My images are credited “© Gerald Kapounek”. This page explains what that credit stands for: how I use AI in my work – and where I draw the line.
How much AI goes into my images?
Kept to a minimum. Every person you see in my images genuinely stood in front of my camera. I use AI solely as a tool in post-production – it streamlines the process without altering what the image conveys.
Levels of AI use
Whether AI is involved is rarely a simple yes or no. The five levels below make the distinctions clear – and show where I stand on each.
- Level 0In use
-
No AI – Captured in camera and edited by conventional means: light, colour, contrast and cropping.
- Level 1In use
-
AI-assisted editing – Captured in camera, with AI-based tools supporting the edit – for noise reduction, masking or retouching. The substance of the image remains unchanged.
- Level 2Rarely
-
AI compositing – The subject is captured in camera; the background or setting is AI-generated.
At most, I use AI to complete parts of a background – I never generate an entire background.
- Level 3Not used
-
Predominantly AI – A real photograph serves merely as a starting point.
- Level 4Not used
-
Fully AI-generated – Created entirely without a camera.
These levels are a voluntary framework, not a legal category. Both the model and the idea behind this page originate with Martin Dörsch: a /C page that is as standard as an imprint. He explains the concept here (in German).
The people in my images
You should recognise yourself in your images. That is why I handle all retouching myself – and stop precisely where an image would start to stray from who you are, with or without AI tools.
Your images and AI
You may use your images for the agreed purpose, with appropriate credit. Any alteration requires my prior written consent – and that includes AI edits such as replacing the background, extending the frame or “enhancing” a face. This is set out in my terms and conditions. If you have something like this in mind, get in touch first.
Any questions about my use of AI?
Last updated: November 2026. Should my approach change, this page will be updated accordingly.
