The titles are actually very similar to the original 1970's titles which featured a similar, if reversed, pan across a real valley. We started out with a very detailed brief from the BBC with the elements they wanted in the scene. The valley needed to reflect the makeup of a generic Welsh town in some unspecified location: urban, industrial, rural, historical all at once. They also wanted to move from morning to midday throughout the 20 second piece so we had to plan for a lighting change throughout. The foreground detail was part of the brief from the start and the decision to go completely digital for the piece made this a lot easier. The budget was very low in comparison to perhaps what the BBC would pay for a new Eastenders title sequence and to do the camera move and lighting change completely in live action would have taken weeks of prep, location scouting and filming with a motion control camera halfway up a mountain to get it right.

With help from the production team in BBC Publicity, we found a suitable valley about an hour north of Cardiff. To be honest, I've no idea where it is because I didn't take much notice of the map when I was being driven there but the valley is actually a military training ground and was therefore mercifully clear of pylons and modern constructions. Apart from a few sheep, it as a blank canvas and so I took two panoramic photos of the scene, one in the morning and one in the afternoon to get the two lighting states. To make things even more interesting/difficult for myself, I took the photos at high dynamic range and then stitched these HDR images together later. This meant that I had a gigantic HDR panorama to work with and could extract any lighting issues and recolour the landscape without damaging the underlying image. The eventual Photoshop files were over 1.5GB in size. I removed all the sheep and any visible construction artifacts that would betray this as a still image. I also lowered one of the hills to make it a better frame. Due to the images being taken at different times, they didn't completely match so I had to do a little rewarping so that it didn't look so odd when one dissolved to the other during the course of the sequence.

The town of Cwmderi is actually an HDR picture of Cwmbran mirrored horizontally. I removed a few landmarks and carefully repositioned it so that you didn't see very much of the town but the majority of it was taken on the same day as the landscape photos from a hillside above Cwmbran. The rest of the town, particularly the industrial part, is all constucted from photos I found on Google Image Search, by far the best way of finding matte painting elements quickly and at high resolution. Like the landscape, I had to recolour them all for two different lighting states, although these images were a bit easier to line up! The pylons were CG, as were the rugby posts and the wind turbines. The flock of crows were also CG and were a very quick basic render. The sky was the last thing to be made and that was done in Vue, a very awkward program that I hope never to have to use again. Although I thankfully wasn't the one operating it, it was such a pain to get right and keep consistent that I'd never consider employing it again. Why it should be a struggle to animate and render something as simple as a sky, I'll never know. A low res CG version of the landscape was made to create the shadows of the clouds moving over the ground.

The foreground detail was entirely constructed with CG and a lot of it was taken from reference photos I took on the same day as the photoshoot when we were returning along the A470 to Cardiff. We found a field with a drystone wall so I quickly nipped in to take some snaps before we lost the light (or were seen by the farmer!) Romano used the images as projection maps on top of basic geometry and it worked really well. The tree was built from scratch and the ivy was 'grown' using a really great little program that just created ivy in 3D. It's the only program that didn't cause us any problems during the post production, although the render times for the ivy was considerable.

Rendering was a big headache on this production as we didn't have the render farm yet and had to rely on the office workstations to get everything done. Not only that, this was our first proper HD CG production and that caused lots of issues in of itself. The amount of layers and passes it required created a huge volume of data and you could feel the network slowing down every time we started rendering a composite. Now, it would be done a lot quicker and without fuss. In some ways, it proved the need for a farm and a higher quality storage system, without which we wouldn't be able to do similar work again.

My biggest responsibility of post production was on the compositing side. I decided very early on that I was going to use the project to try out Fusion as After Effects wouldn't have been able to cope with the large image files required to make it work. After Effects is good as a quick and efficient compositing system but to do anything HD or more heavy duty than simple broadcast effects it just can't deal with the memory management. Fusion, on the other hand, imported and rendered the gigantic images fine and its 3D system was far superior and allowed me to create a curved surface to put the landscape on and mimic more accurately the dimensions of the scene. The camera was prevised in 3DSMax before being signed off (at least initially!) and then used to create both the 3D foreground and then imported into Fusion to match and create the matte painting landscape. This sychronisation meant that everything worked together without us really having to think about it. If there was a change in the camera movement (and there was almost right at the end!) then we could easily update both the CG and the comp and rerender. Everything was composited in 3D in Fusion for this reason and it proved to be a very robust system. I was able to colour correct and rewarp a lot of elements very quickly and easily using the node based workflow, something I would never have been able to do in After Effects. The only downside was that Fusion isn't as flexible as After Effects at dealing with timelines or animation. It's still very clunky while After Effects is a lot more slick and easier to manage (although the later CS3 and CS4 versions of AE have taken a few steps back). I'd say that Fusion and Nuke are really good at doing individual complex shots, while After Effects can deal with a large number of simple shots more efficiently.

Due to the high profile nature of the work, the pressure to complete everything on time was considerable. I had to sit in various meetings with BBC staff, sometimes seven or more, and go through every details many times over. In the end we delivered a week later than planned but it made for a better sequence so I've got no regrets about that. If this had been a live action shoot, I don't think there would have been much scrutiny and the end result would have been what we got on the day but the digital nature of the production and that fact that most of the frame was being constructed and animated in CG meant that all details were up for discussion. This can be a double-edged sword and something that I have learned since is to set very definite boundaries for approvals and to say no when something can't be done in the time.

It was strange that we had to deliver an HD master for a show that was still being shot in SD (and probably still is two years later!) but it got done in time and it was great to go to the 'premiere' at the BBC clubhouse and have people asking us where we filmed it. People still assume it's a live action scene with some extra CG bits here and there and it's a testament to the work we did that it continues to surprise people.