Saturday, 30 July 2016

Festival of Archaeology 2016 contribution: Have a Go at Photogrammetry!

To celebrate the end of the Festival of Archaeology 2016 (brought to you by the Council of British Archaeology) I've dug out some files I've been working on a side project to demonstrate how easy it is to get involved in archaeology, even from your bedroom (which is where I'm writing this from!).

My side project has been experimenting with photogrammetry to see what changes occur to monuments over time. What is photogrammetry? Well, according to a quick google search, it is "the use of photography in surveying and mapping to ascertain measurements between objects". Historic England recommend using terrestrial photogrammetry (as opposed to aerial photogrammetry) when you want more than 1,000 points per object (for reference, consider that an image from a camera can contain millions of pixels) and the object is less than 100 metres in size. Traditionally archaeologists used aerial photogrammetry, where photographs taken from the underside of a plane are placed side by side, and seeing if they overlap, creating a much larger picture of a landscape and comparing features from different times of the year. A bit like a giant jigsaw puzzle! This is still done today sometimes. However, in the last 20 to 30 years, the digital revolution has allowed digital images to become a lot more detailed, and software has become very sophisticated too, and they can start to analyse these images with algorithms, allowing the process to be automated. When you combine these ideas together, you can start to see the potential for photogrammetry, particularly applications like Google Earth, which now have a large number of 3-D models created by photogrammetry. These are often buildings, which have been "rendered" by specialist software from these images and then exported into Google Earth, so you can see buildings that are both in the real world (what you and I see) and the virtual one (i.e. Google Earth)! In these 3-D examples, which I will show you my example later, the user doesn't have to be in the air. They don't need expensive cameras (but it helps!) and you don't need expensive software. All you really need is a steady hand, a good eye for detail and some time for processing the images!

But why photogrammetry? It is cheap and relatively easy to use, unlike laser scanning or other means of digital recording, but why should we photograph everything and put it in a 3- dimensional digital format? Well, there are times when you need to preserve a monument or archaeological resource digitally because it is about to be destroyed, whether that is commercial development (such as housing), vandalism, pollution, or extreme weather events. For most objects in Britain hopefully this won't happen! If we destroy something, then it isn't coming back, particularly if it holds a lot of value or information about something, such as a Roman military collection, or a rare set of Saxon gold. A digital model can also help us with new interpretations about the past, and answer questions about how an object was made, or what materials were used, or even who made it! A very detailed model of a stone axe could show cut marks that reveal how it was cut or made that are difficult to see with the naked eye. You could even use it to create a mesh (to represent a geometric object as a set of finite elements) that can be 3-D printed, and then used as an educational resource.

However, you should consider whether it is the right thing to do by conserving an object digitally. Not everyone feels that objects should be conserved. Cultures around the world deliberately destroy things, even though they may be very valuable, because it is their tradition to do so. Even you might destroy things to forget about them! In a digital format, it is much harder to delete an object permanently because it is so easy to create copies, especially if it is on the internet. Of course, above all else, photogrammetry shouldn't really be used on its own (unless other methods are not available to you); it needs to be used as part of a wider project to achieve maximum benefit. You should, for example, write down what you see in an object (such as distinguishing features), measure it yourself before survey (if possible) and make use of other resources (such as Historic England's list of listed buildings and scheduled monuments). Think of what you want to ask about a monument or archaeological object and then see if photogrammetry is a good idea.

By coincidence, it is also 100 years since World War One, and there is a monument dedicated to the soldiers not far from where I live, so I have decided to see if there are any changes to the monument over time by creating models at different times of the year. This could help inform other archaeologists and conservation specialists what work needs to be done to the monument, if any, such as cleaning the monument, repairs, etc. However, if you can think of any questions that can't be answered by everyday techniques, such as (for example) "what is the volume of metal in this monument?", you could measure the space in the model when it has been edited and use to to work out how much metal was needed to make the monument.

What you need:

1 x camera (mine is a Vivtar Vivicam, it's rubbish but its cheap). Even your phone will do if you can download the images to your computer!

1 x means of recording what it is you're recording, when you recorded it, and how you recorded it (or the metadata of the project)- usually a spreadsheet on a laptop but pen and paper is fine. Just don't forget to do it!

1 x object to take images of. It doesn't even have to be outside, you can practice with things inside your house!

