Imagery that does not exist yet

Usually not, and that is deliberate. The most reliable use is continuing and adapting real photographs: extending backgrounds, changing formats, removing distractions, matching series. We use freely generated imagery where no photograph is needed and no false impression is created, for instance for mood and background images. Products we show as they are.

With products the results are considerably more reliable, especially for cut-outs, retouching and consistent series for shops with many items. With people the limits are tighter: recognisability, details in hands and clothing as well as personality rights often speak for a real shoot. Where we see limits, we say so beforehand instead of letting them show up in the result.

Through rules rather than prompts. We define perspective, light, colour mood and message beforehand, work with references from your existing material and iterate until the image fits the visual language. Without such rules you get technically clean images that look like catalogue stock. Every image is reviewed before publication – otherwise mistakes in the details are first spotted by the customer.

Yes, if the image carries the necessary resolution and detail. For large-format applications we check this beforehand and, where needed, combine AI with classic production or retouching. For web, shop and social the quality is usually uncritical; there the effort lies more in processing large volumes consistently.