MARTIJN PANTLIN

Every photo print is already dying. The moment it’s made it starts to fade, and if you leave an old print in a drawer long enough you’ll find it years later looking nothing like the day it was taken. I’ve always loved that. A photograph is supposed to freeze a moment, but time keeps working on it anyway. The Decay Machine is my attempt to speed that process up and watch it happen in front of you digitally.

What actually happens to an old photo

A colour photo holds its image in three dye layers: one for the cyans, one for the magentas, one for the yellows. These dyes are not equally stubborn. Some give up faster than others, so a print doesn’t just get lighter, it changes colour. That’s why old photos drift toward those warm pink and magenta tones, or turn a strange sickly amber. The picture is still there, but the balance has shifted.

Then there’s everything the world does to the paper itself. Damp gets in and spreads in slow rings, leaving those brown tide lines you see on water-damaged prints. Where the moisture sits, mould starts to bloom in little dark colonies. Tiny rust-coloured spots appear, what collectors call foxing. The surface can lift and crack. Salts rise to the edge of a stain and dry into a pale crust. None of this is random, it has a shape and an order to it, and that’s the part I found fascinating, but was truly hard to create.

Trying to simulate it

I didn’t want a filter that just makes a photo look “old”. I wanted the decay to actually grow, the way real damage does, starting small and creeping outward over time. I took inspiration from the beautifully decayed photos in each section of this herri issue. So I built it up out of several different effects that all run together:

• Fading — the colours drift and wash out over time, each dye at its own speed, so the picture slowly turns toward that faded pink-and-amber look.
• Edge diffusion — colours bleed into each other where they meet, like wet dye running across the paper.
• Stains — damp patches that spread from little starting points, each one growing its own interior, a darker rim, and a ragged edge.
• The crumbler — the crusty business at the edge of a stain: mineral flecks, the pale line where salts have dried, the dark tide mark.
• Growth — mould, blooming in grainy dark colonies in the damaged areas and spreading further the longer you let it run.
• Foxing and speckles — the little rust spots and dark specks that freckle an ageing print.
• Emulsion lift — the surface swelling and peeling where it’s been wettest.

Each of these has its own timing, so early on you mostly see the colour shifting, and only later do the stains and mould really take over. I spent a long time on that order, because if everything happens at once it just looks like noise. Decay should feel like it’s happening to the photo, not painted on top of it. Even in the age of AI programming, this was not a small feat to direct.

There are sliders to play with all of it: how fast the colours die, how many stains there are and how big they get, how wet the paper is, how much it stains, how far the mould spreads. You can push it gently toward a tastefully aged print, or all the way to something almost eaten alive.

Upload your own photo and save it

The best part is that you can feed it your own picture, from a file or your camera. Upload a photo and the machine looks at it first, works out where it would most likely rot, and then ages it from there. A bright sky decays differently to a dark, dense corner, so no two photos fall apart the same way. You can scrub back and forth through time, watch your holiday snap slowly turn into something that looks like it’s been sitting in a damp box for fifty years, and then reset it and try again with different settings. You can save your photo in the decayed state the timeline is currently set to.

It’s oddly moving to watch a photo of people you know slide into ruin and then pull it back to new again. That tension is really what the whole thing is about.

The controls

There are two ways to play with it. The featured picture on each Section page runs in its own mode, tuned to an already heavily decayed photo. When you upload your own photo you get a slightly different set of tools built around the way real prints rot. Here’s what everything does.

In the default mode, the machine is really about colour dying. The three channel sliders (R, G, B) each control how fast one dye layer disappears:

• R — warming. As red rises, the cool tones turn pink and salmon.
• G — contrast. Green shifts push the image toward rose or olive.
• B — coolness. As blue fades, the image loses its grey-blue cast and turns amber.

Then a few more:

• spatial — how unevenly the decay spreads across the image, from a smooth overall fade to something blotchy and patchy.
• beyond 2026 — pushes the fade past the natural end point of the Section photos, drifting the whole thing toward a warm, parchment-like state.
• edge diffusion — the borders between different colours dissolve and bleed into one another.
• form crumbler — all the crusty stuff at the edge of a stain: rim flecks, mould creeping along the drying line, and the little shadows where the surface lifts.
• growth — the mould and mineral deposits that bloom at the stain rims once the crumbler is doing its thing.

There’s also a whole mould panel (press M) if you really want to get into it: patch size, opacity, how much it blends with the photo underneath, when it starts, how far it spreads, and colour pickers for each layer of a colony from the pale halo down to the dark core. You can even export a look you like as a preset and load it back later.

In upload mode, most of the same controls are there — channels, edge diffusion, form crumbler, growth — but described more simply, and there’s an extra set of sliders under the sliders icon for shaping how your photo falls apart:

• decay speed — how fast the damage spreads across the print.
• seed density — how many stains appear, from a few to many. Stains seed randomly when loading and upon Reset or Shuffle.
• seed size — how big each stain grows.
• moisture — the size of the wet halo; more water means more bleed and flow.
• staining — the overall yellow-brown chemical staining across the whole print.
• mold diffusion — how far the mould creeps out from the stain cores.
• emulsion lift — the peeling and swelling of the surface inside the wettest areas.

Two buttons worth knowing: reset puts everything back to default and reshuffles where the stains land, while shuffle just randomizes the stains to new spots and keeps your settings.

Shortcuts

A few keys make it easier to play with:

• Space — play or pause.
• ← / → — step through the timeline (this pauses playback so you can inspect a moment).
• Esc — hide or show the controls, so you can just watch.
• D (upload mode) — a debug view that shows where the stains are seeded and how the surface is lifting, handy if you’re curious about the machinery underneath.

Technical

For the curious, a few words on how it’s built. The whole thing runs realtime in your browser, on your graphics card, using something called WebGL — the same technology games and 3D websites use to draw fast. Nothing is sent to a server, it all happens live on your own machine, which is why you can drag the sliders and see the photo change straight away.

Underneath, the decay is a little simulation running many times a second. Instead of laying a fixed “old photo” texture over your image, the machine treats the picture like wet paper: it works out where damage would start, then lets moisture, stains and mould spread outward step by step, each frame building on the last. Those calculations run as small programs called shaders, thousands of them at once, one for every pixel on the screen. That’s what lets the damage grow organically rather than just fade in, and why every photo — and every reset — falls apart a little differently.

Why bother

We keep everything now. Nothing we photograph ever gets the chance to age, because it lives forever as a perfect file. The Decay Machine is a small way of giving photos their mortality back, of remembering that images used to be objects that lived in the world and took its damage. There’s a kind of beauty in the falling apart that we don’t get to see any more.

Go and break one of your own photos on top of this page and see what happens!

Header photo by Markus Spiske on Unsplash