For my PhD I gather information on ceramic sherds of the 5th millenium BC in Brandenburg. These Stroke Band Pottery, Rössen, “Guhrauer” Group and maybe Brześć-Kujawksi Culture wares are decorated with impressions. Therefore they lend themselves very well to documentation with RTI.
RTI? What was that again?
Reflectance Transformation Imaging (RTI) is a method where 64 object photos are taken from one perspective but the light source moves with each picture taken. The RTIbuilder then “merges” these into a “polynomial texture mapping (ptm)”, a data format I can view with the RTIviewer. Within the viewer I can interactively change the light source, use a number of filters and explore my object in more detail than I could with a normal image. The technique was developed by the Cultural Heritage Imaging corporation. They have a great explanation how it works on their website. Also, Sebastian wrote a self-study tutorial for his students, if you’re interested in learning more.
Check out this RTI image of one of my sherds as an example to understand better. You can change the light by clicking on the surface and dragging the mouse.
Nice, eh?! 😀
In March and April I borrowed the Kölner RTI-Dome from Sebastian and ran my first RTI campaign. RTI-domes are fixed installations for the lights (in form of a dome) and a control box that is also connected to the camera. I only have to set the timer and push a button and in the dome the first light goes on – a picture is taken – the light goes off – small pause – the next light goes on – a picture is taken – the light goes off – small pause – the third light… times 64. Easy right? Yes. But still, it needed some fiddling before I got into a good workflow.
Some things I learned
- I did not have any experience with the RAW photo format beforehand, so I was quite stunned to find my 64GB SD disk to be full before the first day was over. RAW takes SPACE — especially if you take not just 64 images per sherd, but some more, because you’re still fiddling with the settings. 😀
- I had been told to use the Aperture Mode for the image taking. In this mode, I set ISO and F (aperture) value and the camera decides the shutter speed by itself. I noticed early on that this led to quite different images. Some were perfect, but every three or four photos I could hear by the very long pause between the click – – – – click of the shutter, that this image would be way overexposed. I tried a lot. In the end I switched to Manual mode and set the shutter speed to a fixed 1/10, which seems to have worked fine.
- The F value will define the depth sharpness. For more rounded sherds I had to use larger values than for “flatter” ones, otherwise the edges would blur.
Time management and post-processing
Once I had my workflow, it took me about 5min per sherd to take the 64 pictures, so I managed to document up to 30 sherds a day. That was much more than I anticipated. Sadly, I did not manage to check all of them for quality in the evening and now I sometimes find one that didn’t turn out perfect. But that was to be expected and I plan to do a second “RTI-round” later this year.
Now, I use RawTherapee to post-process the images. I correct the angle and white balance and export the images to jpeg. That’s the format the RTIbuilder likes. RawTherapee is great, because it has a useful batch process, to which I can keep adding new images (that will go to different folders). It runs on all platforms and is free and open source!
In RawTherapee and in the RTIbuilder the workflows always need a few clicks and then some waiting time, so I try to run this next to other tasks. Multi-tasking is not very efficient most of the time, but in this case… I’m not sitting around waiting while my PC works. 😉
As you can see with my example above, it is possible to upload the RTI images to the Ariadne Infrastructure. This is a great service for 3D objects and “re-lightable images” to which RTI belongs. Sadly though, they are not a long-term repository. For publishing my database at the end of my PhD I am still looking for an infrastructure that will support an RTI viewer and give a DOI to a database entry. It’s something the Edition Topoi Repository can do (see example here), but as the Topoi Excellence cluster ran out of funding I am not sure I will be able to use its infrastructure.
Do you know anyone who offers such a service?
What do you think about the RTI technique?
Hi – nice write up – and I love this use of RTI for pot sherds.
Regarding the things you learned – you ended up with the right solution. It is not a valid RTI image set if you set the camera to anything other than manual. Everything should be fixed, including the focus. The only thing that should change is the position of the lights (which the dome does automatically. That means that some images will be under exposed – if the light is near the edge of the dome (a.k.a. raking light) This is a correct capture.
I’m Carla from Cultural Heritage Imaging – so I think you can trust that I know what is required for a correct RTI 😉 Also – we healped develop RTI – butthe original work came from Tom Malzbender and Dan Gelb at Hewlett Packard Labs back in 2000. There is some exciting new research with RTI – and some new tools that will be announced shortly. You might want to check out the CHIForums forums.culturalheritageimaging.org
Dear Carla, thanks so much for your feedback! It’s good to know the solution I found was the right one. 😀
I didn’t know about the history behind the development, that’s useful, thank you.
Oooh, and new tools sound exciting. I will be checking the CHI website and forum to make sure I don’t miss them.