(2022-2023) experiments with early-generation AI image and video generators

Exploration of early-generation AI image and video generators (2022–2023) to create visual content. Trampuž researched how these tools could be used in his artistic process, including the creation of collage artworks and animations. The resulting generated results reflect the creative potential of early AI technology while exploring the boundaries of what was possible at the time. The experiments focus on themes such as environmentalism, the relationship between humans, nature, and technology.

Extreme Weather Animation

(2023) AI animation

In May and June of 2023, during early experiments with AI animation, a spinoff artwork was created using some of the first models capable of generating animated frames—Deforum and Stable Diffusion 1.5. This piece emerged organically while developing visuals for the Statehood Day Ceremony of Slovenia. As part of the creative exploration, it stands as a snapshot of a formative moment in AI-assisted animation, capturing both the technical limitations and the imaginative possibilities of the medium at that time.

AI tools: Deforum & SD 1.5 in ComfyUI

Synthetic Mirror

(2022) AI-generated images

In 2022, Trampuž conducted an experimental project using one of the most advanced AI image generators of the time, Midjourney V4, to mimic his own distinctive visual style. By feeding the AI with a selection of his earlier collages along with written text prompts, he explored how closely the technology could replicate his aesthetic. The resulting images depict surreal sceneries blending the natural and the artificial. The project served as both a creative investigation and a reflection on authorship, originality, and the evolving relationship between the artist and machine.

AI tools: Midjourney V4

Dreaming Is All I Ask For

(2022) digital collages in light boxes, assembled from AI-generated materials

With the collage series Dreaming Is All I Ask For, the artist reversed the often pessimistic thought process and speculated on what an idealized society of the future might look like—one capable of effectively addressing today's challenges, particularly in the field of environmental protection. He focused on questions of how to transform our egocentric mental framework, renounce excess and exaggerated comfort, re-establish a connection between humans and nature, and learn to coexist with it.

In creating the material for the collages, the artist used an early-generation artificial intelligence model—specifically Midjourney V3—which he regarded as a tool. The AI served as an experimental source of imagery, which he then selected, cut, combined, and transformed into digital collages. Through this process, he explored the possibilities of human-machine collaboration and tested the boundaries of artistic production emerging through technological cooperation.

Although he critically assessed the impact of modern technology on nature and society, he does not believe it should be entirely rejected. He maintains that technology—when used responsibly, thoughtfully, and with a clear purpose—can play a positive role. In this context, artificial intelligence functions as a tool for speculative reflection on the future, rather than as an answer to it.

AI tools: Midjourney V3

10sec/piece, 16h/day

(2022) digital collage in light box, assembled from AI-generated materials

In production process of the digital collage that explores the theme of fast fashion, I set myself two guidelines. These symbolically reflect how this industry operates:

1. The visual material used to create the digital collage was not made by me. Instead, I relied on a third entity – I "enslaved" artificial intelligence (AI). I used the Wombo Dream application, which is based on the open-source process VQGAN+CLIP. This process uses AI that has learned, through deep learning, the relationships between the meanings of words and their visual representations. The AI has the ability to visually imitate the meaning of words. This means that the user can input textual prompts describing the desired image – and the AI will generate it. The generation of each individual image takes about 10 seconds. For the collage, I selected from 272 freshly generated images.

2. From the beginning to the end of the process, including the writing of this text, I worked without interruption, and the piece had to be completed within 16 hours from the start. Sixteen hours is roughly equivalent to (or even less than) the length of a workday for industrial laborers, who often operate in toxic and otherwise inhumane conditions.

AI tools: Wombo Dream / VQGAN+CLIP