Digitizing physical objects, 2D and 3D

[lto]

Hello,

I have questions on other's experiance with probing/scanning in both 2D and 3D, and what seems to be the better method or technology currently available. I am talking about the situation where someone brings in an assembly composed of 15 to 25 parts, most needing holes drilled, few being rectangular, and many non radius curves. How would you rapidly digitize these physical patterns, to an accuracy acceptable to the average woodworking trade?

The above example includes only 2D needs, What is the preferred method when asked to reproduce originally hand carved/cut architectural details, where 3D machining is needed?

The Shopbot probe is an excellent place to start, and I am glad they provide all they do, to facillitate the creativity of the small shop, but I'm seeking greater efficency in moving older existing patterns into the digital world. I've tried having Staples scan large tracings in the past, but that experiance was far from optimal.

Thanks for your input.

[Brady Watson]

For 2D, it depends on the size. You can use the SB probe to get an outline tracing. You can put the part on a flatbed document scanner and then trace out in software, like PartWorks, Aspire or ArtCAM. For larger parts, a good digital camera, contrasting background (sheet etc), a tripod and several 'dead on' shots stitched together via your camera's panoramic stitching function, then traced in software to vectorize. Many parts can be digitized with a tape measure and calipers & drawing in CAD that way. Other parts require a method mentioned above.

3D parts can be digitized with the SB probe, Roland Picza, sometimes with a NextEngine scanner et al or sent out to be professionally laser scanned at a cost cheaper and faster than you doing it yourself.

-B

[lto]

Brady,

Using the photo method, with the camera focused squarely, can you give me an idea of the tolerances to be expected if 2D replicating an organic shape 1.5" thick x 10" wide x 48" long? I'm just wondering if this will be accurate enough for parts that will be mating to other parts in joints of an assembly.

I'd also like to get a better understanding of how your laser scanning service could help on some projects. To produce a file from a physical object, say the actual size of the side of your head, as used as your avatar, detailed to the level allowing finnishing with a 1/8" ballnose; approximately what would our cost be?

Also, Are you a reseller of the scanning system you use? Having the capability of taking the job from staert to finish in house would allow offering rapid turnover on 3D replications, which may, or may not be worth the investment in learning curve and hardware.

You may contact me offline on this last issue if you'd like.

Thank you!

[Brady Watson]

1)Using the photo method, with the camera focused squarely, can you give me an idea of the tolerances to be expected if 2D replicating an organic shape 1.5" thick x 10" wide x 48" long? I'm just wondering if this will be accurate enough for parts that will be mating to other parts in joints of an assembly.

No, I can't quote accuracy unless I am the one doing it. There are too many things that will influence accuracy, including how much contrast there is between the background and subject, lighting angle, how square and parallel the camera is to the subject, and how experienced and good you are at manually tracing and scaling bitmaps to vector artwork. There are some things that only come with experience.

2)I'd also like to get a better understanding of how your laser scanning service could help on some projects. To produce a file from a physical object, say the actual size of the side of your head, as used as your avatar, detailed to the level allowing finnishing with a 1/8" ballnose; approximately what would our cost be?

Scanning services are quoted on a model by model basis. If the customer has multiple parts or is a consistent customer, then quantity or OEM rates may apply (OEM for customers who leverage my technology to extend their range of "in house" services and make money off of my scanning services - win/win) If you would like a quote, please send a photo or two of the model and overall rough dimensions.

3)Also, Are you a reseller of the scanning system you use? Having the capability of taking the job from staert to finish in house would allow offering rapid turnover on 3D replications, which may, or may not be worth the investment in learning curve and hardware.

Yes, but I do not sell my systems to the general public. It is a far better deal for small to medium shops to just send things in to be scanned. There is WAY more to it than just 'pushing a button' - most have absolutely no idea how much goes into it.

Feel free to contact me off list regarding scanning services.

-B

[Mayo]

I recently had to duplicate some old original corbels.
The method I selected was to photograph the original, with a ruler included in the photo. I then opened the photo in my drawing program and re-sized the photo so that 1 inch on the ruler was as close to 1 inch as I could get it.
Then I manually traced over the parts and after tracing, I cleaned up the lines and curves.

