Martin Disley

Habsburg AI Portrait Studies Martin Disley Habsburg AI Portrait Studies is a series of cloth-printed generative portrait images and a short frame-interpolated animation produced using a bespoke diffusion model recursively retrained on its own output. This autophagous training causes the model to collapse on itself, constricting the model’s distribution around the mean, forcing it to produce images in an amplified version of the model’s default style. The images not only demonstrate the potential retrograde effects on model development caused by incorporating synthetic data into training but also reveal the average aesthetic character of the target model, MidjourneyV5, by exaggerating the formal properties to which it characteristically reproduces. The artwork was produced as part of a research project responding to the implications of growth rates in model parameter size outpacing growth in available training data and an investigation into the particular aesthetic character of popular synthetic image generation models. The project produced Aesthetic Distillation, a software application for recursive training that amplifies the fundamental aesthetic characteristics of a target model, making them amenable to formalist analysis. The software was written in Python and utilises the Hugging Face Diffusers library to manipulate and train the model. As a proprietary platform, the Midjourney model is not accessible to researchers. Therefore, the base model used in this work was an open-source clone of Midjourney. As an exploration of the aesthetic properties of this collapsed version of Midjourney v5 model, the images in the Habsburg AI Portrait Studies series took a limited subject and prompting approach: investigating the models' representations of human subjects through prompting for "portraiture of men and women". Given the degenerate development of the model and the grotesque portraiture, it has a tendency to reproduce, the series takes its name from the medieval Central European dynasty with a famed legacy of inbreeding. To create the final works, a selection of image was curated from hundereds of candidate images. These were then refined through inpainting with MagicQuill (CVPR 2025) and traditional photo editing techniques in addition to upscaling using ESRGAN (ECCV 2018).  The physically installed works in the series (Habsburg Portrait Studies 001 & 002, 2024) were created by printing these images on banners in custom frames that give them the impression of slowly collapsing under their own weight. The interpolated animation shown at CVPR 2025 was created by running a selection of these images through FILM (ECCV 2022) using a custom script to modulate frequency.