Reconstruction of a historic racing boat

[bernhardb]

Hi together,

here my current project (not ready until now):

Reconstruction of the plans of a historic racing boat which is now located at the "Deutsches Museum" in munich, germany. It was build in the "Pontier Werft" in Switzerland and used from the 1930s to the late 1950s:

pontier3.jpg (80.67 KiB) Viewed 6897 times

I show You this because I think it might be interesting how to get from a object like this (where You have limited access - You can not crawl into the boat for meassuring or turn it around for better access in a museum!) to a nearly correct CAD model. My target is to make a 1:6 RC model out of it...

The solution I choose was to photograph it from all directions an combine the photos to a point cloud which I could import in FreeCad. The combining was done with a opensource-program called "VisualFSM". The yardsticks on the ground help to control the dimensioning of the results. Thats what I got:

pontier2.png (234.43 KiB) Viewed 6897 times

And without the point cloud:

Pontier1.png (161.12 KiB) Viewed 6897 times

All done with FreeCad-Pre17-version.

[Joel_graff]

The solution I choose was to photograph it from all directions an combine the photos to a point cloud which I could import in FreeCad. The combining was done with a opensource-program called "VisualFSM".

That's amazing! I was playing with visualSFM a couple of years ago. It was tricky getting it to run, but it worked surprisingy well.

I'd love to see a detailed workflow for how you created this as well as some of the issues you encountered getting it into FreeCAD. This is a really valuable workflow that, I think, isnt given enough exposure.

Great work!

[chrisb]

Some time ago we had here something similar with a railroad model built after ancient photographs.

[bernhardb]

To the workflow:
In my opinion, the really tricky part is to get a satisfying good set of pictures of the object. I don`t know how often I was in the museum to get it - I have a annual entry card It helps if You put some marks like flayer and "postits" on the deck, on the floor and so on,,, this helps visualFSM to get its job done.

The workflow in VisualFSM is quite simple: Load pictures, press "Feature Mapping", press "Compute Sparse", press "Compute dense"
If it works, it is good, if not You have to think about which photos You gave the program again... for example, until now I was not able to reconstruct the motor of this boat - I have to make a lot of more good photos and propably put some markers on the motor to make it easier to detect. After the "dense" reconstruction You can via "SFM->Save nview matches" save the result, it creates a "PLY"-file.

Now, sorry to say, but the "points" workbench/objects - in which You can load the "ply" file - is not very clever. The bigest drawbacks: The colors get lost and - worse - You have no posibility to scale the points in FreeCad. So, I loaded it in "meshlab", an other open source program, and scaled it here:

pontier4.png (489.99 KiB) Viewed 6839 times

After this, I exported the point cloud again and loaded it in FreeCad.

The workflow is not ready there: Many details get lost in the point cloud, some things - like the positions of the frames - won't be catched at all. If the hull is constructed correctly, the position of the frames dosn't matter so much - the resulting hull is the important thing. But You should have an idea where they are. Also, the structure of the lower part of the hull was only partially reconstructed. So, there is no other way then some more visits to the museum with a yardstick and meassure all the details that were lost.

I even managed to get a yardstick via the cockpit in the boat to estimate the positions of the frames:

pontier4.jpg (95.88 KiB) Viewed 6839 times

I think the combination of photo reconstruction for the "big picture" and meassuring the details is the way to success in this case.

[Pintuxgu]

If I see that boat, I wonder if any improvement in boat design was done after it...
Great work.

As an idea for contrast:
Sprinkle some confetti over the boat (or other big contrastless objects) the next time

[bernd]

Cool project! I wonder if it would be possible to measure anything you need directly on the boat, by get in contact with the authorities of the museum? Hey they get paid by the taxes you pay Even more you could may be give them a digital model of one of their objects for free!

Bernd

[bernhardb]

Indeed, I did measure directly on the boat (with a yardstick) as much as I could, corrected the plan after the measurements (no much correction was needed) and added much details of construction which was not seen in the point cloud. For more (or more exact) measurement than I did You would have to move the boat from its stand and may be to remove the seat so You can crawl into the hull to meassure the distance of the frames exactly... I don't think that would be allowed by the local authorities (if I would have to decide this, I would be afraid for the boat...)