18 Ornagh | Broke Bad | Video Edition ArtIn

Broke Bad by Ornagh 3:30 min, 2024
Soundtrack by Anja Kreysing

 AIs: GPT, Midjourney, Meshy, Deep motion, Radical.  

     Art Statement: Broke Bad is a surreal exploration of obsession, addiction, and the intoxicating pull of a love that feels both thrilling and destructive. Inspired by personal experience, the work captures the high of being drawn to someone who is all wrong yet impossible to resist—the electric danger, the reckless abandon, and the inevitable crash. It is a Bonnie-and-Clyde kind of romance, where passion burns too bright to last, and loss is as dramatic as the love itself. Through distorted reality and heightened emotion, Broke Bad transforms raw experience into something mythic, unraveling the beauty and devastation of desire that consumes.
      At its core, the work interrogates systems of control—who gets to have a voice, and who decides what is heard. The journey to reclaim the tongue becomes an act of defiance, a reassertion of agency. In the end, the silenced erupt into movement, tongues breaking free in a final act of liberation.

      My use of AIs in XR:
Orna’s creative process is deeply intertwined with AI, using it as both a tool and a collaborator in her artistic practice. Across her work, she blends digital and physical realities, crafting immersive experiences that unfold across XR technologies, performance, and storytelling. AI-generated imagery and video play a crucial role in shaping her visual language, allowing her to build layered, surreal worlds that evolve dynamically. Motion capture, augmented by AI, breathes life into her digital performances, seamlessly integrating real-time animation with immersive narratives.
     In her work, AI is not just a tool for efficiency but an enabler of new creative possibilities. It accelerates ideation, refines concepts, and opens up new avenues for storytelling, from VR painting to digital sculpture. Her upcoming website is conceived as an AI-assisted interactive experience, where visitors step into a world that unfolds dynamically, revealing layers of story and meaning as they explore. She envisions a system where generative AI contributes to procedural storytelling, allowing narratives to shift and evolve based on user interaction.
      Her performances and mixed reality works incorporate AI-driven animation and Unreal Engine environments that respond in real-time to live performers. The integration of AI into these spaces enhances both artistic expression and audience engagement, creating an interplay between the digital and physical that feels alive and ever-changing. Her approach to AI is both intuitive and exploratory, using it to expand the boundaries of storytelling, performance, and visual art in ways that feel organic and deeply personal.

-------------------------
ORNAGH
XR Artist

info@ornagh.com
www.ornagh.com
IG: ornaghworld
Artist Name: Ornagh
Location: London, UK
Instagram: @ornaghworld
Linked in: /ornagh
AIs: GPT, Midjourney, Meshy, Deep motion, Radical. 

In the press:
The Cover of Get Hastings Magazine
Razvan Chiorean Gallery
Hastings Independent Press
Hastings Online Times 

 

21 Anja Kreysing | Tidal Interferences | Video Edition ArtIn

 Tidal Interferences by Anja Kreysing 3 min., 2025
Soundtrack also by Anja Kreysing

I worked on an experimental video with an own specific soundtrack composition. The video was intended to contend abstract and non-objective images, what was for the actual AI-applications (in March / April / Mai 2025) still a problem. 

Based on my own prompts, I used the resulting video clips generated by the Sora AI and developed them further through targeted variations. In each prompt, I emphasized what visual elements I wanted to keep and what should not appear again. I repeated this process multiple times until I had gathered enough usable clips – including a kind of open-ended closure for the planned non-linear narrative of the experimental video.

The process of working with the AI was slow, awkward, and at times bizarre. The video AI still feels quite stereotypical and limited in understanding – many generated clips were simply unusable, while others worked only when tightly cropped. I often had to intervene manually: adjusting speed, color palettes, or reversing movements to make the material usable.

The soundtrack was entirely performed by me. However, for two widely used synthesizers, I asked ChatGPT to generate or design custom sound patches. Again, while the AI results weren’t simplistic, they were often too predictable or technically unrealistic. In some cases, the suggested settings simply don’t exist on the hardware. So, although the AI may have access to manuals or technical data, it still can’t act precisely or autonomously. That said, the general sonic direction it offers can be quite useful – from there, I take over and personalize the patches manually.

For me, this process of taking, adapting, and personalizing both image and sound material is deeply engaging. It also serves as a research foundation for our ElektroFlux “Kunstkopf” Pavilion / Embassy, which explores AI and experimental electronic music.

In German 

Beschreibung meiner kollaborativen Arbeit mit KI für „Video Edition ArtIn - The Wrong Biennale“: 

Ich habe an einem Experimentalvideo mit eigens komponiertem Soundtrack gearbeitet. Es sollte von den Bildern her möglichst abstrakt bist ungegenständlich sein, was aktuellen KI Anwendungen (Stand März / April / Mai 2025) noch Probleme bereitet. 

Ich habe nach meinen eigenen Prompts die jeweiligen Video-Clip-Ergebnisse der KI Sora aufgegriffen und durch gezielte Variationen weiterentwickeln lassen. Dabei habe ich in den Prompts immer deutlich gemacht, welche Bildelemente ich übernehmen wollte und welche nicht erneut auftauchen sollten. Diesen Prozess habe ich mehrfach wiederholt, bis ich eine ausreichende Anzahl an Clips beisammen hatte – inklusive einer Art offenem Ende für den geplanten non-linearen Erzählfluss meines Experimentalvideos.

Der Arbeitsprozess mit der KI war dabei wirklich langwierig, sperrig und stellenweise bizarr. Die Video-KI wirkt in vielen Fällen noch recht stereotyp und begriffsstutzig – viele der generierten Clips waren schlichtweg unbrauchbar, bei anderen funktionierte nur ein bestimmter Bildausschnitt. Um sie nutzen zu können, waren zusätzliche Eingriffe nötig: etwa Tempoveränderungen, Farbanpassungen oder Umkehrungen der Bewegung.

Der Sound hingegen stammt komplett von mir. Für zwei relativ verbreitete Synthesizer habe ich mir jedoch von ChatGPT spezifische Sound-Patches schreiben bzw. konzipieren lassen. Auch hier gilt: Die Ergebnisse der KI sind nicht unterkomplex – aber oft zu vorhersehbar oder technisch unrealistisch. Manche der empfohlenen Einstellungen existieren so gar nicht an den Geräten. Das zeigt: Auch wenn die KI theoretisch Zugriff auf Manuals oder Gerätewissen hat, kann sie nicht präzise autonom handeln. Die generelle klangliche Richtung jedoch ist meist brauchbar – von dort aus entwickle ich die Sounds weiter und personalisiere sie durch händisches Eingreifen.

Gerade dieser Prozess des wiederholten Übernehmens, Anpassens und Personalisierens – sowohl von Bild- als auch Tonmaterial – ist für mich spannend. Er bildet auch eine Forschungsgrundlage für unseren ElektroFlux „Kunstkopf“-Pavillon / Embassy, der sich mit KI und experimenteller elektronischer Musik beschäftigt.

Anja Kreysing 
Münster, Germany