Introduction
I have visited the PBM Mariner off Point Loma several times over the years. On one of those dives, I created a photogrammetry model of the wreck, and we also conducted exploratory dives that located additional debris areas. Resting offshore in about 100 feet of water, the wreck is both historically significant and visually compelling, making it an ideal subject for detailed photographic documentation.
My first dedicated photogrammetry mission produced a model with good overall structure, but both of my lights failed during the dive, leaving the final result with a strong green cast. Since I plan to include this wreck in the upcoming book, I wanted to return and document it again—this time with better color, better completeness, and a more faithful visual record.
The first model captured the wreck reasonably well, but the color cast obscured some of the detail, as shown below.
Back on the Wreck
Surface conditions were less than ideal. A large swell was building in Southern California, and we were already feeling it by the time we headed out. The ride was choppy, but I was glad to be back on site with Matt B., Tyler, and Ryan aboard the Marissa with Lora and Chris.
Matt and I dropped in together, and my plan was to spend about 40 to 45 minutes on the wreck while keeping total runtime under 75 minutes, with about 20 to 25 minutes of decompression built into that.
Underwater visibility was fair—better than on our first dive here, but nowhere near the conditions I had when I built the earlier model. It was manageable to keep track of the downline while I was on the wing, but once I moved across the open sand toward the tail, navigation became more difficult. Even so, I was able to cover most of the debris field and make it back to the wing and the downline with about 20 minutes of deco after roughly an hour in the water.
I also had some equipment issues during the dive. It was unusually difficult to maintain loop volume, and I felt as though I was constantly leaking gas and adding diluent. On ascent, I heard a slow bubbling sound and initially thought it might be coming from my loop. Matt pointed out that it was actually my wing inflator; after I moved it around, the issue stopped. I still do not know what caused the loop-volume problem, but it added one more complication to an already demanding dive.
Side-by-Side Comparison with the Original Model
Below are several side-by-side comparisons between the original model and the new version. Seeing them directly, I was struck by how much of a difference the new pass made.
Prop Close-Up
This close-up shows one of the clearest benefits of improved lighting. The added color and detail make the prop stand out much more clearly, and the slider highlights the difference well.
Wide Angle Prop and Wing Section
This wider comparison shows the front portion of the wreck before and after the new documentation pass. Along with the improved color, the increase in overall model fidelity is easy to see.
Model Completeness
Another improvement is the completeness of the wing on the port side, which appears on the right when looking forward because the wreck is upside down. In the first model, less than half of that wing was captured. In the new version, the entire wing is present.
Model Fidelity
The new model is also noticeably sharper and more detailed. I thought the original version was good at the time, but that was nearly five years ago, before I had much more experience with both image capture and processing. In particular, compare the definition of the engine near the top and the propeller in the lower right corner between the two versions.
Overall, I am very pleased with the new version, and revisiting the project for the book was well worth it.
The Biggest Technical Challenge: Color Consistency
The dive itself was only part of the challenge. Once I had the photos, the bigger technical hurdle was creating a consistent texture—the photographic skin wrapped over the model’s 3D geometry. Even when the shape is accurate, inconsistent color across the image set can make the finished model look patchy or unnatural.
This version of the model used the JPG files straight out of the camera:
In particular, note how the color “temperature” changes in the areas circled in red.
I had my white balance set to Auto, which is how I normally shoot. During the dive, I noticed the in-camera JPG preview shifting between different color palettes. I am not entirely sure why it happened, but the camera appeared to be switching back and forth at several points.
Below are two photos I shot less than a second apart. The camera produced these in-camera JPG versions from the RAWfiles, and the shift in color temperature is obvious.
The fix was to import the RAW files into Lightroom, apply the same color temperature across the full image set, re-export them as JPGs, and then build a new texture. The comparison below shows the difference between the inconsistent version and the corrected one.
A Closer Look at the Finished Model
Below are several screen captures from the new model that will appear in the book, along with a link to the online 3D version.
On-Line 3D Model
The 3D model below can be moved, rotated, etc.
More than just a better-looking reconstruction, this new version feels like a more accurate and useful record of the wreck. That is what makes the effort worthwhile for me: creating a clearer visual document of an important underwater aircraft site before time, conditions, and corrosion alter it further.
Acknowledgements
First, thanks to Steve Lawson and Chris Gilmartin for finding and identifying the wreck. I also appreciate their willingness to share the location five years ago, trusting that we would keep it private—which we have—and document and photograph it responsibly.
Thanks as well to my dive partners, who have helped with both this version of the model and the earlier effort.
And thanks to Chris, Lora, and Marissa Charters. It is always a pleasure to dive with them, and I know I can count on a well-placed downline and the boat being there when the dive is over.
