on soft cinema, Harry Bayley
April 2025
Left: Lev Manovich in collaboration with Andreas Kraty, Soft Cinema: "Texas" - Screenshot from the DVD, 2003/2005
Right: Subway Surfers brain rot on TikTok, 2024
Staring blindly at the screen, scrolling through the fifth Family Guy x Subway Surfers TikTok in a row, there is a moment of overlap. Our subway surfer Jake almost appears to be running through the Peak District as the two videos and my interests collide. For a moment, I wonder if there is some kind of poetic potential within the TikTok machine.
In 2002, Lev Manovich opened a media installation he called Soft Cinema. Soft Cinema was constructed from a large media database and custom software that edited films in real-time.* Manovich’s installation was an algorithmic, semi-automated cinema machine that often utilised multiple screens within the frame and different formats. It combined different kinds of media to interrogate the influx of images that came with the introduction of the digital image in the late 1990s. Soft Cinema sought to provide a way of processing this imagery, navigating an increasingly confusing, exponentially expanding media landscape.
Soft Cinema looked to the future of “online” filmmaking and the forms moving images would take in the digital world. While flicking through Manovich’s dossier on the project, I couldn't help but notice the similarity between soft cinema and TikTok's iteration of picture-in-picture form.
As you scroll through your For You Page, the TikTok algorithm curates your content stream like a tastemaker. It reflexively recommends content it believes you will engage with, plotting a path from the material to the immaterial plane. The longer I engage with TikTok, the videos I find tend to drop off a cliff edge in quality. As my brain tires from the onslaught of images, I start heading for the quickest dopamine hit. These come in various forms of “brain rot content”, but the ones which interest me the most are those “brain rot” videos which fill the screen with two (or multiple) concurrent videos. Often these videos are constructed with footage from popular mobile games on one half and the other, clips from films or TV shows.
This method of consuming media injects dopamine straight into the user's eyes, not giving them a moment to think. It sweeps up their attentional gaze with an overwhelming amount of information. These videos are generally seen as poor, low-effort attempts to farm interaction and watch time. Despite this, they are still wildly popular. I find this form of picture-in-picture videos conceptually engaging. The practice is closer to experimental film than much else, utilising multiple frames and montage within the image to create something new.
The common thread of the visual similarity between Soft Cinema and TikTok brain rot, the multiplication of the image, is attractive as it enables an individual to touch the surface of cyberspace, and caress multiple objects at once. An attempt to access the other side of the screen through overexposure to the immaterial, to leave the physical world behind, even just for a moment.
Unlike TikTok, Soft Cinema’s content is exclusively created by the machine, despite this, films within Soft Cinema feel more genuine, and alive than that of the majority of human-made TikToks. Why is this? With the rise of AI-generated content on TikTok this question becomes more important, as this becomes a reality. What purpose do these machine-made videos serve and what effect do they have on the user?
In 2002, Lev Manovich opened a media installation he called Soft Cinema. Soft Cinema was constructed from a large media database and custom software that edited films in real-time.* Manovich’s installation was an algorithmic, semi-automated cinema machine that often utilised multiple screens within the frame and different formats. It combined different kinds of media to interrogate the influx of images that came with the introduction of the digital image in the late 1990s. Soft Cinema sought to provide a way of processing this imagery, navigating an increasingly confusing, exponentially expanding media landscape.
Soft Cinema looked to the future of “online” filmmaking and the forms moving images would take in the digital world. While flicking through Manovich’s dossier on the project, I couldn't help but notice the similarity between soft cinema and TikTok's iteration of picture-in-picture form.
As you scroll through your For You Page, the TikTok algorithm curates your content stream like a tastemaker. It reflexively recommends content it believes you will engage with, plotting a path from the material to the immaterial plane. The longer I engage with TikTok, the videos I find tend to drop off a cliff edge in quality. As my brain tires from the onslaught of images, I start heading for the quickest dopamine hit. These come in various forms of “brain rot content”, but the ones which interest me the most are those “brain rot” videos which fill the screen with two (or multiple) concurrent videos. Often these videos are constructed with footage from popular mobile games on one half and the other, clips from films or TV shows.
This method of consuming media injects dopamine straight into the user's eyes, not giving them a moment to think. It sweeps up their attentional gaze with an overwhelming amount of information. These videos are generally seen as poor, low-effort attempts to farm interaction and watch time. Despite this, they are still wildly popular. I find this form of picture-in-picture videos conceptually engaging. The practice is closer to experimental film than much else, utilising multiple frames and montage within the image to create something new.
The common thread of the visual similarity between Soft Cinema and TikTok brain rot, the multiplication of the image, is attractive as it enables an individual to touch the surface of cyberspace, and caress multiple objects at once. An attempt to access the other side of the screen through overexposure to the immaterial, to leave the physical world behind, even just for a moment.
