[ AI Short Film ] The Twilight Folk

A long time ago, in the forgotten age of a distant world, a hidden ritual took place in the depths of a sacred mountain that stood tall in the heart of an untamed jungle. The ritual carried a fading civilization’s hope that beyond the darkest night lies the warm embrace of a new dawn. Unbeknownst to her folk, as the chosen daughter of twilight descended into the dark, cavernous labyrinth beneath the mountain, a distant light was turning its gaze towards her fading world.

In a nearer future, at the twilight of a civilization facing a shrouded plight, the radiance of progress pierces nature’s once impenetrable veils. Unbeknownst to his folk, as the son of dawn navigates the labyrinth of a culture lost to time, the light of yore is obscured by the moon, casting a dark shadow upon his new world.

ABOUT THE FILM

The Twilight Folk is a short AI film that I have created throughout August 2025 to February 2026. The film’s visuals have been produced using various AI generative models accessed through the Google AI Studio and Leonardo AI platformsThe soundtrack in the background is composed by a niche, Swedish ambient composer named Ulf Söderberg. The film’s story and narrative flow have been devised following a significant degree of subjective inspiration garnered from Ulf’s track and greater body of work throughout the span of several years.  

Pragmatically and operationally speaking, the film is a conceptual teaser for a bigger series that I wish to direct at some point in the future when certain requirements are met. Moreover, the film aims to showcase a practical and artistically purposeful way of using generative AI technologies within the modern creation process of audio-visual experiences. I believe such showcases are necessary within the current socio-cultural climate, which is becoming increasingly saturated with the more nefarious uses of these technologies.

Artistically and narratively speaking, the film is an interplay of dualities that seeks to bridge seemingly disparate polarities within a synchronistic framework; man and woman, past and future, nature and technology, regression and expansion. Two civilizations are facing a slow but definite end, yet they resist it, resulting in unexpected twists and turns that redefine the relationship between cosmic alignment and human willThe eclipse during which a ritual was carried by an unknown, jungle-dwelling Bronze Age civilization acted as the announcement of the paranatural salvation, an outcome enabled by a major scientific undertaking in our own world’s near future that took place within the depths of the same mountain. As the conditions in one plain mirrored the ones in the other, both the man and the woman were copied and transported to their counterpart’s world, an intrusion that, unbeknownst to them, would signal both destruction and rebirth.

From a more expanded perspective of the film’s story, there are two ways of looking at the events that have transpired, one for each of the folks presented in the film.

The jungle-dwelling group, that the main female character belongs to, is a hypothetical, Bronze Age, Amazon jungle off-shoot of the Caral-Supe civilization that could have formed during their ancient, famine-driven migrations. The Caral-Supe are said to have had strong textile traditions and considerable relations with the Amazonian hunter-gatherer tribes to a point where intermingling is considered plausible.

For them, the mountain whose depths house the labyrinth, is both home and sacred space. The labyrinth was shown to the original hunter-gatherer tribes in the region throughout repeated, intergenerational visions during shamanic ceremonies and even individual dreams. Once the Caral-Supe off-shoot migrated to the jungle and began intermingling with the local population, the visions started to appear throughout their own rituals as well, with the mountain standing as a central figure in them, along a white labyrinth illuminated by either a blinding, white light, or an ominous, red glow. Burdened by the challenges of their existence in the jungle and the potential limitation or even extinction of their lineage, a convergence of spiritual beliefs belonging to both the migrants and the hunter-gatherers had led to the creation of a young, yet rapidly expanding new religion that regarded the creation of the labyrinth and the subsequent ritual around it a top priority; and perhaps the only salvation of their civilization. The visions had also given them a hint as to the timing of the ritual; a total darkening of the twilight skies, when the sun’s light turns white along the round edges of the moon.

Alternatively, the main male character, along with the scientists depicted in the film, is part of a mid-22nd century Amazon research group that works within a sprawling complex built around the area of the mountain that the jungle-dwelling folk call (or once called) home.

