You could hear the argument from the next room. On one side, our registrar, who was fairly sure that animating a detail of a Monet for a temporary study label was "the end of museum work as we know it." On the other side, the education director, who was equally sure that a twenty-second loop beside the frame was "the difference between a look and a glance." I was on the registrar's side. I want that on the record, because it makes what happened next more embarrassing.

The Immersive Backlash
My snobbery had history behind it. The last few years have been very good to the immersive exhibition � the big-budget shows in disused warehouses where you stand inside a projection of a painter's oeuvre while the music swells. Some of these are genuinely popular. Most of them, in my opinion, are a projection screen wearing a famous name. The colours have been made louder, the edges softened, the artist flattened into a screensaver, and a room full of people leaves without once having looked at a painting. I've said versions of this at conferences, at dinners, at one point on a panel, and I was never short of agreement.

So when our small museum floated the idea of "animated studies" � short moving details from our own collection, placed beside the originals � I was ready. I had a position, a quotation from a critic I'd read once, and a very confident tone of voice. What I did not have, as it turned out, was a school group.

I should also admit that I went to one of the big immersive shows, for research, with a notebook. I stood inside the projection, I watched the room, and I spent most of my time in the gift shop, which is where I suspect these shows actually live. The catalogue was a scarf. The scarf was thirty pounds. I left with the scarf and the opinion I'd arrived with, which is to say I learned nothing except that my position was very comfortable.

The Small Experiment
Let me be precise about what we did, because the scale matters. We did not turn the gallery into a cinema. We took one water lily painting � a small Monet from the permanent collection, the one with the pale light � and we made one detail move: the water between the lily pads, the top third of the canvas, for twenty seconds, on a screen no bigger than a laptop, on a stand beside the frame. Under it, in small type, we said exactly what it was: an animated study based on the original, made with an AI video tool � the Wan3.0 Video � which runs free in a browser, built on Alibaba's Wan model, the latest version of which opened to the public in August. The label was the point. Nothing was hidden, nothing was passed off as the work, and the painting never moved. The screen did.

What the School Group Did
The school group arrived on a Tuesday, forty of them, in the usual state of being elsewhere. They moved through the first rooms the way water moves downhill. And then they reached the Monet room, and the kid in the front � nine years old, I remember the orange coat � stopped. He looked at the screen, and then, this is the part I keep telling people, he looked at the painting. Not at the screen instead of the painting. At the painting because of the screen. He pointed at the top third of the canvas and said something to his friend that I couldn't hear, and his friend looked, and then the whole group was looking at the water in a painting that, up until that morning, they had been scheduled to pass.

I stood there with my confident tone of voice and I watched a nine-year-old find the water. That was the whole experiment. The water moved on the screen for twenty seconds; the looking lasted about four minutes. I have not stopped thinking about the ratio.

Two weeks later the class teacher wrote to us. The kids had drawn water lilies all week, she said, and one of them had asked whether the real painting was "the loud one or the quiet one." We wrote back: the quiet one. That's the correct answer.

The Comment Book
The museum keeps a comment book at the front desk, and the week the animated study went up, the comment book did something I have never seen it do: it became an argument. "The moving water made me look at the painting for the first time in thirty years of visits" � that one was from a retired art teacher, we found out later. "This is gimmickry" � that one had no signature. "Is the painting damaged?" � that one we answered in person, twice. Our registrar, who remains unconvinced, filed all of them. I think the filing was her way of keeping score, and I don't blame her. The book disagreed with itself for six weeks, and honestly, that's the closest our little museum has come to a public debate in years.

The Honest Part
I still think the criticism is real, and I'd rather say that plainly than pretend I came around to everything. A screen can flatten a painter into wallpaper. A museum that animates everything has stopped trusting its own collection, which is a kind of failure. And the big immersive shows � the ones that turned a hundred paintings into one projection � still bother me, and I think they should bother everyone. The tool is not the problem. The shortcut is.

The tool also has its limits, and I'm not going to romanticise them. The first render gave us water that rippled like open sea, which my colleague the printmaker said she'd have paid to keep, and we re-did it. The clouds in another version drifted sideways, which in a different painting would have been a disaster and in this one was just wrong. We re-did that too. We spent a week getting one detail to look like itself, which is exactly how long you'd expect a museum task to take.

And one more thing, because I keep being asked: no, we didn't animate the whole collection, and no, the water didn't stay moving. The label said the study was a study. When the six weeks ended, the screen came down, and the painting did what it's always done � sat there in the light and waited for someone to look. That's the test, I think. The screen is gone now, and the painting is still getting looked at.

The Principle
Here's where I ended up, and it's not the position I walked in with. Motion doesn't replace looking. It can't, and pretending it can is how you get a room full of people watching a screensaver. But motion can earn a first look � a first look at a painting that most visitors would otherwise walk past, in a museum that is in the business of making looking happen. The screen is not the artwork. It's the door. And a door, used honestly and labelled clearly, is not a betrayal of the collection. It's a service to it.

We kept the animated study for six weeks. The registrar still doesn't love it. The education director is planning the next one � a contemporary work, this time, and the artist has strong opinions about it, which I love, because the argument should keep happening. That's what a museum is for. And I, the confident voice on the panel, have a new line for conferences: the painting didn't move. The people did. That's the only animation that ever mattered.