Scanning larger objects

Printable View

10-15-2021, 06:49 PM
drdachel

Scanning larger objects

I am way out of my element here, so my apologies if I'm in the wrong group. I'm starting some research on the possibility of making some custom aftermarket items for boats. Ideally, I'd like to use a phone, ipad etc to scan a boat that i want to work on and create a 3d template with precise angles etc for the topside of a boat (smallish one, say 25-35 feet). For example the bow of a cruiser to know it's curve. From a quick search, I'm seeing Heges and Itsees3d as apps that might work. Any suggestions or insight would be most helpful. Of course, making the part after this is the next step, but with boats that can be 40 - 50 years and still afloat, I need a way to map the surface of the bow and not go off of blueprints etc. Thanks Dan
10-19-2021, 06:16 AM
curious aardvark

Hmm, what's your budget ?

With 3d scanning the more money you spend the better the result.
Industrial handheld 3d scanners are really bloody expensive. And most would not do something the size of a boat.

You have several budget options.

1) photogrammetry. long winded and often involving several different programs - can be really effective. Autocad 2018 and newer comes with a built in photogrammetry program Recap, that is suppose to be very good - I have yet to try it - I've got it, just haven't ever tried it.

2) https://3dprintboard.com/showthread....D-Scanning-app - not sure what size it will go up to.
But if you ave a recent ipad or phone - it's the cheapest option.

For the record a 'small' boat is still a massive object to scan :-)

3) drone based 3d scanning. I don't know much about this, but it's becoming more and more popular with architects and builders. And even a good drone and scanning software will be tens of thousands of pounds cheaper than an industrial scanner that can do those sized objects.
It's basically photogrammetry - so you could probablky use a mid level drone - say £500 and the autodesk software would probably be the best cost-per-result option:
https://www.autodesk.co.uk/solutions...metry-software

A cheap handheld scanner - that might work - starts at around £5000 and goes up to the full industrial versions at £20-50,000
https://www.einscan.com/
'Budge't handheld scanners.

For most things I have found that simply measuring stuff and a cad program is generally a lot faster and more accurate than 3d scanning.

But if I were going to make models of boats, then autodesk recap and a drone with a good gimballed camera would be my approach.
10-20-2021, 04:32 PM
Martin_au

I’d be looking at the Leica blk360. Cheap terrestrial scanner. Easy to use. Can be mounted on poles or inverted to see the hull from the deck or deck from the ground.

Photogrammetry will be problematic on a lot of boats due to the lack of texture in the surface. On a wooden boat it would be ok, but on anything painted, it just won’t find patterns to match, especially on the hull.

You’ll also need to establish decent control for photogrammetry to get everything scaled correctly which may require something more sophisticated than a tape measure.

If you’re feeling talented, and budget limited, the iPhone 12 and 13 Pro’s SLAM LiDAR may do the job.
10-22-2021, 01:33 PM
curious aardvark

the leica is around £20,000

so in the category of industrial scanners.
Not exactly a budget option.

Also modern photogrammetry, with current high resolution cameras, works just fine with solid colours.
10-22-2021, 02:10 PM
Martin_au

Quote:

Originally Posted by curious aardvark

the leica is around £20,000

so in the category of industrial scanners.
Not exactly a budget option.

Also modern photogrammetry, with current high resolution cameras, works just fine with solid colours.

Got some examples to support the claim it works on solid colours? Are you sure you’re looking at straight photogrammetry and not structured light scanning?
10-23-2021, 08:27 AM
curious aardvark

you got any example that it doesn't ?

using a modern hi-resolution cameras and autodesk photorecap ?

I've done a few scans recently - with solid colour objects and qlone - a photogrammetry app.
it's not an issue.
Photogrammetry works more on the background colours than the colours of the actual object.
Which you'd know, if you'd ever done any.
It's frequently used for buildings - objects famous for large blocks of solid colour.

It does seem that you just automatically say the opposite of everybody else. Without any experience or evidence.
10-23-2021, 02:38 PM
Martin_au

Quote:

Originally Posted by curious aardvark

you got any example that it doesn't ?

using a modern hi-resolution cameras and autodesk photorecap ?

I've done a few scans recently - with solid colour objects and qlone - a photogrammetry app.
it's not an issue.
Photogrammetry works more on the background colours than the colours of the actual object.
Which you'd know, if you'd ever done any.
It's frequently used for buildings - objects famous for large blocks of solid colour.

It does seem that you just automatically say the opposite of everybody else. Without any experience or evidence.

Quite a bit.
First, please don’t try and reverse the burden of proof. That’s a bad faith method of argument.
Second, I apply photogrammetry in both my work and as a side interest on a regular basis (daily to weekly). I work regularly with the state’s museum, the university’s palaeontology group, and others groups in research and industry. I’ve done photogrammetry of objects down to sub 5mm, and up to full site captures. I use a range of equipment, typically a high end camera, but also drones and more recently the Leica GS18i.
Third, the entire principle of photogrammetry (sfm-mvs) works by matching identifiable points in photos. No identifiable features, no matches.
Fourth - one of many: https://www.sciencedirect.com/scienc...96207415000230

Quote:

One of the factors that determine the data quality produced by targetless photogrammetric techniques is the feature richness of the surface being captured. The Structure-From-Motion and Multiple View Stereovision (SFM-MVS) pipeline is no exception to this rule as it relies on the ability to identify corresponding points within a collection of unordered images

Quote:

Furthermore, the complete artefact's reconstruction without the use of NFPs proved to be problematic as SFM-MVS failed in image spatial alignment due to lack of features. Thus, we repeated the data collection phase by including additional target objects in order to provide features. This time, the alignment was successful but the poor quality of the reconstruction (e.g. noise presence, mesh incontinuities, etc.) indicates an unfriendly SFM-MVS surface.

If you like I can drop 50 or so links to photogrammetry papers that describe how the process works.