Psychological Temperature
In dialogue with Poppy De Havilland
FVTVRIST Magazine // Interview by Anna S.
21 May 2026
The figures that move through Poppy De Havilland's paintings are suspended between the familiar and the strange. The same young woman appears again and again, not quite a character, not quite the artist, not quite a symbol, rarely meeting the viewer's gaze, holding something inwardly that stays just out of reach. Gravity feels uncertain. Bodies balance in improbable ways. Objects hover. Nothing is quite grounded.
Sylvia Plath arrives in the titles after the paintings are finished. So does Hammershøi. So do Piero della Francesca and Fra Angelico — painters De Havilland came to partly through her studio proximity to Von Wolfe, who introduced her to early Italian painting and to the open-source side of AI image-making. The references don't explain the images. They shift them slightly, add what she calls a psychological temperature.
De Havilland works with AI, but not in the way that conversation usually goes. She trains models on her own archive, paintings, travel photographs, old masters, sculpture and takes what comes back to oil on canvas. The model distorts, misreads, doubles figures unexpectedly. She keeps the accidents that interest her. When it returned a figure bent double with another on her back, she removed the second figure and replaced it with a swan. The result was Twinborn. The accident was the point.
Fvtvrist met Poppy De Havilland ahead of her exhibition in Los Angeles, visible until the June 13.

Iron Bone, 2025
Oil on canvas
Canvas dimensions: 150 x 150 cm
Framed dimensions: 152.6 x 152.6 cm
Copyright The Artist

Poppy de Havilland, portrait © Courtesy of the artist.
Sylvia Plath is named twice in these works: Lady Lazarus, Colossus Reimagined. The exhibition carries her even under its title. Plath wrote the body as both theatre and ruin, performance and relic. What drew you toward her writing, and what can a poem become once translated into paint?
The reference to Plath came after the paintings were made, rather than before. I wasn’t consciously working from her writing in the studio, although I have studied her in the past, and I think it’s likely her work remained somewhere in the background of my thinking. When the paintings were finished, I began to recognise an atmosphere in them that felt close to aspects of her work: transformation, performance, doubleness, resurrection, a sense that the self is never stable, and the body as something both vulnerable and monumental.
With Lady Lazarus and Colossus Reimagined, I wasn’t trying to illustrate the poems. The titles came later, almost as a way of naming a psychological charge that was already present in the paintings; Plath’s language offered a point of recognition rather than a starting point.
I’ve become interested in what happens when a literary reference enters a painting after the fact. It doesn’t necessarily explain the image, but it can deepen or destabilise it, adding a psychological temperature. A poem can suggest an emotional architecture, but once it enters a painting, it loses the sequence of language and becomes a single suspended moment.
In that sense, Plath was not the subject of the work, but a way of further understanding something I had already made.

1/ Lady Lazarus, 2025
OIl on canvas
2/ Interval, 2026
Oil on canvas
© Courtesy of the artist.
Cold Fire. Iron Bone. Unseen Gaze. Twinborn. Your titles often feel internally divided, as though each image is holding two psychological conditions at once. What does a painting gain from contradiction, and why does paradox interest you as an emotional structure?
I think painting is particularly suited to paradox because it doesn’t have to unfold in a linear way; multiple conditions can be present at once. The figure can look exposed and unreachable; an object can feel ordinary and symbolic; a scene can seem quiet but psychologically charged. I find this doubleness gives the work its emotional structure, allowing the painting to resist closure.
Ultimately, I’m interested in how these contradictions keep the image unstable and prevent it from becoming too legible.
Twinborn, 2026
Oil on canvas
Canvas dimensions: 150 x 130 cm
Framed dimensions: 152.6 x 132.6 cm
Copyright The Artist

Exhibition View, Richard Heller Gallery
© Courtesy of Richard Heller Gallery
The same young woman moves repeatedly through your paintings. Her expression often hovers between intimacy and distance, vulnerability and opacity. She rarely fully reveals herself to the viewer. What kind of emotional condition interests you most in this figure, and what must remain unknowable about her?
I often think about the lost or inward gaze found in Italian Renaissance painting - the sense of a figure looking beyond the viewer, or through them, as though absorbed in another psychological register. In my paintings, some figures look directly outward, such as in Twinborn, with an awareness that they are being observed, and that awareness creates a distance of its own. Other figures turn away entirely, as in the paintings seen from behind, such as Lady Lazarus and Unseen Gaze, where I was also thinking about Hammershøi and the quiet psychological force of a turned back.
There is a charged stillness that runs through many of the paintings. It is an atmosphere I return to often: a sense that the figure is holding something inwardly, but that we can’t quite know what it is. That ambiguity is important to me because it protects the figure from becoming too available to the viewer.
Many of your images feel suspended between dream-state and ritual. There is a sense that the objects, animals, and gestures obey an internal symbolic logic. How consciously do you construct mythology inside the work, and how much emerges intuitively during the painting process?
When creating the paintings, I try to build an atmosphere that feels slightly otherworldly, or difficult to place in time. I’m interested in images that sit between the familiar and the strange, where bodies and objects obey an internal logic rather than the ordinary laws of physics. Gravity often feels uncertain in the paintings: figures balance in improbable ways, as in Iron Bone, objects appear suspended, and the compositions sometimes resist a stable sense of weight or ground.
I don’t deliberately place mythology into the paintings. If mythological associations arise, they tend to come later, either through my own reflection or through the viewer’s interpretation. For example, Twinborn began much more as a psychological piece. I was thinking about power dynamics, shared psychology, and the possibility of the animal and the figure becoming one structure within the composition. It was only after the painting was complete, when I was searching for a title, that I began to think about Leda and the swan and Castor and Pollux. Since then, many viewers have brought those references to the work themselves.
I'M INTERESTED IN IMAGES THAT SIT BETWEEN THE FAMILIAR AND THE STRANGE, WHERE BODIES AND OBJECTS OBEY AN INTERNAL LOGIC RATHER THAN THE ORDINARY LAWS OF PHYSICS.
Before the brush comes the diffusion model. You feed it fragments from your archive: travel photographs, Renaissance painting, sculpture, previous works, textures, memories. The machine returns something suspended between recollection and invention. What is the algorithm to you ultimately: a collaborator, a subconscious, a translator, a stranger working in your handwriting?
I think of the algorithm as a collaborator, although not in a straightforward or traditional sense. I trained my first model around four years ago, and since then I’ve trained many more, each time feeding in the strongest and often most recent of my artworks. Over time, the models have developed alongside the work itself. They are shaped by what I have already made, but they also return possibilities that push the paintings somewhere else. In that sense, the process can feel almost reciprocal, as though we are feeding off one another.
I often begin with a concept I know I want to paint, but the model will return something slightly different, and very often that difference is what becomes interesting. It might misunderstand the image, distort a body, merge two forms, or suggest a compositional relationship I hadn’t anticipated. Sometimes I prefer what it gives back to my original idea. At other times, the opposite happens: the generated image clarifies what I don’t want, and I return more firmly to my own intention.
Twinborn is a good example of this. The initial idea was for the figure to be bent over, but the software returned an image with another figure on her back. That unexpected doubling opened up the idea of a power dynamic and a shared psychology. I removed the second figure and replaced that human weight with the dense presence of a swan, but the original accident was important. It shifted the emotional structure of the final painting.
So in the way I use it, the algorithm is not simply a tool for producing images, it is a space of exchange, where my previous work, my current intentions and the machine’s distortions meet.

Artist’s studio

