Impressions art series

With the Impressions art series, I break a little with my usual abstractions. This is an art series about how we collectively treat and portray people, the noise and fall-out of attention, whether sought after or not. The lurid pictures of a frozen moment, the gossip, the accusations…
Once in the spotlights, the protagonist, or victim of exposure, has to be constantly on guard, often has to live with repeated invasions of privacy and public attacks, which often shake them to the core of their being. Think about the #meToo accusers and the accused, what they have/had to go through in the public eye. The resulting scars may disappear with time or lead to destructive behaviour, depending how well they cope with stress.

Let’s face it; we all have this a little in us, devouring the rich and (in)famous in press and social media. However, this series is not about defending them, it’s about how we collectively tend to be remarkably judgemental, and in times of upheaval as we are today, we lose so easily our capacity for empathy and tolerance.

The artworks are, against my usual style, quite emotional displays, although without showing a clearly discernible context. This allows the spectator to let their imagination run wild. There are on the whole no intentional clues (although I sometimes add elements that seem to suggest a connection), only vague impressions, atmosphere and compositions that feel recognisable, although they are imaginary. What’s behind this scene/image? And what do you feel about it?

The artworks are relatively big, namely 5760 x 4224 pixels, in JPEG format.
The first 5 artworks are minted on the Solana blockchain ( since Friday, 19 January 2024: series


Any resemblance to existing people is pure accidental, no malice intended or hidden agendas. I did not use names of real people in any of the text prompts. In fact, I made sure to generate anonymous, artificial images for this project as not to jump on the bandwagon of celebrities.

Medium & technique

I used StableDiffusion with a fixed seed image I prepared upfront to create a big series of images. I tested with various AI models, like deliberate_v2, lyriel_v16, abstractor-yiu_v10 and my own dataset (which was useless in this case). Then I curated the useful images and reused those in turn as seed images, with the same text prompt, but another AI model which produces more realistic images, namely cyberrealistic_v32, although the intentional noise and distortions from the first generative images remain for most part.

Seed images for the second round.

Seed images for the second round.

After that, I used Affinity Photo to add selectively a rough dot matrix, so typical for newspaper print, and finally applied tonal corrections where I saw fit. Quite a few artworks are collages of various generative images. Here and there I retouched an eye or removed disturbing details, although I left a lot of artefacts in order to;
A: give the artwork a touch of grunge and rawness.
B: clearly show this is artificial intelligence I worked with.

100% detail of Impressions #2

100% zoomed in detail of Impressions #2

To create this series, I had to generate 1516 images and did a real life photo shoot of mobile devices to feed to the AI.


Leave a Reply

Your email address will not be published. Required fields are marked *