And meanwhile, Milan Kundera was busy writing his novel. And then, I think a couple of years later, he was ejected out of Czechoslovakia. The earth goes around the sun 20 times, and Jean Claude Carrière and Phil Kaufman write a screenplay based on the novel, and Saul Zaentz produces it. And now, in Lyon, in 1986, so not quite 20 years, we are shooting this invasion again, using Soviet tanks that somehow materialised. And 200 extras dressed as if they were Czechoslovakian in the central area, the old town of Lyon, which looks remarkably like Prague, if you're careful about your angles. And the idea was to interleave this new footage with Daniel Day-Lewis, and Juliette Binoche, and Lena Olin with archival material. And it was all carefully storyboarded, so that we knew the footage that we had.

And we knew the archival footage. And we knew, well, if we have that angle, then we can shoot an angle like this, which is seemingly the reverse of that angle. And now we can have our actors in it. So we had tried, as an idea, to think about, let's call it 'the Zelig approach', which is: digitally insert your actors into an existing. But this was 1986, very, very early in digital technology. And you could do it. But it was very time consuming, expensive, and cumbersome. So we won't do that. We'll just rely on intercutting to create this illusion. And that meant a scouring of the world for all of this footage.

So I followed on the heels of Peter Kaufman, Phil's son, who had done a lot of the research. I then went to Stockholm, and London, and Amsterdam, and some other cities, asking them, 'Do you have any footage from…' 'Yes, we do.' 'Can I look at it?' 'Oh, this is good. Can you give me prints of this material?' So we would get dupes of this material. And then, figure out a way to technically give them the resources that existed at that time, 30 years ago, to degrade our new image shot by Sven Nykvist, beautiful 35 mm film, with the old images. And we did it.

And it looks convincing to me, even today. I look at the film, and yes, I believe that. But it was, by our standards today, very complicated pieces of doing certain things in the lab that increased the contrast, and popped the density, or increased the grain, or decreased the grain. Or diffused the colour, and reduced it from 35 to 16 mm, and then, okay, I think that works.