Having had my first personal foray into digital 3D modelling from photographs last week when I worked on the model of Bradford War Memorial (fortunately I work with many clever people who are experts in 3D modelling), I thought I would give it another try. Oh Dear, did I come down to earth with a bump! It is much more tricky to do than I found last week.
I decided to have a go at making a 3D model of a Minoan pottery kiln (with another job hat on I study Minoan pottery). Partly this was out of interest to see a model of the kiln, but mostly it was to learn more about the software and how the whole process works. My main summer holiday this year was on Crete as my partner had never been there (you may have seen the picture we took of Knossos in the Curious Travellers ‘name that archaeological site’ game we run each week on Twitter and Facebook @curioustravell2). On our trip to the Mesera plain in the south of the island to see the Palace of Phaistos we also stopped at the site of Hagia Triada, a Late Minoan Villa nearby. On that site are the remains of a well-known Late Minoan Pottery kiln and I took quite a few photos to show the key features. I didn’t plan to make a 3D model with them.
My partner suggested I try making a 3D model as I had many photos of the kiln and so I thought I would give it a try. However, when I went back to look at them I realised I had only taken photos from three main directions. This was partly as these showed the key features of the kiln and partly due to the slope of the land and to the protective roof over the kiln blocking some views.
I thought I would try to get other pictures from Internet searches, but it turns out that everyone takes their photos from roughly the same three viewpoints. Undeterred, I tried to give it a go.
My first attempt did show the kiln, albeit very truncated, but mainly showed a fabulous image of the wall around the fire pit rather than the wonderful structure of channels running under the firing chamber in the main kiln structure. Clearly the wall was the most distinct feature in the photographs, and so it is what the software picked out. Although an important part of the kiln it is not the most exciting bit. For some reason the channels appear very short. Perhaps the perspective of the images made it appear short or perhaps the software ignored this part of the image.
Take 2. This time I decided to digitally cut out the wall in the hope the software would focus on the channels. Cue epic software fail – there must not have been enough identifying points between the photos to distinguish the key features from each angle. In effect the software could not identify the same feature(s) from the different angles. I’m too embarrassed to show an image of this stage – it was so bad.
Take 3. I finally decided to ask for some advice from a friendly colleague (I realise now that perhaps I should have started here – Thanks to TS). If the software cannot distinguish the key points to link the images, try making separate chunks of the model from each of the three viewpoints and then join the three chunks together. Sounds sensible. This is the resulting image.
This attempt was more successful, although there were still problems. The software has identified the channels 🙂 but they are not aligned correctly 🙁 At this point, I’d run out of time as I had an appointment for which I needed to leave.
I will have another go at realigning or reprocessing these images when I have more free time and, hopefully, have found more images. None of the images above are anywhere near a finished model – I just thought I would share my experience.
This has taught me some valuable lessons about processing the images
a) ask for advice from the experts (obvious I know).
b) you need images from a variety of angles to make a decent 3D image, not just the angles that give the most pleasing photo composition.
The most important lesson was how important it will be to have the many and varied images of sites and buildings that are being sent in by the Curious Travellers community. We cannot do this project without your help and we greatly appreciate any images you are able to send us. Don’t worry if you think we may already have photos of the site or if it is an unusual angle of the building. It may just be that your image is the exact one we need to join the images together and make the model work: get involved