The problem of choosing the right colors for stained-glass compositions 1 bis/2

by adrien

(above) two sheets of Antique mouth-blown glass presenting strong color irregularities versus (below) the original measured samples of the same references at the Glasshütte Lamberts factory last summer. The chemical composition of the material is the same, but the thickness greatly varies. How to deal with such differences?


Stained-glass is a very special material to work with.  One one hand, the colors of some glass have absolutely unique properties: the light that passes through it is so well filtered that it produces a color that is more pure than any other material (with the exception, maybe, of some dyed gelatin filters – however gelatin filters are very fragile, while glass is almost unalterable). On the other hand, Antique stained-glass is produced by manual workers (it is mouth-blown) in coordination with chemists that “tune” the glass chemical composition in massive ovens. The result is a very beautiful material that present a lot of natural variations of thickness and of color.

Some stained-glass are colored in their mass, while some others are “plated”: a sheet can be made of 2 or 3 layers of different colors that are superimposed. The upper plate is usually very thin (up to +- 1/10th millimeter or less), and very “concentrated”, i.e., if the plate is thick the color can easily turn almost black, while it would be Pink at 1/10th mm.

Depending on the spectral characteristics of the chemical components involved, the thickness can matter or not. Strangely enough, some orange selenium-glass sheets don’t change colors if they are 3 or 5 mm thick, but become strongly lighter if they are less than 2 mm thick, for example. This is a consequence of the spectral transmission curve of this particular colored glass. When I was visiting the stained-glass studio Parot in Paris a few months ago. I saw a similar phenomenon: a 10 cm thick piece of fused glass (silver-yellow I think), made of a bunch of yellow glass plates, had exactly the same color as one of these plates! (The explanation is easy but would take a long time to write).


So… with such a difference between the data and the reality, what do we do?

There are several possibilities: accepting the natural variation, excluding the parts of the glass sheets that are too different (if we have enough stock!), or correct the data and redisign the pictures progressively, pannel by pannel, when it is possible to introduce more variety.

The fact that the glass “natural” color-variations introduce another level of randomness in the pictures is also interesting.

So far, the comparison between the computer model and the stained-glass pannel seems fair enough. However, the real test will be made within a few days, when the pannel will be ready to be lighted from behind! (right now the colors are not appearing correctly and it is impossible to tell how it’s gonna look like).