UB88 Photogrammetry Project – Dive 4 (Stern Torpedo Tube)

Summary: not quite a “wasted” dive but I also didn’t get the results I wanted.

I had a feeling that the “ghost net” at the stern of the ship would be a tough one to build a model for. There are quite a few complications like the fact that it isn’t a solid surface and that it “moves” with the currents.

It is a shame because I really love the single torpedo tube at the stern end of the wreck. It is so narrow and then the stern rapidly flares out and the ghost net gives it a creepy look and feeling.

In addition, the “end” of the torpedo tube is very cool looking:

I ran into multiple problems both during the dive and during processing:

• It was the second dive of the day and the natural light was better but quite a few of my pictures turned out dark so I used Lightroom to bump up the exposure. This might have caused problems.
• I ran out of room on the memory card during the shooting of the net. I think this is secondary though.
• I also used Lightroom to conduct a better white balance and, when applied to all the photos, it wasn’t the best
• I didn’t get quite enough angles of the torpedo tube end and so the pictures didn’t align and I had to use manual markers to get them to align. Since it was done at such a pivotal point, one small error in an angle gets magnified.

Below is a picture from Drew of me photographing the ghost net. This will give you an idea of the scale:

Here is a picture of the textured model as it stands right now:

It almost looks like a cool watercolor or pastel painting that somebody made!

Below is a view from an angle. You can see that the software is picking up a lot of “artifacts” and the port side of the wreck doesn’t even really show up correctly. I think this is due to the mis-alignment issue I mentioned earlier.

You can see what I mean by the wreck “flaring out” from the torpedo tube.

Also note the strange angle of the torpedo tube. I do NOT believe this is due to the software or alignment issues. I believe this is due to the damage from the net.

Example alignment problem

I thought it would be interesting for people to “see” how the software “aligns” the photos and what it can and can’t interpret.

If you look at these two photos, I bet you can immediately align them in your head and figure out that the second one is just angled from the first:

There is no way the software can reliably align those two photos! You would need about 4-5 photos minimum for that amount of angle transition for the software to work! 🙂

Even using manual markers would be tough in this case because of the change in angle.

Next Steps

I have another dive (Project Dive 5) coming up this week during which I will concentrate on trying to photograph both sides of the hull from the conning tower forward so I will hopefully have 50-ish percent of the wreck done at that point.

I will also try building another model of the stern (this will be my fourth attempt) without any color correction or exposure changes from Lightroom and just use the original photos. We’ll see how that turns out. It is a somewhat slow process due to the need to manually put in markers and because I don’t have all the pictures correct, I have to play around with the order in which I add them to get them to align.

Either way, I have a stock of about 300 photos of that general area that I can use to help fill in the blanks the next time I get around to photographing that section.

But, I was right. This is going to be the hardest part of the project.