USS Saratoga (CV-3) Island (Bikini Atoll)


I had been doing a lot of photogrammetry work on the Saratoga and had one idea that was a “well, if it happens” project.

Namely, it would be interesting to build a photogrammetry model of the island (or tower) structure on the wreck. In the photo below, the 5″ guns are on the far left, the island is in the middle and the smokestacks are to the stern of the island. The smokestacks are gone at this point.

The primary reason for the “side-project” is that the island is starting to fall apart (like other parts of this wreck) and I thought it was worth trying to preserve a record of what it currently looks like.

Given that this was a side-project, I used some time at the end of a few dives to build up a part of the model to see if it was possible or worth spending time working on. Below I will document how it built up over a few different dives/days and some of the challenges along the way (they were numerous).

The “Short Version” / Result

I was able to get the photos needed using some “spare time” at the end of four different dives on the Saratoga. The details of each dive and how the model built up over time are below. The summary is that it came out reasonably well given that this was just a side project. Below is a link to the on-line model.

This is a screenshot of the model in about the same orientation as the historical picture above:

Dive 1 – The Guns

The first step was to shoot some photos of the 5″ guns after I had done a dive on Plane #1 from the USS Saratoga Airplane Project. I knew, at a minimum, it would be cool to build a model of the guns.

After I got back up to the flight deck, I headed over to the guns and shot about 400 images.

Gun barrels are a bit tricky because the are relatively sharp at the ends and getting photos to align can be hard if you don’t get enough photos. For this dive, I did a reasonable job for a first pass with only a few minutes at the end of a dive. It was literally 7 minutes of shooting photos.

Note how the port side barrel has a cool reef structure on top (similar to the one on the USS Lamson #2 gun). The screen captures below are of the early-version of the guns. It is clearly missing some coverage around the back top section of the guns and on the starboard top edge. But, it was worth pursuing further…

Here is a sample picture of the coral formation on top of the gun barrel:

Dive 2 – Guns v2

Once again, I used a little bit of time at the end of a dive to build upon the guns since I knew I definitely wanted to document & show those in a model. I only added about 120 photos but at least made sure that the blank spots were filled in. This time, it was after the dive to build photogrammetry models of the TBM Avenger and the Helldiver off the starboard bow of the wreck.

I now had at least a base to build from. Low resolution model screen capture is below. The top is now filled in (note that it has some kelp growth on top).

Dive 3 – Island Lower Level

These photos were taken after shooting the Helldiver that is off the port side a little back from amidships (Plane #5). My goal was to see if I could shoot the lower level of the island and “connect” it to the guns. I took about 370 more photos and, at this point, the model was about 900 photos in total.

Two quick notes about this evolution of the model. (1) The water had turned a bit “green” so the color balance is different (see sample comparison photo below). (2) I didn’t bother building a dense cloud or mesh. I really wanted to just see if I could get all the photos to align.

Screen captures below are the sparse (or tie point) cloud.

Here is a photo that illustrates the “green-ish” color of the water/photos. It reminds me a lot of what we sometimes get in Southern California. This can largely be “corrected” but at the moment, I did’t think it was worth the effort.

Dive 4 – Final Chance

I was rapidly running out of time and still had a few key objectives. It was challenging to forfeit cool penetration dives on the Saratoga but such is life. I had visited the Machine Shop (might post about that at some point) and after built the “Wall” for other divers to use as a reference.

I was really running up against a proverbial “wall” and needed to figure out a plan. I decided that finishing the tower would be a good use of time and, if I could do it, would be amazing. My dive plan was to spend a few minutes poking around the Combat Information Center (CIC) since it was reachable via the bomb elevator right by the island, and then finish the island.

I somehow missed to redpoint the CIC and gave up after a little bit since I had the bigger objective in mind.

I went to the island and started to shoot photos and make my way around the island. The section towards the stern is especially challenging given the ropes that hang down.

The big question is whether you go around the outside or inside. That decision changes also depending upon the profile of the obstruction.

However, I had a bigger problem looming that I was aware of but hadn’t fully planned for: running out of memory card space.

I started making “laps” around the island. It was a little challenging to keep track of exactly what I had taken pictures of, but was managing pretty well. I was also trying to manage my run-time and finish the project. All of a sudden, I got an alert on my camera: I was out of memory card space. This presented an immediate problem since I wasn’t finished with the project and it was my last dive on the Saratoga.

I’ve mentioned in the past how I generally shoot photogrammetry models when using my Sony a7rIV that has two memory card slots (see the section on Camera Details). When I shoot photos in this mode, I generally get 2100-ish RAW photos and 2800-ish JPG photos. My camera had run out of space for RAW photos but still had space for additional JPG photos.

I paused and switched around settings to shoot only in JPG. An interesting thing happened: the white balance / color of the photos changed drastically. Below are two photos in sequence where you can see the difference:

Those two photos are literally less than a minute apart. I haven’t researched the reason, but the color balance of the in-camera JPG seems to be different when shooting in RAW+JPG vs just JPG.

I finished up the dive knowing that I was really limited in the number of additional photos I could take. I also figured that I could fix the color issues in post-process. After I got back, I started the alignment process hoping that I could get a full model. I had to start being a little “picky” about the number of photos I shot at the top of the island.

It ended up being 3,106 photos.

Post Process Color Correction

Clearly, there is a color difference on the top section of the island in the picture above. Part of the reason is that it is shallower and gets more light and color but it is more than that. I imported the “JPG only” photos to Lightroom and made some global adjustments.

The entire model was shot without artificial light so my goal wasn’t to color-correct the photos. It was simply to get a uniform color tone. I then built a new texture based on the photos. Below are the results of the final model.

It isn’t perfect by any definition, but it isn’t too bad for a “side project” that took place over the ends of 4 different dives and no artificial light.

Leave a Reply