photokina: 08.–11.05.2019

#photokina

Intelligent functions provide support from recording to issuing

From intelligent, automatic image optimisation to the first cameras that can take pictures independently, innovative new features and devices are bringing a fresh breeze to the imaging scene.

Self-photographing cameras that extract the relevant photos from a video stream with the help of AI are based on the automatic recognition of people, animals and emotions. When these are known, it is only a small step to, for example, filtering out several images of a toddler smiling frontally from a half-hour video recording.

When the images and relevant sub-areas are recognised, they can also be automatically changed with the help of intelligent functions. The spectrum is in this case very broad: it starts with the perfect release and supplementation of missing image areas, as are sometimes required, for example, for retouching. In addition to this, scene-dependent corrections and optimisations can be carried out automatically. One bonus of self-learning systems is that they incorporate the individual preferences of the respective user over time and can adapt their automatic optimisation to these. This is of special interest to professional photographers, who can in this way retain their individual look despite automated processing.

KI Bildoptimierung © Werner Pluta

However, the spectrum of possibilities today extends so far that even automatic montages of several images can be produced, like with group photos, for example, where there is seldom a photo found in which everybody is smiling. AI-supported image processing composes the laughing faces from several photos into a single new one. This even goes so far that strange modifications become possible and, for example, a new hairstyle is created, or a woman is created from a portrait of a man. The results today are astonishingly realistic and can hardly be differentiated by laypeople from an image taken directly with a camera.

The photographer is still needed to take and process relevant photos, because true creativity still remains the domain of human beings. However, approaches can also be detected there indicating that the penetration of algorithms is probable even here. In the project "Inceptionism" from Google, for example, the program "DeepDream" was already capable in 2015 of creating independent images of bizarre beauty, but still as unintended side effects. Since then, development has progressed very rapidly, meaning that it is in the meantime possible to generate photorealist image of a cat from a simple line drawing.

Artificial intelligence, which is becoming increasingly relevant for the entire imaging workflow, from the taking of the photo through processing and archiving to the issue of the image, will be one of the trend themes of photokina in September 2018 in Cologne.