Computational photography (som artikeln pratar delvis om fanns i olympuskameror redan för mer än 10 års sedan. Finns i alla kameror idag mer eller mindre. Ska vi förbjuda kameror eller bara få fotografera i M-läge med alla spakar sätta till standardvärdet? Dessutom tycker jag artikeln är väldigt dåligt skriven ur teknisk synvinkel helt utan förklaringar. AI används som ett generellt begrepp för allt som en modern kamera eller mobilkamera kan göra helt utan särskiljning. Helt klart för att locka läsare och inte för att förklara vad som menas när "kameran använder AI".
Vi får alla fota precis hur vi vill. Jag kommer dock aldrig att fota med en telefon.
Lite qoutes från artikeln
- Your phone goes far beyond collecting the light that hits your camera's sensors. It's guessing what the image would look like if the camera was better and then building it for you,
- When you click capture on a phone, you don't just capture one image, you actually take anything between four to 10 images usually, in normal lighting," Attar says. Your phone blends these images together for a picture that's supposed to be better than an individual snapshot.
- iPhones, for example, employ a feature called
Deep Fusion, using AI trained on millions of images. In addition to handling many of the techniques described above, these neural networks can identify objects in a picture and process them differently, changing individual pixels based on other pictures the AI has seen before. "It's very high-level segmentation,"
- Phones are working so hard to eliminate problems they can even introduce
bizarre distortions that
look like AI hallucinations if you zoom in on the fine details.
-your phone is often making creative or even artistic decisions about the memories you're capturing. Users may have no idea it's happening – and on some phones, AI is doing a lot more than tweaking parameters.
