This is my 200th post and I wanted it to be somewhat special so I am posting what I consider one of my best photogrammetry models of one of my favorite wrecks.
Many might not understand why the USS Hogan is so appealing to dive — after all, there is little left of the structure of this Wickes-class destroyer. I previously posted about the wreck in general along with some photos I recently took while creating the photogrammetry model.
I don’t remember exactly when I got the idea to build a model of this wreck, but it was quite some time ago. I knew that I needed more practice due to the “blank spots” in between areas of the wreck and lack of consistent structure with which to align photos. I also figured that it was an ideal target for a model since it is spread out and hard to get an idea of the “big picture” with only photographs of individual sections.
This post is going to be somewhat long but has some very interesting information and a ton of details about the challenges of building the photogrammetry model.
Below is a “Table of Contents” with links to the different sections:
- The Model
- Screenshots of the model
- Identifying areas of the wreck
- Photogrammetry Model Details
Instead of creating a link to the model at the end of the post, I’m going to put it near the top in hopes that people get to experience the benefits of an on-line 3D model.
If you click on the link below, it will take you to Sketchfab which is a site that hosts 3D models. You will be able to move around the wreck, zoom in, rotate, etc. all on-line from your browser. If possible, I recommend doing this full screen mode from a laptop or desktop computer vs. a mobile device. The extra screen size will help and sometimes big, detailed models don’t render on mobile devices.
For those who don’t want to navigate to the on-line model, here are some screenshots of the final model.
Identification of Wreck Areas
One of the real benefits of a photogrammetry model is to see the “big picture” and identify ares of the wreck in relation to others.
Often times, a picture doesn’t really convey the object in total. Below are some of the interesting and noteworthy features of the wreck, along with photos of the feature and original blueprints.
First, below is an overall “top-down” view with those key points noted along with a blueprint of sister ship DMS-5. Each numbered area will be explored in depth in the sections below.
1 – Stern
As I covered in the previous post, the stern of the Hogan is very iconic. It is completely covered in strawberry anemones and has been “flattened” during the conversion to a destroyer minesweeper from the original profile on the destroyer. Here is a “side by side” of a photograph and the model I built.
I have also included blueprints of the original stern design on the destroyer (left bottom) and the “flat” design of the minesweeper (right bottom). The flat design was needed to aid in the deployment and recovery of the minesweeping equipment.
2 – Prop Guards
The propellers on the Wickes-class stuck out far from the profile of the ship — you can see how far out in the blueprint. Therefore, they needed props guards to prevent hitting the prop into the pier when they docked. Below are a few sections from the original blueprints for this class of destroyer along with a screen capture of the model and a photo of one of them. I’ve also included a photo of a sister ship that shows the flat end but also the prop guard.
3 – Crew Wash Room & Mess
Interestingly enough, the section in front of the stern was always referred to as the galley. I believe this is because tiles and fixtures were found in that structure.
However, according to the blueprints, it is actually the crew mess and water closet (bathroom). You can actually see the note “Ceramic Tiling” in the blueprint for the washroom area. Tyler mentioned that he has also seen a sink in that area so it makes sense.
4 – Condenser, Reduction Gear, Turbine, Prop Shaft
This is a great example of how you really can’t see the “big picture” until you built a photogrammetry model.
You can see the individual “bits and pieces” but not how it all fits together. Below are a couple detailed photos, screenshots of the model, and the blueprint for this area of the ship. You can see the prop shaft running back towards the area where the prop would be right behind the crew wash area.
5 – Boilers
This was one of those mysteries that was solved / resolved. One of the things I noticed when I initially built the model was that there were only three boilers. What happened to the fourth boiler for the “four stack” destroyer? Maybe it got destroyed when it as bombed? Maybe it floated away to somewhere else? Maybe it got crushed? For various reasons, all seemed unlikely.
I reached out to Steve Lawson to see if he knew anything. He ended up finding a picture of the Hogan after it was converted to a DMS and….it had three smokestacks (see picture in the upper right corner). I then did some more research and, sure enough, they had removed one of the boilers from the original destroyer configuration. Mystery solved.
Below are some pictures of the model, the boilers themselves. some blueprints and the photo from Steve Lawson. Also note in the diagram below that the two boilers that are “together” is towards the stern which is reflected in the photogrammetry model.
6 – Top Deck of Galley House
Right near the boilers, there were some round “discs” that I thought might have been end caps for the boilers or somehow related to them. I started looking through the blueprints for any similar structures/patterns and I think I have identified what they are. The blueprints show guns mounted on top of the galley which is right on top of the boilers. The shape of the model and the photos is exactly the same and the location matches perfectly!
7 – Bow area
The bow on the destroyers is very narrow. Most of the bow is collapsed, but the very front area is relatively intact.
Photogrammetry Model Details
This section will provide an in-depth guide to the process of building the photogrammetry model. I’m going to go into a high level of detail with hopes that it can help others.
I shot the photos for this model with a Sony a7rIV with the 28-60 lens and the Nauticam WWL-1B wet-lens. All photos were taken at 28mm which gives a field of view of 130 degrees with the WWL lens attached.
This camera has two UHS II slots for memory cards. The two slots can be configured in different ways.
