Hacker News Reader

Comment by astrange | Hacker News Reader

There are pretty good ways to detect photo editing, using things like measuring levels of JPEG artifacts across the image. Or if you save pictures raw I’ve never seen anyone try to fake a camera raw file.

View on HN

Couldn't you just "rebalance" the JPEG artifacts as a final pass? Or some sort of slight blur that re-jumbles the least significant bits of every pixel?

Yes, this is standard in the forgery business - dust and speckle filter, drop the res, ensure the light is balanced, and maybe blur if you can get away with it.

These general techniques are weak against specific detection models that understand how the original image should’ve been generated. Another example is camera noise modeling since the injected noise won’t be consistent with how a phone camera etc actually behaves, or an ML model for image depth might see that it's at the wrong distance somehow.

Scaling the image down seems like one of the best tricks if you can get away with it. Another cheap one would be shifting the image 4 pixels over to break up the DCT artifact grid.

> the injected noise won’t be consistent

With a good forgery it will be. Even then, the fact that it's noise means that you won't ever be sure about this.

Why would it be harder or different to fake a raw file?

I didn’t say it would be hard, just that I haven’t seen one. But they contain a lot of information that can’t be displayed on a computer monitor and so it would be a lot of work to check all of it.

Reply on news.ycombinator.com