Imagine you are walking around a South Indian town – say Kāñcīpuram – with the printed edition of your favorite māhātmya – I am sure it will be the Vaiṣṇava Kāñcīmāhātmya – trying to define some of its sacred geography. You are starting to get lost: all the toponyms mentioned in the texts do not really coincide with the contemporary ones, but suddenly…you see it! A cow is using it as its personal manicure-tool, but that is definitely the piece of art you were looking for. The pedestal that was mentioned in your colleague’s article, an object that might help you date the Kāñcīmāhātmya, since it is the only object mentioned in the text that can be historically contextualized. You take out your camera and you begin taking pictures, a lot of pictures. Locals wonder about your behavior: they stare at you, while you go around an object that has just been used as nail-scissors by a cow. You try to explain to them (and to yourself!) that you are doing it for the sake of research, since you want to do…some photogrammetry.
Such a scenario is not so distant from our daily lives as researchers – or at least not so distant from our time spent conducting fieldwork – and that is why today I am going to provide a very simple introduction to how to perform some basic photogrammetry.
The hardest part of a photogrammetry project might well be the act itself of taking pictures. To start with, depending on the size and complexity of your object, you might want to have between 20 (min) and 200 (max) pictures. 40 to 100 might be enough to keep the software busy for a while.
You should pay attention to various factors while you take the pictures.
- The object should always be kept at an equal distance from your camera.
- You should turn around the object and avoid changing the object’s position: this will help the software to align the photographs better and – therefore – produce better results.
- The light should be stable and you should pay more attention (and take more pictures) to the darker areas of your artifacts – the software will thank you!
- Having a reference-grid under the object, or a surface with a non homogeneous pattern might be of use to align the pictures better.
We are going to use a prebuilt software, Agisoft Metashape, that can be downloaded here: a one-month trial version is available, after which you need to buy a license – or make your boss buy it for you.
Another nice tool is Meshroom (open-source and free) which has different levels of difficulties in usage, providing “just one button” for the casual user and “the extremely finely grained control of every step and parameter” for the experts.
Once the software is installed and ready to be used, start by saving your new project via the “save as” button. In workflow, click on “add pictures” so that they can be added to the workspace (on the left).
On the bottom, you will see the state of your photogrammetry-flow.
Information on the status of your flow.
In fact, at this point, we are actually almost done! The following steps are easy; you just need to tell the software to perform some tasks, which can be done by clicking on workflow and selecting the following tasks, one by one.
When choosing the quality of the tasks, at least for the first trials, keep it on “medium.” Moreover, try to keep your computer light on software, browsers, etc. while performing the various tasks, since they demand a lot from your machine, both in terms of memory and power; if you work on too many things at the same time, the process will be slowed down even more.
Photo alignment and information.
After aligning your picture, you can proceed with the build dense cloud command, that will results in something like this:
Dense point cloud.
At this stage, you can polish your image by cleaning the “noise” around it. The free-form selection tool on the toolbar will help: circle the area you want to eliminate, and then click on the “x” on the toolbar.
After this, you can proceed with the build mesh and the build texture commands, that will get you to the final steps: the dense cloud will be literally meshed into the (almost) final object and the texture will give it the final touch, providing its real texture.
All this, as already said, will take some time but – at this point – you should be able to see your wonderful object in 3D. Here, I have a pedestal and a Buddha head, coming directly from the Puducherry Museum.
Buddha head from the Pondicherry Museum (India).
Pedestal from the Pondicherry Museum (India).
Remember to bring your results to the office tomorrow and, if you wish, share it with us!
One thought on “Once upon a Camera…”