The Old Times

Nick Gellatly

Putting the Words In

Fourteen frames of painting, ninety seconds of timeline, and a title sequence with no title on it. This was the week the words went in, which turned out to be much less about choosing a typeface and much more about whether anyone could read the thing at all.

The ‘homework’ from last week was to create a styleboard:

And to create a visual poster at 16×9 landscape (1080) format that can be used as a ‘benchmark’ visual style for you project:

The objective for this week was to move to the ‘Production’ phase including each of the following:

• Asset creation
• Text
• Layouts
• Artwork ready

With this in mind I created the assets using Canva (with support from Claude AI as my virtual assistant and online instruction manual). This phase was completed and can be seen in the previous post for this module on this blog: https://theoldtimes.co.uk/2026/10/16/motion-graphics-week-4-part-3/

Credit where it’s due

I had been carrying a rough credit list in my head since the storyboard, and I was wrong about two things in it. The first is billing order. Because he is my favourite, I had Gordon Warnecke as the opening cast card, on the reasonable grounds that Omar is the lead. In the actual film Daniel Day-Lewis is billed first and Gordon seventh, which is how 1985 saw it, and since I am re-imagining the titles rather than rewriting the contracts, that is how my version has to run. Day-Lewis moved to frame four and Warnecke to five.

The second is the production card. I had it as Working Title alone. It is Working Title and SAF Productions, made for Channel 4, so the opening is now three cards rather than two: Channel Four Television, then a Working Title production, then in association with SAF Productions.

There are thirteen cards in total. I had drafted seventeen and cut four, because at ninety seconds anything beyond thirteen flashes past too quickly to read. Derrick Branche, Souad Faress and Richard Graham all came out. They would be on an end roller if this film had one.

In total, I went through seven versions in an attempt to get it right (in so far as I have) and the first iteration with titles can be seen here:

James had asked for ‘dynamic typography’ and I was pleased with the result – frankly, I wasn’t certain a few weeks ago that I’d get this far with this project.

Dynamic typography means the type carries meaning through how it moves, not only through what it says. A credit that fades up is static type that happens to be sitting on a timeline. A credit that assembles the way the pictures assemble is doing some work.

The temptation here is obvious, and it is to show off. Kinetic type is the easiest place in a title sequence to do something clever that nobody asked for, and my artwork is already calm and flat and deliberately unhurried. So I gave myself a narrow brief. The type moves the way the machine moves, and it never moves for its own sake.

Three behaviours, each borrowed from something the washing machine already does.

1. Tracking close. The letters start spread wide and gather as the card settles. This is the drum: things fly apart, then come back together. It also solves a practical problem, in that widely spaced type is hard to read in motion and tight type is hard to read against a busy picture, so moving between the two gives you a moment of each.
2. Per character offset. The letters do not all arrive at once. A range selector walks along the line so each one fades and lifts a few frames after the one before it. This is the tumble, because objects passing the porthole go by in sequence and never all at the same time. It is about eight frames across a whole line, so it reads as a ripple rather than a wave.
3. The settle. The ripple shortens as the sequence goes on. Eight frames on the first card, two by the time we reach the folded laundry and the door. By then nothing in the picture is moving either, and the type has stopped fidgeting along with it.

If, like me, you have watched the version above on YouTube, you will notice that not all of the titles were legible – while I liked the colours and general design the main title itself wasn’t sufficiently bold or, indeed, ‘beautiful’. The shots that worked best were the ones sitting on a navy background. Everywhere else I had put flat type straight onto a painted, faceted, mid tone picture with nothing between the two, and it only survived where the painting happened to be empty and dark. Moving the cards around would not have fixed it either, because frames five, six and eight have no empty quarter to move into.

For the next version, I moved to include a piece of ink navy, slightly tilted, with uneven edges, drawn to fit that card’s own text and sliding in six frames ahead of it. You won’t notice but I also changed the typeface, or at least I tried to. Poppins has perfectly circular bowls and even strokes, which reads soft and contemporary against artwork built entirely from straight edges. A condensed face would sit better and carry far more ink at the same height, which matters on a busy frame. I picked Oswald, spent a cheerful half hour failing to install it, and discovered via a diagnostic script that every title I had looked at for two days had quietly been Arial Bold. Having now seen it against the shards, Arial Bold is doing a perfectly respectable job, so Arial Bold is staying. Not every decision has to be the one you planned and i think this works well and I don’t think sits as an admition of defeat. Though maybe it is but I didn’t want to sweat the small stuff!

The neon sign

In an early version of this project, I made a neon logo – with ‘Beautiful’ in pink script, MY in cyan, LAUNDRETTE in yellow block capitals, all on black. It had been sitting in a folder ever since, but I knew it would make the difference here.

