Introduction
This has been a long time coming …
I’ve been thinking about building a photogrammetry model of the USS Emmons for years. It is an epic wreck that is on the edge of recreational diving in terms of depth. It is, in my opinion, one of the best US Warship wrecks within dive-able depths.
I did 5 dives from 8 Jun – 12 Jun 2025 and took 7385 photos to make this model. It is unheard of to get that kind of continuous window to dive the USS Emmons. It is usually only dove a couple times a week (primarily on weekends) and when the conditions permit. To get 5 straight days of diving the Emmons (Kurt had 6 in a row) is very unlikely.
As background, I wrote a Trip Report post and a detailed post about the USS Emmons. Both are good reference materials and background for this post. This post will focus on the results of the photogrammetry model itself.
I’m planing to write a follow-up post that dives into the gory details of the process to build the model.
Panorama View
One of the things I wanted to do was provide a “panoramic view” of the wreck with detailed photos for some of the key points. Given that you can’t get all of the wreck into a single photo, one way to do this is to base it upon a photogrammetry model. Below are two of the views I made of the wreck with key points noted.
My wife quickly pointed out that my first attempt was “upside down” and didn’t make much sense. 🙂
It is a good angle from a perspective of understanding where things are, but she had a very good point so I produced an alternate version which has the ship “upright” and makes more sense:
At full resolution, the panorama is 6 feet x 3 feet (72 pixels/inch and 5184 x 2592 pixels). I had a version printed at FedEx/Kinkos that is full size that I mailed to the Emmons Association.
Photogrammetry Model Details
The model is comprised of approximately 7,500 photos that were taken primarily across four consecutive dives (I took a few extra for one specific area on the beginning of the 5th dive). People often ask how long it takes for the computer to build a model like this so I will provide some quick statistics. This is for a reasonably powerful machine (Mac Studio 2023 Apple M2 Max with 12 cores and 64GB of memory).
- Photos: 7385
- Alignment (3.97M alignment points, generic preselection, key point limit 9,000, tie point limit 3,000) : 62 minutes
- Depth Maps (High Quality, Mild Filtering) : 13 hours 53 minutes
- Solid Model : 8 hours 43 minutes
- 16k x 16k Texture (Blending Mode = Average) : 11 hours 22 minutes
The total compute processing time is basically 35 hours.
It takes about 25GB of RAM to fully load the model and 16K x 16K texture.
I will go into a lot more details in a “behind the scenes” follow-on post, but my original plan was to start on the hardest parts (debris field) and work my way towards to the stern / amidships and then go to the bow. I made a real-time decision to change that plan and start at amidships and cover the bridge, then the bow, then the hull, then the deck and then cover the debris field. I’ll go into details of the “why” on a separate post.
Here are the coverage areas for each dive:
If I had the chance, I would re-do Dive #2. It was a lot of current on that day and I chose to stay close to wreck to avoid being blown off and to avoid the current. The result is that you can see the “tracing” effect of my lights on the hull as I stayed close. In my follow-on post, I will detail what I have done to reduce that effect in the final model, but I wish it wasn’t there.
Metashape allows you to see how well you have covered an area (confidence model) and it is generally pretty good on this model. I will always want “more” but it is good given the constraints. Blue is good coverage and red is minimal coverage.
Kurt took a couple videos of me during our 3rd dive. They are near the stern area and the first is around the huge winch and the second is at the stern section — both out in the debris field. It gives a sense of scale of how big this wreck is.
Model Screen Captures
Below are some screen captures of the model. On any model of this size, it is hard to pick just a few angles and screen captures.
On-Line Model Link
Below is a link to the on-line model. Note that this version is relatively “low resolution” and contains 8M faces compared to 330M in the full resolution on my computer. However, it does provide a really good overview of the wreck. Note that this model might not render on older phone with memory constraints:
Diver
In the past, I have built a diver into a mode to show scale of the wreck. Kurt and I tried to do this and it didn’t work out as I had planned. Without going into a lot of detail in this post, previous photos occluded Kurt even when I removed them from test models. However, I did make a model from just those photos and thought it would be cool to show it here. Kurt is by the mines at the stern in this model.
Model Screen Captures
These show the area both with and without Kurt to give the reader an idea of scale:
Model Link
Below is a link to the on-line 3D model of Kurt by the depth charges:
Atomontage Model
I have been working with Atomontage for a couple of years. They are developing some very interesting technology to display photogrammetry models on-line with very high resolution. They put models into a “montage” and have done that with a few of my models and recently added the USS Emmons.
Below is a link to the URL in case you want to investigate the technology.
As you zoom in, it renders the specific area in higher resolution. They also have a feature that allows you have the effect of “real life” so you cannot see the entire wreck when you are close (they call it “fog”) and it simulates what a diver would actually see when you are in the water. You can see that effect in some of the screen captures I’ve included.
https://client.atomontage.app/experience/wrecked-in-my-revo
Here is a screenshot of all of my models in the “montage.” It includes the Yukon stern (which also has the engine room in it), the USS Moray, the USS Emmons and three airplanes.
Below are some screen captures of the USS Emmons and then a link to the online URL.
Time Comparison
Wrecks are constantly changing. I’ve highlighted that in previous posts including this one on the local UB88 WW 1 German Submarine. Luckily, the research team in Japan that produced the excellent Emmons article in the International Journal of Nautical Archaeology. I believe that the phots used for the model were taken in June 2015 (exactly 10 years ago). Since then, there have been numerous changes in the wreck but the two most obvious changes are the crack in the hull near the bow and the collapse of the twin 40mm gun mounts near the stern.
I took screen captures of their online model on Sketchfab and compared to my on-line model to show the differences.
40mm Guns
To orient, the 5″ #3 Gun is on the left on each of these near the back of the raised platform. The port-side 40mm gun can be seen pointing skyward in the model from photos in 2015 and can now the underside of the gun platform is visible with the 40mm guns buried in the sand. Most of the raised deck has collapsed.
Port Side Bow Crack
This change is even more obvious. You can clearly see that the current and time have taken a serious toll on the hull and that the crack has been greatly expanded.
Part of the reason I wanted to create this model is that the wreck is starting to deteriorate and, at some point, it will accelerate and we won’t have an accurate record of the wreck.
Thanks To…
Once again, I owe an immense amount of thanks and gratitude to both Kurt and William. This project would have never been possibly without the full support from both of them. Behind the scenes, Ben Lair introduced me to Mike Boring who then introduced me to Kurt. It is, indeed, a small community with very few degrees of separation.
References
Assessment and Significance of a World War II battle site: recording the USS Emmons using a High-Resolution DEM combining Multibeam Bathymetry as SfM Photogrammetry in the International Journal of Nautical Archaelogy
Too cool for words, Brett. Well done!
Thanks, Ken. Much appreciated!