For them, the mountain is the main site for one of several scientific hubs in the Amazon region that belongs to an international research conglomerate, covering areas of research such as edaphology, hydrology, hydroponics, equatorial climatology, biotechnology, environmental engineering, nanotechnology and autonomous AI systems, zoology and zoonotics, anthropology, and various endemic studies within the jungle’s microclimate. However, unbeknownst to many of the scientists at the complex, there is one novel area of research shrouded in secrecy, which is Resonant Spacetime Dynamics. This young area of research focuses on the study and manipulation of unusual, georesonant energy signatures, which in this context, stem from the mountain itself, and are most concentrated in a hollow room where the remains of an ancient, labyrinthian structure have been discovered. Moreover, well-preserved drawings on the structure’s walls depict a process taking place under certain conditions, where two worlds must bring forth one of their own. Considering the curious nature of the mountain’s energy signatures, perhaps there could be more to the ancient instructions than meets the eye, something that could take the research down an unexpected path. All this is essential, as any scientific undertaking, no matter how exotic, could lead to new avenues of undoing the main threat to their world; the slow genetic degeneration of the human species as a result of the many abuses conducted throughout the previous, 21st century. While the degeneration’s first observable effects will become more accentuated after at least three more generations, the scientific consensus is clear on the fact that humanity’s lifespan as a healthy biological race has been reduced to only a few centuries, thus prompting the need for expedited innovation throughout all sciences, without the luxury of selectivity.

As far as the vision behind the ritual and experiment is concerned, the blindfolds and the labyrinthian progression towards the “meeting” point in the centre where the two characters touched hands plays into the idea of quantum superpositioning, which implies that a system could exist in multiple states until measured or observed. In a way, I have imagined that by depriving the two characters of the sense of sight, they, which in this context represent the sole “observers”, would become more malleable to the dimensional thinning around them caused by the experiment of the 22nd century scientists. These conditions had allowed the transition of each character to a different time(line), world, or dimension, which would have been impossible if left fully aware of their surroundings through the sense of sight, as the human eyes are wired to construct a fixed reality instead of conveying a more flexible one. Moreover, the undulating progression down the labyrinth and the necessity to slide one’s hand on the wall for guidance channels the brain’s focus towards a singular point of contact, and consequently, the hemisphere it is associated with. Upon reaching the entrance to the centre of the labyrinth, the right hand loses contact with the wall, and thus the person, as instructed before the ritual/experiment, must switch to their left hand in order to meet the wall that will guide them down to the centre. Considering the brain’s increased focus on the right hand upon losing contact with the wall, and the sensory deprivation enabled by the blindfolds, switching to the left hand would intensify the brain’s hemispheric switch from left (logic and delimitation) to right (intuition and holism). This allows for a very brief window where the “observer’s” mind loses its ability to concretize an external reality and can thus be influenced by the external energetic conditions enabled by the experiment, resulting in an interdimensional switch. If the right conditions are met, one’s hand would then meet their counterpart’s that started from the other end of the bicursal labyrinth, and a different world altogether. Ultimately, however, both “observers” are copied to the other dimension upon what could only be considered a coin toss, as one perceptual instance of the individual stays in their own world, while another one makes it to the other side.

While I am aware that my understanding of these scientific concepts might be skewed, I nonetheless believe this framework could add believability and perhaps even plausibility to the paranatural element introduced in the film. Moreover, I believe the outcome presented in the film could constitute a very solid premise for a future series that would base its narrative upon it. Some components for the series’ expanded story currently exist in a very conceptual state. 

Apart from everything, however, there is one key consideration I wanted to respect throughout the film, which is the implication that neither the world of the past nor the world of the future invalidate the nuance and importance of the other, and that humanity’s growth and existence is defined by omnidirectional movement. The people of the past are not all savages, and the people of the future are not all arrogant, Saturnian tyrants. Ultimately, this is a story about two people and their folk, gazing at the twilight of their civilizations, and learning that the salvation of one is enabled by the other.

PRODUCTION JOURNAL

The production of The Twilight Folk had spanned across 5 months, having started on August 26th 2025, and ended on January 28th 2026. The work on the film was conducted on a part-time basis. The film had cost around 450 USD to produce.

