I see in the news that Germany has made it illegal to morph photos used in passports.
For me, this is one time where the law is ahead of technology.
Apparently morphing a photo allows one to make subtle changes to the photo to fool facial recognition systems. Allowing another person to use the passport. According to the article, if the two people look similar the human eye can’t tell the difference in the morphed photo.
I am confused. What technology is being used and what does it do?
Can’t see why you couldn’t do it in Photoshop. Facial recognition works by mapping facial features, such as nose shape and position on face, so it would be fairly straightforward to tweak that in photoshop whist keeping the image still looking recognisable as the real person.
It might be in response to things like Snapchat filters where you can very easily, for example, artificially age or faceswap those in the frame.
It could be done with the ultra-advanced technology of the mid-1990s.
The thought certainly occurred to some of us, in those advanced future years of the mid-1990s, of maybe digitally enhancing the passport photos a bit before sending off the application, and, who knows, maybe it would have been fine, but ultimately it did not seem like a kosher thing to do.
Even a little earlier, viz. the famous “Black or White” music video, which dates back to 1991.
From what I understand, the idea is that you take photos of two people, morph them into one, and get a hybrid photo which is “halfway between the two” and looks sufficiently similar to both people in the original photos to fool face recognition systems at automatic border checkpoints. Such systems take pictures of the people arriving at the checkpoint, map their facial features against those in the passport photo, and if they match will accept the person as the legitimate holder of the passport. That is something which off-the-shelf graphics editing software should be able to do.
I suppose the difference is that today complex face morphing is easy and completely automatic. You don’t need to be a brilliant digital artist.
I am aware of morphing one image to another. As others have said, that has been around for a long time. Long enough even for me to hear about it. 
What I am curious is whether something new has been developed. The implication I got was that the passport photo looked like the original person, but it had been morphed just enough to allow another person to successfully use the passport. Perhaps all that is going on here is that the current technique doesn’t morph one image to another, but takes the original image and changes it just enough to fool a given recognition algorithm. Since the image may not be visible altered, the passport hasn’t been tampered with and no crime committed. Hence the new law.
Or something else. I don’t know.
“50/50” averaging is certainly known in this context, and can be done with a single click. In that case, the point is that the image (by definition!) resembles both people, so that the official issuing the passport OKs it as a picture of the first guy, and so does the border official comparing it to the second guy.
Now, if you know more about the computer or manual face-matching being used, I suppose you can fine-tune things using some of the more sophisticated morphing techniques available today. For instance, blend just the eyes and/or nose and/or mouth. But, if this is just submitting a doctored photo rather than a genuine forgery, you still have the basic constraint that the photo should sufficiently resemble 2 people; studies show even straightforward 50/50 morphs fool people at least 2/3 of the time. Newer/better techniques are going to produce morphs with fewer or no tell-tale artifacts even without any manual editing to clean things up.
Example study: Face morphing attacks: Investigating detection with humans and computers | Cognitive Research: Principles and Implications | Full Text
Thanks.
Based on my further reading that is what is going on here. There is a combining of images but the measurements important to facial recognition algorithms are factored in to allow the photograph to look like the person carrying the passport-even if it is a second person.