What Artificial Intelligence can do depends a lot on the capacity of the algorithms that are carried out in the calculation flows in order to achieve curious results. As with the new Microsoft Artificial Intelligence that get to make face caricatures.
If you could pass him a photo of your face so that he can "draw" a caricature close to the most characteristic features of your face. Actually what Microsoft has done is create a cartoon drawing neural network.
The network is called CariGANs and is made up of a technology called CariGeoGAN. This is able to determine the geometry of a face of a photograph so that I can map it and thus create a cartoon model.
Finally, CariStyGAN, will be the "manager" to give that artistic aspect to the mapping of our face to result in a whole cartoon. Come on, those who make a living from making cartoons can now work harder because Microsoft's Artificial Intelligence is going to make it very difficult for them.
In order to achieve these given results, Microsoft has had to train the system with thousands of hand drawings. The funny thing about this system is that they are capable, 22,95% of the time, in tricking the user into believing that this cartoon was actually made by a cartoonist and not a machine.
And the thing has not stopped here either, since Microsoft wants to go further with this Artificial Intelligence, and be able to analyze videos so that we can create caricatures of ourselves in a video capture.
A technology that will allow you to "draw" caricatures from photographs and at the same time carry out the reverse process, that is, look at a cartoon and "put it back" so that we can see the person behind it.
Don't miss our tutorial for lose weight in Photoshop.