Do you get given the scans or do you have to do it yourself?

I work closely with a multimedia firm on each re-creation - they supervise high-resolution digital camera rostrum photos of each page.

Are you allowed near the books for obvservation (if not)?

Yes, I get to look at the books while a curator turns the pages. We film each book from different angles, and then spend ages peering at the curves each page adopts and the way the spine of the book deforms over time.

How did you get the job?

I think it must have been word of mouth, but I was running some magazine adverts back then - can't actually recall.

Is this a "British Library" work, or an Imaginetix one? (I mean have you been employed by them, or are you doing the work on a contract basis?)

I'm on contract, to a multimedia firm called Armadillo Systems who create the whole installation - from the initial photography, through programming the application that it all runs on, and all the really clever stuff. Effectively, I do this on a freelance basis, but over the years the working relationship has become a close one.

If you've been working on the project for four years, there must have been changes to LightWave in that time you could have done with at the start of the project. Is there anything you wish you'd had in earlier versions?

I think I must've been using LW5.6 or something way back then, so Modeler's drawmode was a progressive one, rather than fast OpenGL. Everything was done with MTSE morphs rather than endomorphs, so it would take ages and ages to set anything up, and tweaking page-shapes was a howling nightmare compared to today. I recall we didn't have enough RAM or CPU grunt to do everything in a single pass, so I was ramping Object Dissolve on pages two or three layers back from the ones we were currently rendering. Background curve conforming wasn't even a dream then, or having the Spline Guide so I set up virtually everything with the Magnet and Drag Vertex tools. Oh, and one thing we had endless problems with was the clip-maps that make each page irregular - realtime OpenGL transparency sped this particular job up noticeably. Back in 2000 or 2001, it would have been some kind of miracle. As it was, I melted the RAM clean off my old GeForce 1 card.

There wasn't any Layout Spline Guide then either, so I had the spine rigged with a skeleton. Everything is so much easier and quicker these days.

Tell me more about the research you went through for the different paper types?

Given that each book is wildly different, each one's a little R'n'D project in its own right. Peering intently at the videos of the page deformations, looking at all the different rostrum shots and some off-angle pics with a positioned light-source, and looking very, very closely at the books themselves in the Library. The light-reaction properties of 600-year-old Vellum (which is basically thinly-scraped goat hide) are totally different to those of, say, 200-year old rag paper. Also with the Sherbourne Missal, I had to keep staring at the gilding on the page to convince myself how reflective it was - it wasn't like Humbrol Gold Enamel, this stuff is seriously shiny - mirror bright, almost.

Do you just get scans for colour? If so, what do you do for bump, diffuse and so on?

If there was a camera which could just photograph the diffuse or reflection channel of it's subject, they'd sell in the millions! The channel maps are created in photoshop. It's a team effort, with the photoshop guru separating out gilding from page, and the various pigments from each other. I then muck about with the various levels and wearing the print off my F9 key until the responses from the page match our observations and shot footage. It's an iterative process, but experience with the task has made it go pretty quickly now.

Do you have to do lots of correction for perspective and the like? I imagine that the pages aren't completely flat when photographing them?

The Sherborne Missal is over six inches thick, so the sheer parallax distortion was very noticeable, let alone the nose-dive the page edges took into the binding. I came up with a system of pixel displacement mesh-warps which re-flatten the page digitally, giving us a high-resolution template that would be faithful to the page if it was physically disconnected from the binding. This is the start-point that we then work from to create all the channel maps. I'm really careful with using these, as preserving the authenticity of the source material is a primary concern. I'd start off on this with very high resolution images, so we're always working down towards out final target resolution, rather than scaling anything up and losing information.