This is a practical post looking at the results of two 3d-modeling sessions. The objects I’m working on were excavated in the 1960s as part of the American Excavations at Kenchreai, which Prof. Joseph Rife directs. I had written a little bit more intro in my last Kenchreai post so please see that if you’re interested.


The first model is of object KE 1224 (S 57), a marble paw of a lion or other large cat. Click through to see the entry for the object, along with an embedded model, though I will also link to that directly from this post.

The method I’m using to make the models is photo-based. As in, I take a bunch of pictures and use Agisoft Photoscan to calculate a mesh describing the 3d-volume of the object and then to build a more-or-less photo-realistic texture to “drape” over that mesh. Bill Caraher has recently discussed an article by B. Olson et al. that describes the process.

Here’s a picture of the object on a photo stand:


You can see that I’ve used a styrofoam block to help the paw stand upright, and that there is an offset styrofoam block to help Photoscan figure out the physical relationships I’m capturing in the set of photos I took of this piece. I shot 78 images with my iPhone and used 31 to make the model. Why the iPhone? It’s small, so works well as a hand-held, and has a sensitive shutter “button” that I can trigger without shaking too much. I could perhaps get better pictures with a high-end point-and-shoot but the iPhone works for now.

Here’s the result as shown in a Photoscan screen-capture, with thumbnails of a few of the pictures I took and an indication of which of those I used. No red “no entry” icon means Photoscan incorporated that view.


Cooler than that image, though, is the ability to share the model itself so you, the reader, can rotate it at will. Click on this image to go to the site, with the paw shown with no automatic lighting.


Although I said this is a “practical” post, I encourage you to zoom in on the chisel marks by holding down the ‘x’ key. And note the blurred division between intentional marks that indicate fur and the underlying chiseling that also contributes texture. Ancient art was partly a form of craft production and this fragment, when shown in 3D, encourages consideration of that aspect of the piece.


The paw has a broken edge so it wasn’t a great loss to use that to balance it on the stand. KE 235 (L 61) is an essentially complete Late Roman lamp so here the challenge is to get all the way round the object. That being the case, let me say now that I didn’t quite succeed.

I used the same basic process, but this time I suspended the piece using fishing line. Here’s a picture:


Of the 88 pictures I took, I selected 47 to use in making the model. The ones I didn’t use were either very close duplicates or out of focus.

I had to move carefully so as not to make the whole thing shake. But again, I couldn’t really get good pictures from all angles. I was in a narrow space, which didn’t help. But the real problem was that when I got under the lamp to shoot from below the handle end, I was pointing at ceiling lights and getting terrible glare.

Results, as seen in the linked screen capture from, are not, however, unusable.


Click through to see for yourself.

When oriented handle at the left and looking at the side, the transition from the shoulder to wall didn’t come out well. And the base under the handle didn’t work well at all. But again, it gives a sense of the shape and surface of the object so I’m willing to share it as part of a discussion about technique. You can learn from my less than perfect attempt. And I do consider this model something of a “floor” in terms of results. Better lighting, particularly the ability to dampen overheads, more space, and more experience should lead to much better outcomes. Stay tuned for those. Though we’ll all have to wait until next year as I’m back stateside now.