If anyone is still skeptical about the potential of artificial intelligence, these photos may change their minds.
If you look at these portraits, you won't believe they're all fake. No, they're not Photoshopped, but graphic renderings generated entirely by artificial intelligence.
Everyone already knows that it is possible to create realistic faces and characters using computer graphics (or CGI, as you prefer): just look at the excellent work done by the developers of PES 2019 or the flood of video games that have come out on the market in the last year.
This situation, however, has a completely original aspect.
NVIDIA researchers have published a new paper in which they report an interesting result: their generative adversarial networks ( GANs ) have succeeded in creating realistic face models from pre-trained data.
The study was started in 2014 and it is interesting to see the enormous results achieved in less than 5 years.
The creation of these faces wasn't left entirely to the AI. NVIDIA researchers had to train the AI for a whole week using eight graphics cards that managed to create these exciting results.
The ability to create new styles from a pre-established template allowed the system to make changes to the basic template and expand the possibilities to allow for choice and preference on the final product.
The use of this technology doesn't stop at creating faces and anthropomorphic figures, but can expand to the creation of additional models for objects and environments. This is a revolution in the world of photography, as it will be possible to create environments from scratch and bring a new artistic perspective, expanding the boundaries of creation.
Just think about the fact that since 2014, IKEA has only used computer-generated environments for its furniture catalogs. The introduction of artificial intelligence will allow for a very powerful tool to reduce costs and increase yield and impact.
We mustn't fall into the usual trap: AI won't steal our jobs, but rather enhance them. We'll probably have to go back to the books and spend more time in school because we'll need to delve deeper into the programming structure for a one-to-one human/machine dialogue.
On the one hand, planning, on the other, the provision of a unique service that will allow us to make a leap forward.