The Brandy Wreck – Then and Now

Background

Way back 2011 Grahame Knott of Deeper Dorset found a small wreck site in Weymouth Bay. At the time the search was on to find more of the P47-D Thunderbolt but in true shipwreck hunting fashion Grahame found another mystery. Known as the Brandy Wreck, the site had four iron guns, two anchors, a millstone and ballast stones.

At the time both Grahame and Simon were wondering how it might be possible to create a photo mosaic of a wreck in the gloomy waters of the English Channel. The Brandy Wreck was small and around 400 images were shot with the intention of making a mosaic. Five days of Photoshop head scratching later and a mosaic duly appeared. The process was tedious in extremis and not very accurate. The idea of doing more wrecks was put on the back burner.

The First Model

Fast forward to 2015 and Simon started to explore the potential of photogrammetry. With little opportunity to shoot specific images Simon turned to the catalogue of existing work and ran the 2011 Brandy Wreck images through Agisoft Photoscan. To his surprise the images worked and a 3D model revealed itself. At that moment traditional 2D photography died. The future was 3D.

Here’s the first model:

A Retrospective Critique

The first model is fine, but it lacks:

• Black and white – Lacks lifelike colour.
• Incomplete – Missed off  two cannon.
• Was not very refined and lacked detail.
• Was not scaled/could not take measurements from.
• Did not have embedded GPS data – no DEM and no georeferenced ortho photo.

So whilst it was nice to look at and great for diver briefings (we viewed it on a phone last week!) it wasn’t much use. But it was the only permanent record of how the site was in 2011.

The ortho photo of the site did prove useful in 2016 when Wessex Archaeology surveyed the site on behalf of Historic England. Based on a few finds the conclusion was the wreck probably dated back to the English Civil war and was therefore mid-17th century.

A Second Scan

Earlier this week the chance to dive the wreck again presented itself. Close to Portland Bill the site is sheltered from the westerly winds and will become a project site to investigate when the weather keeps Deeper Dorset from other targets.

Scanning such a small site needed just 680 images and took 12 minutes (Yes, 12 minutes) from beginning to end. This included the 4th cannon lying slightly off the site.

Here’s the result:

And this one has everything we now expect:

• Lifelike colour.
• Complete, scaled with accurate measurement.
• GIS reference.

And from this the ortho photo and DEM takes minutes to derive.

Did you really say 12 Minutes?

From the time stamp in the camera the first image was shot at 15:16 and the last one at 15:28. This has covered 222 square metres of Weymouth Bay. It took longer to descend, find and lay a guideline the site, send up the delayed marker buoy and to complete decompression stops. Simon did have an advantage as he knew the site, but to claim it would take longer is disingenuous.

Site Changes

Between 2011 and 2017 the site has been dived maybe once or twice a year on average. We were not expecting massive changes but having examined the 2011 model and compared it to the 2017 version something significant has gone missing – the first to use the contact page and drop an email to Simon saying just what has vanished may well win a small prize…

With these kind of baselines, photogrammetry has the potential to become a repeatable, highly efficient method to accurately record changes in a site.

What next?

Things have come on in leaps and bounds since that first model.

GPS/GIS is a foregone conclusion. The size of the sites that can be scanned has gone beyond what was thought possible. Ortho photos with scale are standard. DEMs are done just because we love the art and the Heart of Gold adds efficiency to massive subjects while cloud computing can churn away eating massive data for breakfast.

But we are not resting on this. There are plans in the pipeline to step things up a gear, to develop new ways and increase efficiency even further. Every bottleneck is under review and will be worked through or around. Watch this space…