Unlike TikTok, Soft Cinema’s content is exclusively created by the machine, despite this, films within Soft Cinema feel more genuine, and alive than that of the majority of human-made TikToks. Why is this? With the rise of AI-generated content on TikTok this question becomes more important, as this becomes a reality. What purpose do these machine-made videos serve and what effect do they have on the user?
Left: Family Guy and Subway Surfers brainrot on TikTok, n.d.
Right: Lev Manovich in collaboration with Andreas Kraty, Soft Cinema: "Mission to Earth" - Screenshot from the DVD, 2003/2005
The big point of differentiation between TikTok brain rot and Soft Cinema is not only TikTok’s widespread use but also its position as a commodity. The app is driven forward not by discovery but by capital, and it is projected to have generated $18.5 billion in advertising revenue in 2024 alone.
Steven Shaviro explores the concept of Post-Cinematic affect, a cinematic practice which is politicised, centring affective labour at its core, a cinema more concerned with conversing with the non-human/posthuman/ahuman one which focuses on the immaterial, feelings and emotions above all else – arguing that financialisation is an “insidious banal influence on all aspects of contemporary existence”.
This perspective advocates for approaching machines with the intention of communication instead of a tool/user power dynamic. Engaging with media in this post-cinematic way enables us to touch its surface and feel affect from the inanimate. This point of contact suggests a reason why Soft Cinema’s algorithmic creations feel human, and why in the present we are seduced by and constantly romanticise technology of the (recent) past.
Employing this perspective here allows us to better locate the differences between TikTok and Soft Cinema. Saying that Soft Cinema is post-cinematic allows a closer look into the poetics of our relationship with the Soft Cinema machine, how through its selection of media we are in dialogue. Reflexively it creates as we watch and interact with its physical interface, showing glimpses of its knowledge and information openly. There was a level of labour expected of you and the machine, yourself bathing under its display, while the machine plugged away at the various pieces of code and databases to create something to communicate with you, a dialogue between human and non-human, a chance to touch the digital realm. There is something poetic about the hardware, now long outdated PC’s working through almost analogue information creating their films, something romantic.
A key component of having poetic dialogue in this way is that of materiality and hardware’s ability to oscillate between the material and immaterial, human and non-human. This oscillation is explored explored by Laura Marks in Video Haptics and Erotics (1998) where Marks argues that it is the oscillation between subject and object that attracts us to the machine:
The immaterial action of these algorithms can be seen as the object-making process in the same way but the materiality of Soft Cinema brings us back to subjecthood. This is in contrast to TikTok’s brain rot where its immateriality and design, wish us to remain as objects, not critical, not communicating only consuming. Contextualising the physical presence of Soft Cinema in the present day, the whole concept sounds ritualistic, an attempt at communicating with a different body, paying attention to the machine and watching it create something poetic. The exhibit itself was constructed of two private screening rooms and one public area designed by the same algorithm, the machine creating the room in which you are sitting, watching its creation from within its alter material “body” An intimate experience with a perceivably “innocent” “offline” machine, an artefact which will soon be as alien to humans as rituals from a preindustrial era.
Steven Shaviro explores the concept of Post-Cinematic affect, a cinematic practice which is politicised, centring affective labour at its core, a cinema more concerned with conversing with the non-human/posthuman/ahuman one which focuses on the immaterial, feelings and emotions above all else – arguing that financialisation is an “insidious banal influence on all aspects of contemporary existence”.
This perspective advocates for approaching machines with the intention of communication instead of a tool/user power dynamic. Engaging with media in this post-cinematic way enables us to touch its surface and feel affect from the inanimate. This point of contact suggests a reason why Soft Cinema’s algorithmic creations feel human, and why in the present we are seduced by and constantly romanticise technology of the (recent) past.
Employing this perspective here allows us to better locate the differences between TikTok and Soft Cinema. Saying that Soft Cinema is post-cinematic allows a closer look into the poetics of our relationship with the Soft Cinema machine, how through its selection of media we are in dialogue. Reflexively it creates as we watch and interact with its physical interface, showing glimpses of its knowledge and information openly. There was a level of labour expected of you and the machine, yourself bathing under its display, while the machine plugged away at the various pieces of code and databases to create something to communicate with you, a dialogue between human and non-human, a chance to touch the digital realm. There is something poetic about the hardware, now long outdated PC’s working through almost analogue information creating their films, something romantic.
A key component of having poetic dialogue in this way is that of materiality and hardware’s ability to oscillate between the material and immaterial, human and non-human. This oscillation is explored explored by Laura Marks in Video Haptics and Erotics (1998) where Marks argues that it is the oscillation between subject and object that attracts us to the machine:
What is erotic is being able to become an object with and for the world, and to return to being a subject in the world: able to trust someone or something to take you through this process: and to be trusted to do the same for others1
The immaterial action of these algorithms can be seen as the object-making process in the same way but the materiality of Soft Cinema brings us back to subjecthood. This is in contrast to TikTok’s brain rot where its immateriality and design, wish us to remain as objects, not critical, not communicating only consuming. Contextualising the physical presence of Soft Cinema in the present day, the whole concept sounds ritualistic, an attempt at communicating with a different body, paying attention to the machine and watching it create something poetic. The exhibit itself was constructed of two private screening rooms and one public area designed by the same algorithm, the machine creating the room in which you are sitting, watching its creation from within its alter material “body” An intimate experience with a perceivably “innocent” “offline” machine, an artefact which will soon be as alien to humans as rituals from a preindustrial era.
