Taking the lines of a hull with the help of photogrammetry

[flo-mo]

Only recently I realized that it is even possibile for an interested amateur designer to engage in something as advanced as photogrammetry. More information on the topic: https://cwbblog.wordpress.com/2014/0...uch-as-a-boat/

I came across this subject when I watched this video recommended to me by youtube: https://www.youtube.com/watch?v=ZIW4XU6Wm8Q

Intrigued by the idea to apply this method for documenting an existing boat design, I did some research to see if there is free software available to give it a try.

The one that worked best for me with my limited knowledge is Agisoft Metashape 1.7.3 Standard Edition which is not free, but you can test it in full function mode with 30-day trial license for free.

I think I was lucky because my first attempt of making use of the program resulted in a decent outcome. There sure is a lot more room for improvement and I am eager to learn more about the possibile applications of this technology.

Here is what I started with to see if it is even worth the trouble.

Granted I started with a 1:10 scale model but the size of the object is only limited by your ability to gather the number of relevant photos in necessary quality of your object.

So I took 181 photos of the scale model of my Flywood Two Sheet Sailing Canoe trying my best to match the requirements for the photos.
Import the photos to the program, do some adjustments and start the processing. Then I realized that my PC is too old and slow to do the calculation reasonably fast (estimated time: 1 day, 12 hours). So I had to use a friend's more powerful PC which did the job in 90 minutes.
Beginner level tutorial of the 3D model reconstruction: https://agisoft.freshdesk.com/suppor...es/31000152092

Random selection of some of the 181 photos:

The program showing the camera positions:


The 3D object represented by a "dense cloud":


From the dense cloud a mesh is extrapolated which can be exported for further calculations with CAD programs like Rhino.

Mesh representing the 3D model:


Comparison of the original design created in Rhino (red) and the 3D model created with the photogrammetric process (green):

[cstevens]

Very nice!

[AJZimm]

Very cool. I have seen photogrammetry done with lasers but wasn't aware there was a consumer-level program that could use regular digital photography.

I see that the green photogrammetric model appears to have regular section, buttock and waterlines. Does the Agisoft software produce these or did you somehow import the mesh back in Rhino and produce them there?

[Nathanael]

Very nice and looks like it will be very useful to to help transfer those last minute changes back onto the computer

[Juan]

Woooow

[Ben Fuller]

Nice. David Cockey and others have been doing this for some time, on various historic craft, ranging from a score or so of peapods to sardine carriers. It cuts down the field work considerably and if you get interior photos can prove a nice document of construction. There is a learning curve to processing and both a cost and learning curve to translate the results into some thing that can be analyzed.

[Edward Pearson]

What's about the minimum number of 'pictures' you would need to render a hull from an historic craft?

[flo-mo]

Very cool. I have seen photogrammetry done with lasers but wasn't aware there was a consumer-level program that could use regular digital photography.

I see that the green photogrammetric model appears to have regular section, buttock and waterlines. Does the Agisoft software produce these or did you somehow import the mesh back in Rhino and produce them there?

The Agisoft software only produces the mesh which can be exported in various formats. From this mesh sections, buttocks and waterlines then were produced in Rhino.

[flo-mo]

Nice. David Cockey and others have been doing this for some time, on various historic craft, ranging from a score or so of peapods to sardine carriers. It cuts down the field work considerably and if you get interior photos can prove a nice document of construction. There is a learning curve to processing and both a cost and learning curve to translate the results into some thing that can be analyzed.

It's probably a blessing and a curse because you can waste a lot of time processing the information you collect with this technology. I think once you establish a proper workflow, it could be really helpful for documenting all kinds of objects. Also, the idea that a series of photos taken in the past of an object that no longer exists could allow you to reconstruct a digital version of it is fascinating.

[flo-mo]

What's about the minimum number of 'pictures' you would need to render a hull from an historic craft?

From the general image capture tips of the Agisoft helpdesk portal: https://agisoft.freshdesk.com/suppor...es/31000149337

• Overlap. In the case of aerial photography, the overlapping requirement can be put in the following numbers: 60% of side overlap + 80% of forwarding overlap.
• A number of photos: more than required is better than not enough. Later on, you can disable or skip excessive images.