The production had been split into three stages:

1. Concept (August 26th 2025 – November 16th 2025)

This stage had been comprised of the following activities:

  • Script writing
    • The song used in the film, Skymningsfolket by Ulf Söderberg, had played a pivotal role during this stage, as the entire narrative flow of the script had been tailored to its progression.
  • Bulk image generation and editing via the free, preview models of Google AI Studio.
  • Auxiliary editing of some of the generated images using Adobe Photoshop to influence the preview models’ results more effectively.
  • Creating a boardomatic of the film with the preview pictures in Adobe Premiere Pro that would be later be used as the primary editing reference during the Video Production stage of the project.

The following image generation models had been used during the Concept stage:

  • Imagen-4
    • The main model used for generating complex scenes from scratch.
  • Nano Banana Preview (Gemini 2.5 Flash Image)
    • The main model used for generating simpler scenes from scratch and editing existing pictures in a creative manner.

These models had been used on the Google AI Studio platform on a free of charge basis. Imagen-4 was limited to around 10 generations per day, and Nano Banana Preview was limited to chat windows of around 32000 tokens each. It is important to note that while each individual prompt window for Nano Banana Preview was limited to a fixed number of tokens, the number of prompt windows itself was not, thus resulting in essentially hundreds of pictures and modifications generated free of charge.

2. Concept Refinement (November 24th 2025 – January 4th 2026)

This stage had been comprised of the following activities:

  • Refining the images generated through Google AI Studio’s preview models using higher quality models accessed through a paid plan on Leonardo AI.
  • The refinement of the images was done through a synthesis of four methodologies:
    • Upscaling the higher quality preview images through Leonardo AI’s Ultra upscaler.
    • Prompting the higher quality models to creatively upscale, add, or remove elements from the rough reference images.
    • Using Adobe Photoshop to manually edit certain preview pictures so that they would constitute better references for the higher quality image generation models.
    • Using preview pictures as a reference for new text prompts through tools such as Prompt Catalyst, so that the higher quality models would generate the scene from scratch using the reference’s contextual information.

The following image generation models had been used during the Concept Refinement stage:

  • Nano Banana Pro
    • A very powerful model used for editing and creating new scene variations from existing elements.
  • Seedream 4.0 and 4.5
    • The main model used for chromatic retouching, as it favors rich contrasts and warm, artistic hues.
  • Flux.2 Pro
    • The main model used for creatively upscaling textures on the rougher preview images and for retouching scenes where rugged, muted realism is favored over high-contrast, cinematic hues.
  • GPT Image-1.5
    • A very powerful all-rounder used for making difficult and more obtuse edits to reference pictures that the other models struggled with.
  • Lucid Realism
    • One of Leonardo AI’s proprietary generation models. It was used primarily on a few scenes that needed to be created from scratch using new prompts generated from rough reference pictures using tools such as Prompt Catalyst.

3. Video Production (January 5th 2026 – January 28th 2026)

This stage had been comprised of the following activities:

  • Using the refined image references to generate videos of varying lengths that added both simple and complex movement to the scenes.
  • Using Adobe Premiere Pro to edit the film in accordance with the boardomatic made during the Concept phase.
  • Using Adobe After Effects to add subtle movement to otherwise static scenes and retouch the lighting and coloration of certain clips. Certain auxiliary effects have also been added to certain scenes.

The following video generation models had been used during the Video Production stage:

  • Veo 3.1
    • One of the main models used to animate the reference pictures. It favors crisp movements and visuals yet tends to apply an additional layer of anti-aliasing over reference images that are already detailed, thus resulting in improperly blended textures within the scene. This outcome has been mitigated through the application of compound blur to affected surfaces in Adobe After Effects.
  • Seedance 1.0
    • One of the main models used to animate the reference pictures. It was the perfect replacement when Veo 3.1’s anti-aliasing was too abrasive for the scene. It favors rich contrasts, smooth textures, and slow movements, yet it is prone to more severe hallucinations if the prompts are too simplistic. Additionally, its bias towards slower movements can result in unrealistic behaviors to elements that would otherwise need to look livelier (water, fire etc).
  • Kling O1
    • A good all-rounder that favors a blend between Seedance 1.0’s contrast and Veo 3.1’s movement. It has good improvisational range on close-ups, yet if the reference image features distant elements, the model might decide to ignore them and keep them static. Therefore, precise prompting is needed to maximize its potential.
  • Sora
    • An expensive yet precise model that had been used sparingly for scenes that the other models struggled with, such as fully top-down shots.
  • Hailuo 2.3
    • An auxiliary model used for scenes with minor movement.