I had two problems to overcome. The original was not square, so I had to decide what to change in making it square, that would affect the look the least. The other problem was not realized until I had cut a sample in cheap eps foam. My tracing was correct and dead on with the photo. What I didn't realize was that the camera or the drawing program introduced a distortion in one direction - the direction I did not have a ruler placed in the photo.
So even though I proportionally enlarged the photo to get my ruler to read 1 inch, it was not proportionally enlarged in the other direction and so my sample was actually 1/2 inch too large in that direction.

I had to re-size the photo purposely distorting the other direction until it was true to the original measurement. Then I had to trace everything again, clean it up again, and cut another sample which turned out perfect.

I was glad I discovered this before I cut up the client's Azek sheets.
The scrolls on each side were cut from 3/4 inch, the middle of the sandwich was one piece of 3/4 inch and one piece of 1 inch. There were 7 corbels, so 28 pieces in total.

A long time ago there was discussion about converting patterns using the probe or a piece of conductive rod chucked into the router against the edge of a pattern which had aluminum foil applied to it. I've never tried it myself.

[dana_swift]

Louis- your question about accuracy for an organic shape 1.5" thick x 10" wide x 48" is easy to estimate.

If your camera has 4000 pixels across and you use the wide part of the image along the 48" axis. If you fill 2/3 of the frame the 48" correlates to:

4000 * 2/3 = 2666 pixels.

The object is 48" long so the distance represented by an individual pixel can be no less than:

48 / 2666 = .018" approximate best case.

That is the best case if you do everything exactly correct with a perfect camera and a perfect lens. Probably it wont be that good, so divide the estimate by half for reality:

0.018 / (1/2) = .04" approximately

The next issue is the effect of not being perfectly flat. The best solution is to use a telephoto lens and photograph the object from a distance much longer than 48". That compensates for the "organic shape" at least somewhat. So the 0.04" best case estimate will degrade. Exactly how much requires a lot of math. To learn more read up on photogrammetry. (The science if measuring things from photographs.)

But a reasonable expectation for an organic shape is 0.1 inches in your case. That number comes from dividing the optical errors by half again:

0.04" / (1/2) = 0.1"

The important thing is the more pixels, the more accuracy. The better the lens and the further you are from the object the more accuracy. The further you are, the smaller the relative image so fewer pixels.

As a friend of mine once said "An experiment is worth a thousand expert opinions". Take some pictures for free and see how well it works. Measure the accuracy in the aspects of the result that are important to you.

D

[Brady Watson]

1)I had two problems to overcome. The original was not square, so I had to decide what to change in making it square, that would affect the look the least. The other problem was not realized until I had cut a sample in cheap eps foam. My tracing was correct and dead on with the photo. What I didn't realize was that the camera or the drawing program introduced a distortion in one direction - the direction I did not have a ruler placed in the photo.
So even though I proportionally enlarged the photo to get my ruler to read 1 inch, it was not proportionally enlarged in the other direction and so my sample was actually 1/2 inch too large in that direction.

2)I had to re-size the photo purposely distorting the other direction until it was true to the original measurement. Then I had to trace everything again, clean it up again, and cut another sample which turned out perfect.


3)A long time ago there was discussion about converting patterns using the probe or a piece of conductive rod chucked into the router against the edge of a pattern which had aluminum foil applied to it. I've never tried it myself.

Mayo - thanks for posting this...It gives others an idea that there is real work to be done on many parts they will get in from customers and there there are 'executive' decisions that have to be made to make chicken salad out of chicken poop. The customer wants a square part, but the original is all warped and whacked out..."now what?" moments...

I never mess with the photo other than to crop out non-pertinent areas. Leave the scale alone & handle it when it comes time to scale the vectors themselves.

Yes - you certainly can use a piece of foil and a rod - which was the ORIGINAL SB probe - before there was an SB probe! ...or you can just use the SB probe. A word to the wise for 2D tracing - offset your imported DXF to the inside 1/2 the probe tip diameter to get a true edge tracing. Clean up the nodes just like you would with a photo, verify sizes and Bob's your uncle.

As I've said many times before...ANY form of digitizing whatsoever has it's challenges. There is ALWAYS some cleanup to be done, regardless of the method. Any way you slice it, it takes work - so don't be fooled into thinking it doesn't, or you are 'doing it wrong' because you have to put some effort into it. It's just one of those things. Use your brain, listen to your intuition and git R done!

-B