Much more information in the user manual: https://agisoft.freshdesk.com/suppor...es/31000149337

[flo-mo]

What's about the minimum number of 'pictures' you would need to render a hull from an historic craft?

If you think of a specific historic craft and you have a set of photos you are willing to share, I could simply give it a try and we will see if it will work.

[Edward Pearson]

I have quite a few photos from a visit to Clovelly village, Devon, of the inshore Picarooner work boat. There is one at the entrance to the visitor center that Peter Radclyffe rebuilt/ built. There is one left working, have a few of that also.

[Edward Pearson]

90B0B702-D7F2-402D-BC11-E5DBE6CEF2B2.jpg

60F78661-3392-44FB-8E45-24C84A3BC079.jpg

90ACD0A8-ABEE-4EC5-B461-FE55BA044387.jpg

FEF9431C-41EA-4FB5-98CF-16EF473610D9.jpg

[Edward Pearson]

D0186116-77D5-479A-B712-28CBEC63157F.jpg

4CBD3FF0-557B-4D46-A15C-D75994108185.jpg2639870C-83D1-42AD-9E97-9EBD4190F484.jpg

63F23014-CEC5-4B8C-8887-D0948AAF842E.jpg

[Edward Pearson]

A458F626-89A5-4F19-B566-14A2C335DBB7.jpg8F013E32-189D-41C8-851E-746FCA5215FF.jpg

C75B29C8-0DC9-4F19-992B-DDAEF67398F4.jpg

[Edward Pearson]

8A4EA229-9495-42F8-BF83-46EB5C75AAF1.jpg

057F5987-4D89-4398-9978-ACA92C79DB7A.jpg

32896014-F757-426C-91D0-C5AD76D98CA2.jpg

3FC6536E-280F-4A79-A382-27EC4BB0FD28.jpg

17DED77C-1CB0-4542-9253-F0F9268A6B2D.jpg

[Edward Pearson]

8690FDBB-B4BD-4A3A-97DD-6E8BAD511860.jpg

6687CAD1-AE50-4B1E-8513-7E12D6947675.jpg

17000766-913B-4E9E-8008-26310D876670.jpg

BC062898-D3EE-4B2E-80EA-3CA4399F91CC.jpg

37EF909B-106C-490E-8B29-00F5D7B5CA52.jpg

[Edward Pearson]

8158D9C4-BDD5-4A37-A7C7-2D7072E02BC3.jpg

0A2DC1A3-673D-43E9-B6DF-310A4A8E82AD.jpgD210CF3E-8659-4EAC-8ADE-011CFF1DEF3D.jpg

[Edward Pearson]

B605F524-4CD1-4649-8032-6C254EADB404.jpg

[Edward Pearson]

Won’t be enough of the one on the beach but some details of construction show for those interested...

9770895A-F14F-4ABD-BC1E-E69093D6E5CD.jpg

59EA1B78-372F-47F1-8ED4-A9361DE6D199.jpg

FB935639-F3F6-4AA2-822C-132C22071A66.jpg

0D607112-7273-4E48-9FC8-848543B8FD68.jpg

[Edward Pearson]

Attachment 88368Attachment 88369

Attachment 88370

Attachment 88371

Attachment 88372

[flo-mo]

Thank you Edward for sharing the photos of this lovely craft.

Unfortunately, as expected, the information content of this photo set is not sufficient to achieve a useful result with this software. At least it only took 5 minutes to find out.

[John Meachen]

A couple of years ago I stumbled upon a program called Visual SFM which appeared perfect for this kind of thing.It ought to work,but I struggled with it.

[flo-mo]

Continuing my adventures in the mystery world of photogrammetry.

Of all of my designs I have actually built, the Gorewood Cottage Cruiser 16 is my favorite.

I haven't taken the lines yet as doing it the traditional way can be a tedious process. Although using photogrammetry for this task will probably turn out to be even more tortuous, I am having fun learning something new.

