I have almost the entire photogrammetry model of the Yukon complete — in fact, I had it almost complete after dive 5. However, I’m still facing the ongoing problem with the moving seaweed and not being able to align photos for the starboard side hull.
Below is the “state of the project” after the “Seaweed Snafu” on Dive 6 (note: I didn’t even both building a “texture” layer):
My plan for Dive 7 was a bit more “complex.” It involved running an orange cave line across the hull and then placing “cookies” along the line in hopes that the software could use the patterns to align the photos. What a pipe dream that turned out to be!
The Professional version of the software can use these special patterns and markers to align photos and provide relative distances, etc. and then also automatically filter out those patterns from the model. I don’t have the Professional version; however, I figured I just needed unique patterns for the software to align.
Conditions were reasonably good for the dive. We tied up near the guns on the stern area. Here is the general process I followed to conduct my “Cave Line & Cookies” attempt:
- I headed to the large cutout opening in the stern area since I know I can use photos in that area as an “anchor point” to align photos between different dives.
- I went inside and tied off the cave line and started running the line up the stern to the hull and then across the hull.
- I would stop at different cutouts and loop the line around the cutout to keep it taught and create a “path.” After doing that in about 5-6 different cutouts, I was back at the place we tied into where I tied off my reel.
- I then reversed course and laid 8 cookies along the line as I moved towards the area I tied off.
- After I finished laying the cookies, I went down the stern and took photos of my “anchor point” and started taking photos as I moved along the line and trying to keep cookies in the photos
- I then got all the way back to the reel, dropped off my camera and then started back along the line picking up my cookies and reeling in the line.
- I got back to the stern area, untied my line and then went back to my camera
That whole process took about 40 minutes or so. Below are some pictures from my follies. I had little hope of this actually working and it didn’t. The cookies would flip over revealing just a white disc, but that wasn’t the “real” problem. The line and cookies would also get completely covered by moving seaweed.
Here is a picture of the patterns and some cookies I made:
In the photo below, you can see the cave line running along the hull. If you look very closely, you can see a white cookie far away. The seaweed would cover the line rendering it useless.
The “problem” is that almost of the alignment points are NOT along the cave line and on the seaweed. White dots are possibly alignment points and blue dots are points that are aligned.
Picture of a cookie on the line that flipped over showing only the solid white side.
Cave line reel tied off on the wreck near the downline from the Marissa.
When I got back, I put the photos into Metashape not sure what the result would be. It wasn’t great. I got some of the hull near the very stern covered, but had made very little progress.
I then started alignment from the tie point of the reel heading back towards the stern. The good news is that a LOT of the photos aligned and I thought GREAT. However, when I looked at the sparse cloud from the side, I noticed that a small alignment error had (once again) propagated and was made worse as I worked my way to the stern:
In short, a waste of time, but it was worth a shot. I won’t go into all the details of the other models I tried to build but I spent many days working on it trying to get everything to work.
Below is a picture of the model after Dive 7:
The Engine Room
If you look closely on the picture above, you might notice something “funny” looking in one of the gaps in the hull…
I had some spare time so I decided to start taking photos inside the engine room to see if I could build a model of an inside space. There are some complexities in terms of shooting inside spaces and also how to “display them” but I’m working on those. It was probably the most productive part of the dive. Preliminary results are below:
The photos were under-exposed so the engine room is a bit “dark” at the moment. However, you can definitely see the big reduction gears in the center of the engine room.
Below are a few other views into the engine room. This model was built with 332 photos. The red circle is the cutout that I usually use to enter the engine room (there are multiple ways inside).
Below is an even closer up view:
Below is another view of the engine room from a different angle:
I will spend more time on the next dive getting additional photos to build more models, but the initial results are promising.