Photogrammetry

On today’s episode, host of the Women in Archaeology Podcast, Chelsi Slotten, tells us about the exciting world of photogrammetry. What is it and how do you do it? It’s a great toolbox episode and will leave you curious and wanting to experiment!

Photogrammetry

Transcript:

Welcome to the 365 Days of Archaeology Podcast.  I’m Chelsi Slotten and on this episode I’m going to talk about photogrammetry.

What is Photogrammetry?

Photogrammetry has been around for years but has recently become a much bigger topic in archaeology as the number of photogrammetry based projects at the 2017 SAA’s will attest.

Simply put, photogrammetry is a way of making 3D models from 2D images.  A computer software program utilizes large numbers of photos of the same thing to takes measurements and figure out what the object would look like in the real world.  It works based on the principle of parallax.  Parallax is the apparent displacement of the position of an object when it is view from different vantage points. If you can imagine a tree in front of a house, when you stand in front of the house the tree appears to be in front of the house, but if you stand far to the left the tree appears to the right of the house, that’s parallax.  By having photographs from many different vantage points the computer program can figure out where the tree and house are relative to each other, or where facets of stone tools are in relation to one another, or the size and shape of a ceramic vessel, or how the location of morphological markers on bone relate to one another or any spatial relationship you care to recreate.

Photogrammetry and Archaeology

Photogrammetry is primarily used in 2 ways in archaeology, for creating models of stationary objects, and creating 3D representations of Landscape topography.  You may be asking yourself, but what about 3D scanning, which is an excellent question.  There are several main advantages to photogrammetry.  First, compared to 3D scanners, photogrammetry is cheap.  You can do photogrammetry with a camera, a computer and a license for the software.  If you prefer there are several open source versions of the software so you really only need a camera.  While it’s preferable to use a nice camera with excellent resolution, cell phone camera’s are decent. I’ve seen some really good looking models come from cell phone pictures.

Second, it’s accurate.  While more work definitely needs to be done on this subject early studies indicate that it is about as accurate at 3D scanning, although some of that will vary based on how good the photo is and what resolution you process your images at.

Next up it’s portable, 3D scanners can be large and often require their own power sources.  Finding an outlet in the middle of the field can be difficult, to put it mildly.  With photogrammetry, all you need is a camera and the software, no need to haul cumbersome equipment around.  3D scanners are getting smaller but it’s still a piece of dedicated equipment and camera’s and computers aren’t.  Lastly it’s flexible, which is to say it’s all of the above and more.  One major benefit of photogrammetry is that you can apply it to old photographs (if they’re good and numerous enough).  This is super helpful as sometimes objects have been lost, destroyed, repatriated, damaged, loaned, reburied, or otherwise removed from study.  If you’ve got photos you can build a 3D model and access some of that unavailable information.

Photogrammetry Basics

So moving on to some of the nuts and bolts of photogrammetry.  The first thing you need to do is take photos, lots and lots of photos.  Since photogrammetry works on parallax, the more photos you have, the more information the computer can work with, the better the model.  Although, as with all things, there is a point of redundancy at which more photos will not statistically improve your model.  The general idea is to take photos that have overlap so the computer can identify specific landmarks in several photos and accurately calculate its location.  In order to do this, you generally want between 60 and 90% overlap of your photos.

The following tips will help you take photographs that will process well in the software. First you want to maximize contrast between the object and the background so it’s easy for the software to identify what you want to model. Second, maximize your depth of field, also known as having a high f-stop. Third, use a tripod for stability, although it won’t kill you if you don’t. Use even light to create minimal shadows. The software program isn’t great at dealing with shadows, Lastly, include a scale bar or control points so you can take measurements from your 3D model with confidence.  All of these suggestions mean that, from a photography standpoint, your photos are going to look pretty flat and uninteresting- this is exactly what the computer program likes to work with, but maybe don’t show them off to your photography friends as examples of your prowess.

When taking the photos, there are two major methodologies.  The first is the turntable method, in which you put your object on a turntable and take photos of it at predetermined points.  You do this from several angles, slightly below, straight on, slight above, and significantly above.  Then flip your object over and do it again to make sure you get every side of your object.  The other option is the walk around method.  Like the name suggests you place your object on a surface and walk around it taking photos at predetermined points from several angles.

Once you’ve got your photos you then upload them into the appropriate software and start processing.  One of the major downsides of this system is the processing part, namely that it takes a long time, especially for really high-resolution models.  There are entire classes dedicated to how to use photogrammetry software.  There are also large online message boards where you can ask questions.  If I went into how to use the software this podcast would be another 2 hours long, at least.  If you’re interested in learning how to use the software my suggestion is to go take a class somewhere.

Software

There are several different options for software you can use. The most popular is Agisoft, which is also a paid program, although their student rates are reasonable, and if you need it for a project you may be able to convince your department to help pay for it, so that’s something to consider.  You can also try it out for 30 days with a free trial, but there is some limited functionality and you can’t export models.  Several open source alternatives are also available which have been reviewed by people who do photogrammetry.   Some of those alternatives include Photoscan, OpenMVG, OpenMVS, VIsualSFM, MeshRecon, and MVE.  Not all of these programs have all the same functionality of Agisoft so sometimes you need to mix and match to get your complete model.  You can find links to the reviews of these products in the notes section below.

All in all, photogrammetry is a very useful tool for archaeologists as we move into a digital age, and is a skill worth developing as  it’s uses and applications will only grow in the next several years.  Thanks for listening!

Notes:

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Up ↑

%d bloggers like this: