Reverse engineering scale models

try this:
http://www.agisoft.ru/products/photoscan/standard/

its a photoscan software. You snap tons of pics of teh object (probably paint or tape lines on it to help the recognition) and the software does the rest. Have not tried but have meant to.

There is a free demo of it.

H

 

I've used PhotoScan to "measure" a number of boats, and have also done other testing with it. It's very useful but does have some significant limitations.

PhotoScan needs an irregular "texture" on surfaces. It does now work with clean, smooth surfaces which are a single color such as paint or gelcoat. It can work with such a surface if the surface is sufficiently dirty, worn or decayed. Unfortunately tape lines by themselves don't help much. Unpainted, grainy wood can work very well as does peeling paint.

The output from PhotoScan is a mesh. The mesh will generally have noise around the edges. Also, depending on the surface "texture" there can be noise on the surface. So a mesh created using PhotoScan will generally need editing before using it as input to a CNC system or similar.

 

The Gougeons had a method of making a model about the size you made and measuring it so they could predict the full size boat.

This is the same technique used by Herrishof.

Both of these people have made very successful boats this way.

The Gougeons manual is available free on their website. Corley previously posted the link in another thread, sorry I don't have it available.

Very simple equipment will allow you to do a good job.

 

maybe you could print some irregular pattern and drape it over the model when trying photoscan?

 

That would work if the surface is developable surfaces. Otherwise the paper won't conform to the surface. Probably the simpliest method to get an irregular pattern is to mix a powder in alcohol or other quick evaorating fluid and put it on the surface with a very coarse, light spray. But the boat's or model's owner may not be agreeable.

 

If the coating is easily washable, perhaps with just water, I guess there wouldn't be too many objections...

 

A mueum curator is very likely to have objections, and most of the boats I've measured are in museum collections. Boats which are used are a different story.

 

Just stumbled over Autodesk 123D Catch

Might be worth a try?

I have acces to 3D modeling software myself. I'm doing this for a living in fact.
The initial price of $1400 you mentioned is not too far from reality. It would mean about 10 hours work plus equipment. But I completely agree to that these are no prices you want to pay for a hobby or fun project.

Another diy-idea: paint or tape lines in regular distances on the model hull and make high-res pictures with a very big tele lens from the front. You could get some fairly nice sections with almost no perspective distortion. put them as "background" in your 3D software and build your shape over them.

The good think: You just need to build half a boat.

 

Boataid,

Have you decided on a process or dropped the idea?

 

I have experimented with 123D Catch. It is similar to PhotoScan with similar requirements for an irregular surface "texture". Advantages of 123D compared to PhotoScan are:

• 123D Cach is currently free. PhotoScan Standard is $179.
• Processing is done on an AutoDesk computer so the speed and memory of the user's computer doesn't matter.

Disadvantages are:

• The user has much less control of the process.
• Photos have to be uploaded which effectively limits the number of photos which can be used. For a simple half model this shouldn't matter. For more complex objects this can be a significant limitation.
• My experience is 123D Catch captures less geometry and does not work as well as PhotoScan with limited surface "texture".

 

With some 3D modeling software such as Rhino it is possible to correct for the perpsective distortion though it is not a simple process.

 

I've got no expieriences with 123D Catch. I just saw a friend playing with it. I know that the quality of the returned surfaces woudn't be sufficient as an input for a milling machine. But it will give you an idea of the volume and shape of the model.

I did a 3 hour exercise with just the picture from post #10. This is the result: No idea though, if this is what Boataid want.

 

I've been working around lifted lines and hull packages for years, and recently into SolidWorks to become my primary design environment.

Hullform (now in public domain), which I have been using for years, is relatively friendly for working from lifted offsets, particularly where you have prominent longitudinal lines - good for chine forms; not so for your canoe.

For the canoe, it would make sense to arrange the lifting points at a number of transverse section positions, with points at consistent heights from the height datum - table top for a model; shed or slab floor, or optical plane for full size vessels.

Don't be too fazed by trying to generate a highly accurate representation of the model. It may well contain some shape oddities which you regard as not valid. Also, don't be overcome by chasing a high degree of fairness and smoothing. You aren't going for display cabinet/yacht finish.

I strongly commend the concept of building some expendable models. For the size of canoe, 1/2, 1/3 size, to give a model that will take 1 or 2 persons has a lot of appeal.

Think of the total process as a design, rather than a replication, inspired and guided by your existing model. You may well want to initiate significant design changes along the way. You should also realise that your model may have had some distortions in its copying of its full size prototype. In all of this, remember that is is your design; your project.

 

The "light painting" concept with laser level and digital camera would be my choice.

Make up a support for the camera and laser to keep them spaced apart at a constant distance, move the boat model thorugh the support and do not change the zoom during the process. It doesn't have to be on telephoto: all the "frames" will have the same Y and Z arbitrary scale where X is the fore-and-aft axis of the boat. Parallax will not matter since each frame is photo'd at the same distance. Mark equal frame locations on the camera/laser support to keep frame spacing constant.

When frame digitalization is done dismount the camera and photo the frame from the side at the same distance as the frames. The middle frame location markings will then be at the same scale as the frames. Note: the outer markings may be distorted by parallax but they're all the same. Now you have an image with the X axis at the same scale.

Some boat design software, such as FreeShip will allow you to import a background image so you can adjust each frame of the digital design. A software package like FreeShip will then allow you to fair the design and get the necessary hydrostatic and hydrodynamic numbers.

Design and construction are interdependent. Developable plates were mentioned in an earlier post; the relevance of that would depend on your chosen build method. Plywood or plank construction requires developable plates and a software package like FreeShip can do plate development. However, as it is the smooth model hull will need to be molded or strip built.

 

Terry, have you used the method you describe or is it currently "theoretical"? Is the camera aimed normal to the light sheet? If not how is perspective "distortion" of the section shapes corrected?

Steve Killing and his son Jonathan Killing developed a similar method which they called "SmartScan" It uses a light sheet source and a digital camera which were mounted together on a carriage. The carriage moves along a track parallel to the boat. The images are processed in a computer to identify the projected curve and deal with the perspective distortion. The SmatScan system is described in http://journals.hil.unb.ca/index.php/MCR/article/view/17908/22050 Around 2000 the Killings built a unit for the Canadian Canoe Museum. When I inquired at the museum several years ago about the system I was told it was no longer being used.