Lev Manovich in collaboration with Andreas Kraty, Soft Cinema: "Texas" - Screenshot from the DVD, 2003/2005
Since Manovich’s Soft Cinema exhibition, our relationships with the algorithmic machine have fundamentally changed, before tethered to the material by an individual engineer, limited by its own hardware both machine and non-machine work in partnership to create something new. A conversation of gestures and tinkering pioneering technology to see into other dimensions, falling into cyberspace. Now a part of everyday life, a self-replicating, independent immaterial machine, not concerned for poetry or connection, engineered for profit and submission. With non-machines reaching into the immaterial for escape only to be thrown back into reality at the first sign of resistance, only with a higher screen time.
Could this poetic dialogue be possible in contemporary online algorithmic practices? It is possible, that while online algorithms are most popularly used mainly for learning and recommending other people's creations, you can still catch glimpses of soft cinema within the neverending TikTok for you page if you try hard enough. In that moment when one video comes into frame and the other leaves in the seconds of overlap you have these split images, existing between the lines and folds of the TikTok algorithm.
If instead, we utilised these immaterial technologies for positive real-world affect instead of exclusively weaponizing them for financial gain. Then the dialogue between us and the spirits inside the machine may open up. But for now, we have closed the embassy and murdered all the diplomats.
It is up to us, the consumers, the creators to realign our relationship with these mysterious algorithms and platforms, currently, we work for them, create revenue from buying drop-shipped items from TikTok shops, and watch content designed to keep us scrolling. But it is people, not TikTok who create its content, we have no one to blame but ourselves.
Could this poetic dialogue be possible in contemporary online algorithmic practices? It is possible, that while online algorithms are most popularly used mainly for learning and recommending other people's creations, you can still catch glimpses of soft cinema within the neverending TikTok for you page if you try hard enough. In that moment when one video comes into frame and the other leaves in the seconds of overlap you have these split images, existing between the lines and folds of the TikTok algorithm.
If instead, we utilised these immaterial technologies for positive real-world affect instead of exclusively weaponizing them for financial gain. Then the dialogue between us and the spirits inside the machine may open up. But for now, we have closed the embassy and murdered all the diplomats.
It is up to us, the consumers, the creators to realign our relationship with these mysterious algorithms and platforms, currently, we work for them, create revenue from buying drop-shipped items from TikTok shops, and watch content designed to keep us scrolling. But it is people, not TikTok who create its content, we have no one to blame but ourselves.
Harry Bayley is a filmmaker and writer living in Taipei. Their films have been screened in the UK and internationally centring around the idea of accessing different dimensions through the combination of cinema and ritual. They recently completed his MA in Film and Screen Media with a focus on emerging forms of spectatorship, in particular transcendental and uncanny modes of watching. Their writing has appeared in Gamescenes, the Big Ship, and Queer East.
site https://www.hfilmt.com/
instagram @hfilmt
*. Soft Cinema incorporated 4 separate “pillars” into its identity:
1. “Algorithmic Cinema.”Using systems of rules, software controls both the layout of the screen (number and positions of frames) and the sequences of media elements which appear in these frames.
2. “Macro-cinema.”Soft Cinema imagines how moving images may look when the Net matures, and unlimited bandwidth and very high-resolution displays become the norm.
3. “Multimedia Cinema.”In Soft Cinema, video is used as only one type of representation among others: 2D animation, motion graphics, 3D scenes, diagrams, etc.
4. “Database Cinema.”The media elements are selected from a large database to construct a potentially unlimited number of different narrative films.
1. Marks, L. U. 2002. Touch: Sensuous Theory and Multisensory Media.
site https://www.hfilmt.com/
instagram @hfilmt
*. Soft Cinema incorporated 4 separate “pillars” into its identity:
1. “Algorithmic Cinema.”Using systems of rules, software controls both the layout of the screen (number and positions of frames) and the sequences of media elements which appear in these frames.
2. “Macro-cinema.”Soft Cinema imagines how moving images may look when the Net matures, and unlimited bandwidth and very high-resolution displays become the norm.
3. “Multimedia Cinema.”In Soft Cinema, video is used as only one type of representation among others: 2D animation, motion graphics, 3D scenes, diagrams, etc.
4. “Database Cinema.”The media elements are selected from a large database to construct a potentially unlimited number of different narrative films.
1. Marks, L. U. 2002. Touch: Sensuous Theory and Multisensory Media.