Graham Davis is Professor of 3D X-ray Imaging at Queen Mary’s Institute of Dentistry. In this blog post, he describes his team’s work in helping the BBC to restore a lost episode of the Morecambe and Wise Show from a disintegrated film reel, as featured in the latest episode of BBC Click.
For those who haven’t heard or read the story, it’s probably a little difficult to see the connection between dental research, X-ray science, and the Morecambe and Wise show.
Since the early 90s, I have been working on the design and application of X-ray microtomography (XMT) scanners, which are essentially microscopic forms of medical CT scanners.
XMT differs from its full-size medical counterpart in that it is only used to scan small inanimate specimens (from millimetres to a few centimetres), it takes longer, and whereas with medical CT the X-ray system rotates about the patient, with XMT it is the specimen that rotates.
In dentistry, we use XMT in research, where we can scan extracted human teeth and precisely map the mineral concentration in three dimensions. This enables us to study the way that teeth lose mineral under acid attack, and where possible, test experimental methods of reversing the process.
Although most of the focus of our facility is on tooth and bone samples, we are often approached by researchers in quite diverse fields with enquiries about the feasibility of scanning some other type of specimen.
We recently collaborated with Cardiff University to successfully scan and digitally ‘unroll’ a damaged 15th Century scroll from the Norfolk records office, using our scanners to detect the iron-based ink on the parchment.
My colleague, David Mills, and I had, on occasion, discussed the possibility of detecting and imaging the silver content of black and white films, but it was not until TV archiving expert Charles Norton came to us with another of those bizarre requests that we actually put this to the test.
Charles relayed to me a problem with acetate film, commonly called ‘vinegar syndrome’, whereby the acetate base of the film breaks down to form acetic acid and the image, at least for a while, turns to the consistency of marmite. Any attempt to unreel the film, or any movement that causes the layers to move with respect to each other, smears the image beyond recovery.
In time, the whole film turns to a black goo. Furthermore, the process is self-catalysing, so once it has started the end is inevitable. Not only that, but acetic acid fumes released can ‘infect’ other films stored nearby, thus affected films are usually disposed of quickly.
Charles then told me how a 16mm film copy of the second episode of the Morecambe and Wise show had turned up in an archive in Nigeria. In hot, humid conditions it was suffering from advanced vinegar syndrome and fit only for the skip. But this was the only remaining copy since the episodes were originally stored on video tape, long since erased, and such film copies were only made for overseas use. Charles had heard of our work with scrolls and asked if anything could be done.
Not wishing to turn down a challenge, we set to work. Early experiments with very short rolls of film (only a few centimetres in diameter) yielded surprising and spectacular results. But there was a fundamental snag; to go from these tiny sample rolls to something 30-40 cm in diameter would require a phenomenal increase in X-ray exposure (probably enough to incinerate the film) and an X-ray detector with inconceivably high resolution and dynamic range.
Even to get close to having something like this would take millions in funding and years to develop, by which time the film would be soup. Even then, we could never come close to having something that would scan the whole film; the laws of physics just would not allow it.
So after back-of-the-envelope feasibility calculations I broke the bad news to Charles and told him that there was no way it could be done, unless we chopped the reel into small pieces and scanned each piece individually. To my surprise, Charles immediately agreed to this proposal (even though I had not been terribly serious) since the alternative was to dump the film or watch it decay into a black sticky mess.
We used a laser cutter to avoid mechanical disturbance and piece by piece scanned the first eight minutes worth of film. We then cut more blocks towards the centre of the reel to give us snapshots of the remaining footage.
Over 5,000 X-ray images were taken as 45 blocks were each rotated around 360 degrees, taking around 18 hours per block to obtain the required image quality.
Each block gave us a 3D data set, which we could virtually slice through to see layers of film (like cutting through a swiss roll lengthways). With a few days programming, I was able to start manually flattening out the images to get my first glimpses of Eric and Ernie in action.
I even managed to semi-automate the process, before conceding that this was simply taking too much time away from my primary research. So when Adam Wiewiorka from BBC R&D came up with an idea for fully automating the process, I handed over the reins and let him work his magic.
The images recovered were distorted and damaged, mostly by the vinegar syndrome, but a few by the laser-cutting. Nevertheless, we had jointly succeeded in bringing these images back from the dead, which gave us all something to smile about.
For media information, contact: