God is a Woman, Ariana Grande

Relationships between films and painting through the scope of our AI

Posted on Invalid Date

Since the creation of cinema, many filmmakers have used the imagery of Art, sometimes for the entire movie's atmosphere, sometimes just for a scene, a set or a costume. Looking for these visual references inside a movie is a kind of game. And with our AI-powered search engine, searching for any visual reference inside a catalog of stills from all the genres (movies, ads, music videos, TV series and anime) is now possible.

Our Artificial Intelligence is indeed able to understand, as a human brain, your search: artistic movement, name of a painter or title of a work of art. That’s why we’ve decided to write an article about paintings and cinema through the scope of our AI. A way to play with our stills database and talk about the dialog between art and video.

From tableaux vivants to cinematic shots inspired by artworks

Tableaux vivants (French for living pictures) are probably the first form of life-size adaptation of a painting. Very popular in Western culture especially at the beginning of the 19th century, these forms of entertainment consisted of static, costumed, carefully and theatrically set up scenes that could reproduce or be inspired by paintings. They were performed live, and from the late 19th century they started to be photographed and filmed.

Nowadays, the term tableau vivant refers both to its original and historical form and to life-size reproductions of paintings or static and theatrical scenes. In a sense, any shot that stages a static scene with costumes is a kind of tableau vivant, such as the series of images with Armenian costumes depicting verses from the life of the poet Sayat-Nova in The Color of Pomegranates.

Forrest Gump (Robert Zemeckis) and Christina's World (Andrew Wyeth), The Marquise of O (Eric Rohmer) and The Nightmare (Fuseli), Cabaret (Bob Fosse) and Portrait of the Journalist Sylvia von Harden (Otto Dix)… Many movie scenes were inspired by art in the style of tableaux vivants. While some are identically rendered, others are inspired by the general atmosphere (Millais Ophelia painting has inspired Justine floating in the water in her wedding dress in Melancholia), the setting (the famous snow scene in The Mirror is inspired by Pieter Bruegel’s The Hunters in the Snow), or the costumes (Leeloo's futuristic jumpsuit in The Fifth Element is inspired by Frida Kahlo’s The Broken Column)… You will find many articles on the Internet about the dialog between art and cinema. But often the same artworks are used as examples and limited to famous references.

Search by artwork: the AI's ability to recognize titles

Aren’t you a lucky one? Our AI is able to recognize the title of an artwork beyond its literal meaning. For example, if you type Girl with a pearl earring, AI understands that you are searching for stills inspired by Vermeer’s famous artwork and not only screencaps depicting a girl who is wearing pearl earring. This kind of search can be very useful if you have a specific search intent, such as the composition of a painting, or if you want to see what other filmmakers have done. And believe us, with Flim database you'll keep making connections you never intended in the first place.

So let us start with one of the most famous themes and paintings in Western culture: The Last Supper. Our AI finds stills directly inspired by the Leonardo da Vinci version, like the famous hippie version in Inherent Vice or the military and dreamy version from the music video Agoria - Embrace.

AI also detects shots with Leonardo's artwork or another version of The Last Supper: hung on the wall and surrounded by everyday objects in ATL, as a background that echoes the foreground in Carrie, or filmed by itself as in The Last Vermeer. These three examples are interesting because they set three ways to show a work of art in a film.

Filming a work of art is about the choice of context and staging, framing, lighting, angles, etc. The result is, in a sense, another work of art realized through the eyes of a filmmaker and with a specific meaning and artistic intent.

In the case of paintings filmed for themselves, as in The Last Vermeer, the place of these kinds of shots in the film edit is very important. They can express memories such as the bedroom in La Jetée inspired by Van Gogh’s Bedroom in Arles, visual and meaningful parallels like the paintings that punctuate the scenes in Pierrot Le Fou, or the close-ups of The Last Judgment in The New Pope etc., allowing us to make different associations of ideas and images.

About The Last Supper, our AI also finds screencaps with similarities (motifs, composition, lighting, framing, etc.) which are not clearly inspired by the Renaissance artworks. And these results are perhaps the most interesting thing about Artificial Intelligence. For example, some shots show only people eating (Back to the Future), people sitting around a table with horizontal composition (The Green Knight) or people standing around a table with religious iconography (Benedetta). Are these frames directly inspired by The Last Supper as Inherent Vice, or is it AI? In the end, it does not matter, because these suggestions are precisely the value of AI: it allows us to diversify the results and broaden the inspiration itself.


Thanks to Flim search engine, you can indeed input any visual reference and make an infinite number of visual connections: some will be obvious, others much more subtle. Let us explain why.

How do Flim’s AI search engine and neural networks work?

Flim is the only platform that brings together screencaps from all genres in one place (movies, TV series, ads, music videos, documentaries and animations) and provides you with a powerful search engine based on artificial intelligence and neural network technology instead of tag systems.

Tag systems can be very precise, but they are also very limited, since the results of a search are only the images associated with the corresponding words. AI search engine, on the other hand, allows for a more inspirational and diverse search.

Neural networks mimic the human brain and its biological processes through a series of algorithms. Thanks to its neural architecture, the Artificial Intelligence is able to interpret your search beyond its literal meaning and find instantly the stills in our entire database that best match its interpretation.

So the search on Flim is unlimited. You can type in whatever you want and get results just as a human brain would create mental representations: objects, feelings, situations, concepts, or visual references like brands or artistic movements.

Search by artistic movements: very useful for creating mood boards

The relationship between visual arts and cinema has existed since the invention of cinema. Artists and filmmakers have explored the cinematographic medium within their artistic movements. For example, the Surrealists used absurd imagery and "special effects" (Buñuel), or the German Expressionists (Fritz Lang) used high angles, deep shadows, etc. to transpose the codes of their artistic movement to the screen.

What interests us here are the films made within an artistic movement, but also those inspired by it. In other words, the filmmakers who allude to artistic movements without being part of the artistic movement to which they refer, such as Stanley Kubrick, who was inspired by the romantic paintings of Gainsborough and Constable in Barry Lyndon, or Hayao Miyazaki, who often refers to Impressionism in his landscapes.

Indeed, many filmmakers have transferred the codes of a pictorial movement into the cinematographic medium in order to assimilate them for their own meaning. Stanley Kubrick, for example, drew inspiration from Romantic music and painting, emphasizing the tragic and fatal destiny of his characters, while paradoxically insisting on the lack of romanticism of his main character, who is driven only by ambition. As for the inspiration from German expressionism, distorted perspectives, high angles, shadows, etc. emphasize the frightening and creepy atmosphere.

Therefore, looking for artistic movements on Flim is very helpful to create mood boards. Try with Pop Art: our AI provides stills from videos directly inspired by Pop Art, such as Warhol-style repetitions from Beyoncé's music video Countdown, but also screencaps from videos not directly inspired by Pop Art, but where the AI detects similarities: colorful stills, comic book-inspired iconography, close-ups on red lips or pop culture motifs.

With impressionism for example, you will get screencaps of outdoor painting's scenes in the late 19th century, impressionist painting or shot inspired by impressionist painting,

nature, lights, water lilies, etc.

Also very interesting results: shots of exhibition or museum spaces. In some of them the AI has detected impressionist paintings or at least paintings from the late 19th century on the walls, while in others the museum visitors are the subject, regardless of the type of paintings.

This last result is very interesting to fully understand how AI works. It mimics the human brain's interpretation process using neural networks and what it has learned about images. The AI was actually trained with a combination of images and text created by humans. All the screencaps that result from a search are, in a sense, the popular view of a subject, just as Impressionism - one of the most popular movements - is associated with the subject “exhibition” in the common view.

Search by painter: the AI's interpretation of an artist's style

Filming a painter is about inspirations and influences, creative process and gestures, life and artistic or historic context etc. depending on the genre and the intent of the filmmaker: tell a story or show the creative process. For example, while the biographical drama Mr. Turner tells us the last 25 years of the life of the artist and his universe, Henri-Georges Clouzot in The Mystery of Picasso shows the act of creating to precisely catch “the mystery” of creation through his camera.

When searching for a painter on the database, the results are shots inspired by his style and compositions (Edward Hopper), images with figures that look like the artist (Salvador Dalí), screencaps of the artwork (Caravaggio) or stills showing the artist at work (Picasso).

Let’s use the example of René Magritte, because this search is very representative of how the AI understands an artist and his whole universe. As explained above, it orders the search results by relevance.

The first stills are images with typical patterns of Magritte’s painting: men in bowler hats, cloudy skies, absurd iconography like these two men walking through an eye, or Christopher Walken flying in front of a seascape.

But AI also goes beyond motifs to try to understand an artist's art more deeply. René Magritte highlights our perception of things and our mental images, as well as the discrepancy between an object in real life and its representation. And this still from Perceval le Gallois, which depicts three red spots as eyes and mouth, or this one from Brazil, where a white handkerchief is flying as a cloud, are very representative of how AI tries to catch the concept / metaphor behind the images.

We could search for a long time for other artists or art movements, but also for situations like “painter in his studio painting his model” or materials like “watercolors”, but we leave it to you to play with the AI!

Back to blog