4 ms·

It's been probably 15 years since I took a class on image processing, but I recall that it was very obvious looking at a 2D FFT and other frequency analysis too

It's been probably 15 years since I took a class on image processing, but I recall that it was very obvious looking at a 2D FFT and other frequency analysis tools if an image had been modified, because you'd end up with a lot of discontinuities. Maybe that is not as telling these days because every image is going through multiple rounds of processing. However I personally find it difficult to follow the "look at the shadows!" type of analysis, and all I can think is there's probably a better way to present this with less ambiguity. Maybe the writer is just using the tools they know and doesn't have the background or tools they need.

max51 4y ago
Those technique don't work very well when the edits are done on the RAW files directly. I creates a lot of false positive and negatives. When a picture is shared by the original photographer directly, it's extremely hard to differentiate between fake and real. Just to name a few reasons for it that come to mind: -The original resolution is wayyy bigger than what you see online. When you go from 300Mpx (eg. a panorama) down to 2 - 10 mpx and add compression, it will hide most of the artefacts. -better lenses means they have less distortion and sharpness problem at the edge of the frame. Makes it easier to blend different pictures together seamlessly -raw files editing allows a lot more modification to the colors without degradation or leaving traces. You are effectively changing the choices that the camera makes when it creates the colors. There is no difference between changing the color temperature in the camera and in photoshop if you use RAW. With jpg editing done by a third party... yeah, those are easy to spot with the right tools