[Gary Campbell]

Jim...
You might have done what I did... try taking out the "U" and make it "ibild"
Gary

[Brady Watson]

Jim,
You have mail. For those who may have tried to email me, it is easiest to go to the website www.ibild.com (http://www.ibild.com) and email me from there. I generally answer emails within 24hrs, so if you don't get a response, check the address.

Thanks!
-Brady

PS - There's no 'U' in IBILD.

[cutitout]

here is one I have played with -- not to bad if you can get the calibration map near the object to be scanned -- and its Free

http://www.david-laserscanner.com/