Previous to these dives, for shooting photogrammetry models, I had two 128G cards and saved RAW + JPG to both cards (i.e., mirror images of each other in case one fails). The reason for saving both RAW and JPG is that I’m lazy and usually use JPG straight out of the camera. Other people shoot only RAW and then use something like Lightroom to convert to JPG after making some global changes (white balance, etc.).
For these dives, I decided to try a different approach to maximize the efficiency of the two card slots but still have a backup.
- The first slot would have a fast (V90) SDXC card with 128G capacity and I would write compressed RAW to that card (about 64MB per picture). That would give me about 2,062 photos.
- The second slot would have a semi-fast (V60) card with 64G capacity and I would write JPG to that card which would give me about 2,810 photos.
The advantage of this system is that I can use a second, cheaper card in the second slot and still have redundancy. If the first card slot fails, I still have my standard JPG files. Not as good as RAW, but for what I’m doing, it is acceptable to me. If the second card fails, I can take the RAW files and convert them to JPG to built a photogrammetry model. In the normal case, I get both RAW and JPG and have a ton of flexibility.
I did two dives on the Hogan to build the model.
The first dive was on 15 Oct with a group that had chartered the Marissa but had a spot available. I had a flight to catch the next day so I had to get out of the water with enough surface interval. My dive was a little over an hour and I took 824 photos of the wreck. The visibility was so amazing on this dive that I thought it would interesting to take photos without any artificial light. The photo below is pretty much straight out of the camera.
My second dive was conducted a month later on 12 November and Drew Wilson joined me on the Marissa. The conditions were not as good and I had two primary objectives for the dive: (1) to fill in the gaps at the “tops” of the structures and (2) to add some color to the model. My dive was 98 minutes long with about 40 minutes of that on deco.
As Drew and I have done in the past, he swims or scooters off of the wreck and takes natural light photos of me illuminating the wreck with my video light while taking pictures for a photogrammetry model. Here are a few of his pictures:
On the second dive, I took 1,338 photos for a total of 2,162 photos.
Build Process Details
After Dive 1, I built a preliminary model on “low” accuracy but only 278 of the 824 photos aligned. You can see one of the boilers on the left in the sparse model below and the bow to the right but the rest didn’t get aligned. Hmmmm…
At this point, I thought it might have been a big mistake to only shoot with natural light at 120 feet deep. The photos were a bit dim so I thought about post-processing them to bring up shadows and exposure. Instead, I ran a model with Medium accuracy which produced the model below with 823 of 824 photos aligning.
I knew during the dive that didn’t get the “top” sections of the boilers or the stern as I was just running out of time. You can see the blank spaces on the model screenshot below:
I was pretty happy with this model after a single dive. It was a bit dark and would need some additional exposure, the tops of the boilers and stern were missing, but I had good coverage of the debris field for a single dive of a 300+ foot long site with a lot of “blank” space.
After Dive #2, I had a total of 2162 photos. After running alignment with Medium Accuracy, I had 2153 of the 2162 photos correctly aligned and the sparse cloud looked pretty good:
In the capture above, you can start to see the color added by the photos from Dive #2 and it looks like the top of the boilers and the stern deck are filled in. So, I ran a Medium Quality model process. Below is the result.
Oh No! I still had some “blank spaces” on the model.
I was a bit despondent and thought I would have to schedule another dive to finish the model. Not that I mind diving the Hogan, but it is a long trip and boats don’t often go there.
I had another idea — what if I built a model with High Quality? I didn’t want to necessarily built the WHOLE model on High Quality without first trying to get an idea if it would actually solve the problem. I had an interesting idea: reduce the “region” to be just the stern area and see what happens.
Now that I knew that it would work for at least the stern section, my fingers were crossed for the boiler section and I started the build process for a High Quality model. It wasn’t fast, that is for sure.
At this point, I ran the tool to estimate the quality of the photos and disabled all those with a quality “rating” less than 0.4 (manufacturers recommendation is 0.5) and ended up with 25 cameras disabled, 9 not aligned and 2128 photos in the model.
Here is the top down view of the final model before I made some adjustments to increase the exposure and “lighten” it up.
The model was a bit “dark” due to the natural light photos from the first dive so I used the contrast and brightness tools in Metashape to lighten it slightly.
Model Build Time Stats
- Matching Time (only additional photos) – 2 hours 3 minutes
- Alignment Time – 23 minutes 39 seconds
- Depth Maps (2125 cameras) – 6 hours 19 minutes
- Dense Cloud – 4 hours 1 minute
- Mesh (High) – 1 hour 30 minutes
- UV Mapping – 2 hours 57 seconds
- Texture Blending – 1 hour 31 minutes
As usual, Lora and Chris @ Marissa Charters were awesome and got the downline right at the amidships point and are always so great to do technical diving with.
Thanks to Tyler for the continued partnership. He kept urging me to build a model of the Hogan and I think gave me the initial inspiration to take on the project. I’m also pretty sure the first time I met Tyler it was while diving the Hogan.
Thanks to Steve Lawson for help with the research into the 3 boiler mystery.
Thanks to Drew Wilson for his help on the dive and his great photos.