A Lost World of Shipwrecks Is Found | Hacker News Reader

A Lost World of Shipwrecks Is Found

nytimes.com

372 pointsNaOH64 comments

Anyone that finds this interesting should pick up the book "Ship of Gold" by Gary Kinder. It's a quick read and a great tale of eccentric folks pursuing deepwater wrecks and having to develop all sorts of new tech (and fight off competitors) to do so.

Picked up a copy for my father for Christmas, thanks for the rec!

The fact that the financier wishes to remain anonymous hopefully means they are just humble. Nice to see someone's wealth being put into something we can all enjoy.

Publicity = Risk these days.

Anyone putting their name out, for anything, risks getting someone, somewhere, really pissed off.

One sentence can start a global riot of the press get behind it.

Yep. Anyone smart enough to get that rich usually figures out that the attention that comes with it is at best, tiresome (and often quite aggravating).

Most rich people just want to stay anonymous and protect their privacy.

Most billionaires you have never and will never hear about.

Most? Is there any data to back this up?

There are over 1,500 billionaires.

And those are only the ones that can be guessed to be billionaires based on their publicly known assets.

There are probably quite a few more which don't own big stakes in companies and are hidden behind financial webs.

>Photogrammetry, a process using thousands of photographs and readings, produced a rendering that appears three-dimensional.

Photogrammetry generates actual three dimensional images, not merely appearances of.

If if you want to pedantic, all images are by definition 2 dimensional and can only appear 3D

I believe that photogrammetry can reconstruct 3D geometry, especially with the help of range sensors like the article mentions. The process results in 3D information like triangle meshes in addition to 2d images.

Yeah but in defence of the guy you're replying to, once you render them they are 2d images.

And before that they're 3D models.

I keep finding photogrammetry papers discussing how to get the sizes of objects from photos. But those look like photographs and yet somehow seem unreal. Do you know any good intro to the subject? Something that would show a regular photo and the photogrammetric side by side?

You can try it yourself with commercial software: http://www.123dapp.com/catch

Scale can be hard to capture in photogrammetry, but in general the unrealness probably has more to do with the somewhat smoothed out (triangulated) surface that doesn't capture too fine a level of detail and the textureless (uncoloured) model they are rendering.

No need for commercial software! Try VisualSFM, Insight3D, Regard3D!

Or OpenSFM: https://news.ycombinator.com/item?id=12835771

Amazing find, strange that it took so long to actually search the Black Sea considering the perfect oxygen-less environment for preserving these wrecks.

Most of that was under USSR control until recently. Even after '89, it would take some time to figure out good sites, get funds, get permissions etc.

I don't think this the reason - there were archeological expeditions under Societ Union, and it's been 25 years. Was there some technological breakthrough, or some other political change lately?

There was an expedition in 2000 that located a few wrecks. This new project is locating a lot more, and they are using new 3d scanning things mentioned in the article. Probably the biggest constraint is funding. They mention this is paid for by some anonymous rich donor.

The Black Sea is huge, even sonar sleds have fairly narrow widths, so it would take a ton of work to find potential wrecks, and you need to visit all of them to determine which ones are modern and which are older. And that's a very long and costly process since it's well below diving level and very dark. You have to send down an ROV which can basically see only a little in front of itself and in a narrow field of view.

The Baltic also preserves wrecks extremely well, due to the lack of wood eating creatures.

The Black Sea is famously anaerobic - no oxygen and few microbes to consume organics. Wood and human remains would disappear in decades in other places. And the Titanic is decaying from iron-loving bacteria. The iron and steel parts will be gone in another century.

There is so much awesome stuff to be discovered in our own seas. Would love to see space exploration and deep sea exploration on the same level of excitement, investment and pursuit. I would even argue that Google, the company who's mission it is to organize the world's information should focus on these things vs building Chat apps

Maybe I missed it, but what happens to the corpses? Are they also somewhat preserved, or do they rot slowly?

IIRC they still find human bones around the Antikythera wreck every once in a while. At the very least they did when it was first excavated at the end of the 19th century, and that was in regular medium-shallow salt water.

My guess would be the soft tissue gets eaten by fishes. I don't know what would happen to bones.

They might not be located near the wreck. This is a small vessel with a crew of only 20 or so. At any given time almost all of the crew would be on deck, even if they weren't they would have if the boat began sinking. The crew likely would have been in the water as the boat sank, and even if they did die they may not have ended up in the same location (a dead body in the sea doesn't sink at first).

Is there a gallery with more photos somewhere?

Very cool. I think the article was a bit light on images though, having only four of them and mentioning that 40 ships were found. Would like to see many more images of the ships. Also, the researchers should consider making a VR experience for VR headsets based on the 3d data they have.

Looks very cool - where are the models?

Click bait title. Came to read about affordable care act and was disappointed.

How faked are these photos? They're labelled "photogrammetric" as if to hedge how real they are, but I don't know if that means they're photoshopped. I ask because most large-scale underwater photographs are faked since visibility is usually low underwater and there just isn't enough light. Google images of "Titanic underwater" and you get either close-ups, hyper-realistic drawings, or mosaics. It's easy to think that this guys work are real photos: http://www.kenmarschall.com/wreck.html#Num5

They are not photos, that's made clear from the article, they are renderings.

Think of it this way, the ROV goes down and uses a high resolution sonar instrument to see the 3D structure in front of it down to a centimeter/millimeter level of detail, at the same time it takes pictures. Then they take all that data and process it with a computer in order to generate basically a 3D textured model. Then they take renderings from that model and produce the images you've seen. The individual images from the ROVs would be much less interesting due to the very small field of view and the limitations of the lights.

In no way was it clear to me whether or not the images were the renderings that was produced, I'm aware of photogrammetry and the limitations of the underwater photographs (hence my question), I even have experience doing computer vision with underwater ROVs and their limited cameras (though ours had good field-of-view). The problem is that with their captions, I first assumed that it's a photo used for photogrammetry. Or rather, I thought it was a mosaic of photos, all of which were used for photogrammetry. If I were the NYT and didn't want these to be mistaken for photographs, I'd label them as "renderings of a photogrammetric model" or such. As it is, a "photogrammetric image" could be one of several things.

In retrospect, there are some tell-tale signs (like the edge of the model where the ground is discontinuous).

I can't edit anymore but I want to add that I regret my tone above, I think it's overly combative. I now think my original mistake was thinking that the model that the photogrammetry would generate would 'obviously' be a mesh or point-cloud without color/texture data and so the images being entirely computer rendered didn't fully cross my mind.

To be pedantic, "photogrammetry" means the models would have been constructed from a dataset consisting of photos.

Edit: Sorry, I read your reply below and obviously I'm not telling you anything you don't know. Maybe someone else finds it helpful.

My original reply: "Photogrammetry" in this context, assuming the article is using the term correctly, should mean that the models are a computer reconstruction of the 3d shape based on a dataset of a large number of photos.

Depending on the coverage and quality of the photos taken, some cleanup would be necessry, and potentially a human might have made up particular details after the fact if some portions of the ship were not visible in the photos taken, probably due to occlusions etc.

The article seem to say this was a side product of a geological survey mission, so my guess based on that and personal experience with photogrammetry techniques is the geometry of the model is feature accurate, although some noise may have been introduced. The lighting is likely to be artificial.

Reply on news.ycombinator.com