1x software package- I'm using Autodesk's 123D Catch, but there are lots to choose from. As with a lot of programs they all require a certain amount of playing around to achieve the best results. This is where the magic happens, and your "capture" becomes a mesh you can edit and make videos out of it. A mesh is used because it can be edited and processed, then later exported for other applications.

Take pictures of your object. Now, this is where you have to be careful, because you can't just take a few pictures. You have to make sure that you have a lot of overlap between the images, and a lot of zoomed-in photos to capture details! You need to find what are known as "reference points"- points that are very clear in a lot of you photos, such as an irregular edge, a different coloured part of your model, or some text that is distinguishable in multiple photographs. You can have a go at making you own reference points (such as small archery targets on cardboard) and place these around the monument/object if it helps. Even though the monument I've used it less than 5 metres high by 2 metres wide and deep, I've taken at least 50 photos of it for my "capture". Needless to say, be careful of what is around you (such as steep drops, cliffs, bins etc) to avoid injury while doing fieldwork! Other factors to consider are the direction of the sun (right angles to the sun are good, directly into the sun or in shade will make a difference with poorer-quality cameras). One of the advantages of using photogrammetry software is that they know where photographs were taken, relative to the monument. However, it is worth noting where you took the photos by making notes on a map or a spreadsheet, which will help you when it comes to processing the data.

Location of the War Memorial in the nearby cemetery (Copyright:Historic England 2016). You could use a detailed map like this to show where you took your photos.

NOTE: always ask permission to take photographs if the monument or object is on private land!


When you have taken enough photographs and you are happy with the quality of them, then you can download them to your computer and edit the photos with the software. If you have taken you photographs at different angles, you may want to adjust them so that they are all in the same angle (with Paint.Net or a similar graphics software). Now, a number of the photographs may not necessarily register at first; these may have to be stitched manually, but the software should give you some help with this.

The Results:
The end result before deleting points.

The War Memorial in a photograph... compare with the above model!

On the plus side, you can tell it's a war memorial, with the writing being clearly visible and the statue of liberty (not sure, I think it's liberty) being obvious. However, there are a few issues with the Thornton War Memorial model. Mostly its because the angle of the sun made the statue shine, and the shininess of the surface made it hard for the software to work out what was statue and what was sunlight! So sometimes you will have to delete bits which are not "correct" or representative of the model. In particular, small details that are quite far away, such as the outline of the wreaths, proved to be difficult to capture in the software.
Also, the ground surface was not as lumpy as it is shown here. This is because there was a lot of foilage around the base of the monument that  interfered with the algorithms in the software, so it just shown a very lumpy surface, whereas in fact the . One of the other problems I had was trying to get an aerial photograph, as the top of the monument is about 5 metres high and I don't have a ladder!

Bits in red are the areas that the software couldn't tell were part of the model or not. A quick check showed that they were caused by the sun's reflection on the metallic surface.

The quality of the image depends on many factors, such as the number of photos taken and the angle of the light source.

One of the many things that you can do with the completed mesh is export it for use in an animation (123D Catch has its own animation suite), and for use in other programs that could flesh out the model properly. An animation can really sell your model to an audience, particularly if you use a good flight path to highlight areas of detail or interest and overlay it with audio. You could export the mesh to a software program that can 3-D print your model if you want to! this would allow you to examine a scale model of you model, without having to visit the real thing, which could be difficult to access or has been destroyed. It could even be used as part of a video game!



This bit is often overlooked and doesn't get published (no one wants to see 5,000 data entries on the same thing) but it shows that you didn't copy anybody else's work, you followed correct procedure and showed what software you used so that other people can repeat your surveys and attain similar results. The Archaeology Data Service provide an excellent format for you metadata if you use a spreadsheet. Don't forget to record the position of your photographs!

Overall, there is a lot of room for improvement. The Thornton War Memorial is not quite as jagged in the model as she is in real life, but with better conditions, and better equipment, much of the quality of the model can be overcome. In short, always be prepared! However, compared to even just a few years ago, for a free software package, I am impressed with the quality of the model in some places, and I look forward to creating a second model in a few months time to compare the difference in the models (although I suspect the differences will be caused by the angle of the sun, causing glare).

So, if you want to have a go, just look at this wikipedia page for some free photogrammetry software! A lot of these provide sample meshes to play with. However, what you need to do before you take your photos is to think about why you want to preserve or publish a video about this monument- are there other ways that you could achieve this, or has someone already done it? Other ways might be better, other ways might be more expensive or time consuming.

No comments:

Post a Comment