Taiwanese Artist Li Yi-Fan Will Make You Feel Screen Melancholy

Venice’s Bridge of Sighs is so called because it was where prisoners, transported between the interrogation rooms of the Doge’s Palace and the New Prison, caught their final glimpse of the outside world between the limestone window frames. Now, entering the room that once housed that prison, the Palazzo delle Prigion, you find yourself faced with a surreal sight. Screen Melancholy (2026) is an exhibition by Taiwanese artist Li Yi-Fan that presents a simulation of the palace inside its own walls. On screen, the interior is populated by digital bodies who make movements based on a performance by the artist, who uses a VR headset and hand-held controllers. The mannequin-like figures, whom we see giving a lecture, possess a curious theatricality.

Screen Melancholy, which is nearly an hour long, follows a puppet with milky, clouded eyes and an eyeball suspended over its head as it appears to slide from a screen displaying an anal opening and begin addressing another person, before a larger puppeteering avatar turns to face the viewer, who’s standing inside the palazzo. The digital bodies appear as iterations of a single human: Asian-presenting, naked, almost androgynous bodies with exaggeratedly curved spines. They seem to have just left an embryonic state. 

“As the characters learn to make animations, we slowly come to understand the mechanisms that Li believes are shaping how we see today”

We watch as the eyeball tutors a family of organs—including one figure crowned with a stomach and another with an appendix—on basic animation literacy, through playful, meme-inflected fragments, its tone pedagogical and deadpan. The eyeball figure attempts to educate the other organs about the shifting role of vision, all while searching for the absent brain, which remains missing throughout the story. Each segment is calibrated, the Amsterdam-based artist explains to me, to the “ideal attention span” metrics calculated by apps. It becomes, in his words, “a story of the eye”. As the characters learn to make animations, we slowly come to understand the mechanisms that Li believes are shaping how we see today. 

Li’s acclaim has been growing in recent years, and he was awarded the Taishin Arts Award in 2022, a major recognition in his home country, but his selection to represent Taiwan in Venice still caught him off guard—and this is the first solo Taiwanese pavilion since 2015 to feature an emerging artist. Li’s work is most familiar to audiences in his home country, where its “nonsensical humour… is recognised as a defining feature”. The visuals on show in Venice are funny, but also visceral and discomforting; the artist describes the central challenge in occupying the Taiwan pavilion as tuning his Taiwanese sense of humour in a way that will produce “shared laughter across audiences of differing backgrounds”. He sees this as coming down to whether or not audiences identify with the film. “To laugh at the same joke is to recognise something in common,” he says. “It calls for reflecting on how what Taiwanese audiences find funny translates across other viewpoints.”

“To laugh at the same joke is to recognise something in common”

Li was born in Taipei, the capital of Taiwan, in 1989, and spent much of his adolescence in front of video games and films like Dune (1996) and The Terminator (1984). He sees himself as possessing insight into tech culture because of his familiarity with its texture; he describes himself as being like an “addicted user”. His first short film, howdoyouturnthison (2021), used memories his of his teenage years to produce the architectural space of the film: a dark, basement-like room housing a huge screen. The title references a cheat code from Age of Empires II, a war game set in the Middle Ages. Li’s film adopted a first-person perspective: we move through dim interiors, the viewfinder trembling in a way reminiscent of shooter games. Its subjective, destabilised style anticipated the explosion of machine-generated reel culture, the AI Shrek-style remixes and deepfake mash-ups. As reel-culture has exploded, the association has propelled a rapid rise in his visibility.

This film was followed by What Is Your Favourite Primitive (2023), a 37-minute nocturnal excursion through a puppeteering production studio in a rundown theatre. This film led to broader international exposure: he exhibited at the Taipei Biennial in 2023, the Leeum Museum of Art’s emerging talent survey 2024 Art Spectrum and the 2025 Mercosul Biennial, which was where Li first met Raphael Fonseca, whom he invited to curate his Venice Biennale pavilion.

For Venice, Li wanted to push his experiments with language. In the film, the avatars stage a polyphonic monologue in a strangely neurotic register. The speech drifts between accented English and occasional Mandarin, and they have a slightly impeded, lisped articulation; even for those familiar with Taiwan’s bullshit-talk register, this performance plays it up. It is at times awkwardly unhinged, at others so bad it’s good.

“Even for those familiar with Taiwan’s bullshit-talk register, this performance plays it up. It is at times awkwardly unhinged, at others so bad it’s good”

Li explains that the figures emerge from a simple set-up: he wears a VR headset, controls the puppets with handheld sticks, and performs inside a game environment first developed in Taiwan in 2021. Within this set-up, motion capture does not register tongue movement, so Li deliberately speaks without moving his tongue, while the puppets’ contorted motions stem from his inability to control them precisely—limitations that produces a distinct vocal and performative identity.

Parts of the film come across like meditations on computerised animation. In one sequence, a head detaches from its body, drifts through space, and finally submerges into the screen, reappearing as a talking head. It delivers a lecture on functions such as “random()” and “noise()”, which produce synthetic image textures. In contrast to his earlier works that drifted through loosely assembled comedic fragments and mumbling digressions, this film is tightly structured, and uses looped, at times layered, narrative arcs.

One of the film’s most arresting passages explores the ethics of AI, which Li identifies as centred on “war imagery, pornography, [and] platform regulation”. This strand unfolds as a mini-lecture on a question prompted by a 2023 controversy in which Adobe was found to be selling AI-generated images of violence related to Gaza and Israel through its stock platform. The sequence opens with an ultra-realistic AI-generated tank that suddenly thrusts into the centre of the pavilion, before the lecturer intervenes to clarify that it is AI. The sequence plays out the artist’s approach to image-making: he believes that AI should always be a “protagonist we discuss rather than deployed simply for spectacle”.

Towards the end, the film takes on a Russian doll-like structure, where each level gives way to another showing puppets manipulating puppets, all set within progressively scaled-down models of the palazzo. At the final layer, the structure appears to collapse into its own core. The ground lifts up a hollow animatronic head, its skeletal interior exposed. The avatar enters and the head activates and turns toward the viewer, and begins to cry viscous, gently glittering tears.

“Few years ago I saw someone crying on this island,” the character says. It is a reference to the seminal 2017 animation Old Food by the British video artist Ed Atkins, a work renowned for its artificial effect and its uncanny proximity to human expressivity. Li encountered Old Food during the Venice Biennale in 2019, a trip he now describes as a turning point. At the time, he was technically proficient and involved in post-production work for other artists. “After that trip,” he recalls, “I felt able to move beyond this hesitation and pursue it more fully.”

In Screen Melancholy, the crying scene becomes an experiment: how does crying or, rather, “crying”, mutate across successive image regimes from CGI to AI? As it plays out, we watch each iteration subtly recalibrate what counts as legible emotion. “But I suppose now we all cry differently,” the character adds. As we watch, we realise that Li’s reference to Atkins doesn’t function as an homage. It’s just one piece of data in a huge bank of “geeky” tech references from which the work is pulling. Instead of elevating Old Food as an art reference, Li flattens it, turning it into just another reusable effect within a system of generated images. If Atkins was pushing synthetic affect to its limit, Li shifts the question somewhere else: to the actual systems that render feeling legible. —[O]

Main image: Li Yi-Fan inside the Palazzo delle Prigion. Photo: Mihail Novakov for Ocula.