Although my first attempt was not fully successful it is nonetheless a fascinating journey to explore the complexity of the process.

When I started this attempt I knew that most likely the outcome would be flawed, as shiny objects like a varnished hull are less than ideal for the software to process, but I was interested to see what the result would be like.
Also I already have a plan B for a better result which will take a little more effort, but I was hoping that maybe I will get away without it.

An unstructured surface is difficult to interpret by the software so the first step was to apply pieces of tape to the hull.



It is best to take the photos when the sky is overcast so there are no hard shadows. Conditions were not perfect because at the moment we are having a heat wave with rarely clouds at all, so this may also be a factor for an imperfect result.

I proceeded anyway and here is the outcome.

Dense cloud on top, wire frame at the bottom:



View mode on top - model confidence (blue - high, red - low), bottom - solid model:



And finally an image of the 3D object and the lines drawing created in Rhino:



Although this is not what I was hoping for I am confident that the next try will lead to better results.

[Peerie Maa]

Try giving it a coat of easy to wash off poster paint.

[flo-mo]

Try giving it a coat of easy to wash off poster paint.

Good advice - I will experiment with a highly diluted slurry of dark clay.

[sp_clark]

Featured in the July/August edition of WB magazine:

33FD48D0-E3A1-43EF-BD64-6B17A8E52521.jpg

[flo-mo]

Featured in the July/August edition of WB magazine:

Thanks for the reference -- I may get a digital copy.

[flo-mo]

Here is the next iteration in my endeavor to somehow master the magic world of photogrammetry.

This time my Gorewood 14+ solo canoe is the real life object to be converted into the digital world.

I skip the laborious part of applying pieces of tape as i am not sure if this was helping the software at all.

All I do is generously covering the hull with clay slurry, which is a quick and straight forward procedure and the clean up process should be easy.









The hull isn't perfectly covered, but this may even be beneficial as it adds some structure to the now matted surface. I also tried to make better use of the camera by manually setting ISO, aperture value and shutter speed.

Here is one example of the 130 photos I took at almost perfect weather conditions.




The software indicating the camera positions:



I think for my purposes this is the way to go. I am really pleased with the result. Four different view modes of the 3D object: Dense cloud, solid, wire frame and model confidence.

[flo-mo]

After importing the 3D object into Rhino I am able to create a proper set of lines which was the goal in the first place.



Some more renderings:






For comparison some cropped images of the same canoe from a year ago:




The original images (and some more) I used for this thread probably in slightly better quality: https://photos.app.goo.gl/D59rFCNkumYUthuR8

[Peerie Maa]

I love it when a plan comes together.

[flo-mo]

Also featured in Water Craft Magazine July/August issue

TAKING OFF THE BOAT’S LINES – Photogrammetry, laser scanning: Andrew Wolstenholme on how we’ll do it tomorrow:

[flo-mo]

I'm back working on the Gorewood Cottage Cruiser trying to get a similar, if not better, result to the Gorewood 14.


First, I got rid of the pieces of tape. The Harlequin look makes for a cheerful appearance, but didn't add any useful information for the software to process.




The mud treatment, on the other hand, seems to be the way to go. The hull is covered with dried clay mud.




Now everything depends on the quality of the photos.

Set up the camera properly and wait for the perfect light. This proved to be quite challenging. It took me three sets of photos to get it right.

For the longest time, the overcast sky that provides the lighting conditions that would be best for the photos was nowhere in sight. Finally, storm clouds rolled in and I hurried to get the shots before the rain started and washed off most of the clay slurry I had applied.

After the rain:




This is what the final set of photos looks like:




Now back to the computer -- I followed the workflow described in this tutorial and ended up with the result I was hoping for.

The positions of the camera:

[flo-mo]

I am really pleased with the quality of the result.

Dense cloud:



Wire frame:



Model confidence:



Solid model:



Gorewood Cottage Cruiser 16 - 3D model in dense cloud mode:

[flo-mo]

The 3D model of my Gorewood Cottage Cruiser 16 Canoe constructed with Agisoft Metashape 1.7.3 Standard Edition exported to Rhino, where the lines were added to the model.