Illuminating Objects: Spanish lustre dish

3D model 

 

The 3D model loads in layers and may take a few seconds to complete, so please be patient. If the model does not load at all for you, please try using Firefox.



Model created by John Hindmarch and Ali Hosseininaveh Ahmadabadian. It would help John's research if you could spare five minutes to complete a short survey.

  • Hold the left mouse button to rotate and spin the bowl.
  • Use the left mouse button and [ctrl] to zoom in and out -
  • though we recommend not zooming in too far!
  • Double click the image at any time to reset the display.


John Hindmarch on making the 3D model

First of all, I'd like to thank the Courtauld for giving me the opportunity to create this model of one of their beautiful objects. Secondly, I should point out that this model is a work-in-progress; what you see here is a first attempt, and we hope to improve the quality of the model over the coming weeks, using different and more sophisticated techniques - some developed specifically for this object.

While we did use a laser scanner to capture the bowl's detailed shape, this particular model was created using photogrammetry - the process of constructing a 3D model from a series of 2D images. Using about 60 photographs, our software first calculates the camera's position in relation to the object,then identifies 'features' in the images. Once it has found the same point in two or more photos, it can calculate that point's position in three dimensional space. By building up 'patches' of points, the programme creates a full 3D 'point cloud' model of the entire object.

Problems

'Lustre' is “the state or quality of shining by reflecting light”, and as you can see from photos, this piece of lustreware is highly reflective - a real problem for both photogrammetry and laser scanning. All those 'specular highlights' depend on how you look at the object – ie, they move as you move - which means that different highlights appear in different places in different photos. This confuses the photogrammetry algorithms: they have no way of knowing that the bright spot in the image is 'just a reflection' and not a real feature of the object. You end up with a model with lots of 'noise' and either corrupted data or big holes where the reflections were strongest. Hopefully our next iteration will fix some of these problems...

Why does it look pink?

The pink highlights in the model above are an approximation. The bowl itself does have amazing pink reflections, due to the presence of copper in the lustre (metals, unlike non-conducting materials, have coloured highlights). In reality, the pink highlights only appear on the darker areas of the bowl, we aim to simulate that in future iterations of the model.

Try it yourself...

Behind the scenes, photogrammetry is exceptionally complex, but the idea is simple, and you can try it out for yourself at sites like 123D Catch.