It is now the main title. Screened over the painted shopfront so the black drops out and only the glow lands, it stops being a caption laid over the picture and becomes the shop’s own sign, lighting up in the window I have just wiped clean. It switches on the way neon does, two stutters across six frames and then steady, with a faint hum on the opacity afterwards.

This is better than what I had planned, and I would not have got there by designing forwards. I got there by going back through my own discarded work and finding that the thing I had made for a version of the project I abandoned fitted the version I ended up with.

There is also now an end card, after the ninety seconds rather than inside them, with the neon logo glowing behind it. My name on my own work, which feels like the right place to put it: outside the film’s credits, not pretending to be part of them.

The bit where it all fell over

I have always been open about my use of AI and since I was introduced to it during the ‘Fundamentals’ module last year, it has been a revelation and I am noy ashamed of using it despite the anti-AI sentiment in the media (at least the MSM!) perhaps years in the frame during my professional life as the object of hatred of supermarket development has hardened me. I am not ashamed.

Everything in these fourteen frames is built by script. I am not keyframing a hundred and fifty layers by hand. I describe what I want, Claude writes an ExtendScript file, I run it in After Effects, I look at what comes out, and I say what is wrong. Each pass takes minutes rather than hours, and the whole project is reproducible: if I lose the file, I rerun the scripts and get it back.

Which I did have to do. After Effects crashed yesterday afternoon and took several hours of changes with it. Auto save was set to twenty minutes and five versions, which is not enough, and it is now ten and twenty. Recovery took one run of each script, because the work lives in the scripts rather than only in the project file. That is a real argument for working this way.

The more instructive failure was smaller. For three rounds the names kept vanishing off cards. Each time there was a plausible explanation, each time it got redesigned, and each time something else broke. The actual cause was one undeclared variable: the script used the composition width in a dozen places and never said what it was. Every symptom followed from that single missing line, and three rounds of redesigning the alignment had been three rounds of fixing something that was never broken.

What fixed it was not more thinking. It was a diagnostic that tried each step separately and printed what threw, and a mock version of After Effects that runs the scripts and catches that class of mistake in a second. Both should have existed three versions earlier. The lesson is not that AI makes mistakes, which is unremarkable, but that the way you catch them is the same as it has always been: stop theorising, get the error message, and build the thing that tells you the truth.

Nine iterations

In total – to date – I have downloaded and edited nine versions of the titles and the next video shows where I have got to:

Next step: sound

The pictures are done and the words are on them. What is missing is the thing that will decide whether any of it works, which is what you hear.

There are three jobs. The music bed is already cut to ninety seconds at 96 BPM, which is why every edit lands on a downbeat, but it was chosen for the animatic and needs judging again now the pictures are finished. The sound effects from Week 4 are scratch, dropped in roughly to mark the beats, and need re-timing properly to what is actually on screen: the coin, the drawer, the drum filling, the spin, the door. And then the balance between the two, which is where a title sequence is usually won or lost. I need to test these and establish whether these are the sounds I want, whether it feels – to me at least – that the music works. This is my test for the ‘reading week’ break.

The interesting question is what the sound is for. The brief says the piece should inform, educate and entertain about a time and a place. The pictures are doing the place. Ninety seconds of a laundrette in 1985 South London, is a lot of information to carry in sound alone, and the obvious temptation is period pop. I am not sure that is right. A launderette has its own music, which is a drum turning and water moving and a coin dropping into a slot, and I suspect the piece is better served by leaning into that than by putting a familiar record over the top of it.

Credits and tools

My Beautiful Laundrette (1985), directed by Stephen Frears, written by Hanif Kureishi, produced by Sarah Radclyffe and Tim Bevan for Working Title and SAF Productions, made for Channel 4. The credits in my sequence are the film’s own. The pictures are not. This is a student re-imagining made for Motion Graphics MOD009754 at Anglia Ruskin University, and nothing in it is endorsed by or connected to the original production.

Made with. Adobe After Effects 2026, with every frame assembled by ExtendScript rather than by hand. Still artwork generated in Canva and art directed against a style board of my own. Claude for the scripting, the credit research, the diagnostics and a good deal of arguing about where the title should go.

Typeface. Arial Bold, by accident and then on purpose. I didn’t like Poppins, Oswald was the second plan.

The prompt, roughly. Every piece of artwork in the sequence was generated against one unchanging house style: a bold modern gouache and acrylic painting, faceted and cubist influenced, flat painterly illustration with visible directional brush strokes and matte paint texture, every surface shattered into angular geometric facets, hard edged deep indigo outlines, flattened and tilted perspective. Explicitly not photographic, no film grain, no 3D render, no gritty realism. A fixed palette of nine colours, sampled from the benchmark frame rather than invented: ink navy, laundry cream, powders orange, vermilion, drum turquoise, deep teal, ochre sand, brick brown and pale sage. Each element was generated in isolation on a plain grey ground so it could be cut out and animated separately.

Previous weeks: Week 3, the storyboard and Week 4, the animatic.