PRODUCTION GALLERY & CONCEPT DELIMITATIONS

The Woman and her Folk

The tribal-looking group that the main female character belongs to is a hypothetical, Bronze Age Amazon jungle off-shoot of the Caral-Supe civilization that could have formed during their ancient, famine-driven migrations. The Caral-Supe are said to have had strong textile traditions and considerable relations with the Amazonian hunter-gatherer tribes to a point where intermingling is considered plausible.

Moreover, within this context, the main female character in the film, along with her people, were intended to represent a stylistic synthesis of peruvian, afghani, and dacian styles. 

The peruvian influences ground the characters in the Amazonian setting and could add plausibility to a Caral-Supe connection or other Andean groups or civilizations.

The afghani influences add motifs related to ceremonial veiling, as their textile traditions feature an abundance of head and facial covers. Moreover, the stylistic, desert-dwelling elements of real-life traditional afghani outfits could complement the Caral-Supe’s own desert-dwelling origins in both practicality and symbolism.

The dacian influences add an aesthetically pleasing white and red contrast that could be symbolically associated with innocence and initiations, as well as complementing the other influences through increased visual richness and even fractality. Moreover, the red and white combination was prominent during the experiment conducted by the future folk in their own mirrored labyrinth, thus enabling more mystical storytelling avenues and interpretations.

The basis for the diversity of aesthetic influences in this fictional scenario could be that this group’s world is not our past world, but an alternate timeline where intercontinental travel and trade (for both goods and slaves) is considerably more developed during the Bronze Age than it was during our own timeline. Consequently, in this world-building scenario, certain technological, cultural, and societal advancements associated with the Old World Bronze Age could have made their way to the New World earlier than they did within our own recorded history

Additionally, certain outfits associated with desert-dwelling now become plausible in the context of a desperate migration to the jungle in search for fertile soils and a new foundation for civilization, while the other stylistic elements could represent an either partial or complete societal adherence to an occult religion that has formed as a result of intercultural mingling. and stylistic 

Concepts for the jungle-dwelling folk. The AI models were prompted to synthesize peruvian, afghani, and dacian influences

Iterative concepts for the main female character, whose final design ended up featuring more prominent dacian elements

The Man and his Folk

The main male character, along with the scientists depicted in the film, is part of a mid-22nd century Amazon research group that works within a sprawling complex built around the area of the mountain that the jungle-dwelling folk call (or once called) home.

Their style was designed to be mostly white and minimalist, with a slight touch of brutalism. Moreover, certain elements pertaining to analogue technology have been prioritized to hint at potential cultural shifts that might have occurred throughout the past century.

I did not want to use overtly futuristic elements, relying instead on a more utilitarian approach to interior spaces, considering their purpose is to host a vast array of scientific and logistical processes.

Iterative concepts for the main male character

The Labyrinth

The design evolution of the “past” labyrinth, which is mirrored by the folk in the future (or alternate dimension). This was an exceptionally challenging element to iterate upon as most AI models had seemed to struggle with following fully top-down perspectives.

The design evolution of the “future” labyrinth, the dome that houses it, and the technology within it. While the layout of the labyrinth is mirrored by the one in the “past”, it is the technology of the future that caused the space-time thinning in the room and allowed the crossing of the two characters (an event that the eclipse announced in the past).

Expanded story concepts

Various concepts that expand upon the short film and could be explored through a bigger, full-fledged series, or a short sequel. While the majority of the ideas do not currently have an associated image, I will upload more concepts throughout the foreseeable period. I’ll leave the narrative implications of the pictures below to your interpretation.