At this point, there’s no way you haven’t seen Cruzcampo’s new commercial with Lola Flores. Yes, Lola Flores. The pharaoh. The multi-faceted Spanish artist who has become as famous for his talent on the stage as for speaking openly about taboo subjects.
From now on, the Sevillian beer brand wanted to find the character, the claw, the power, the accent this made her so beloved by the Spaniards for their latest advertising campaign. And, as expected, he gave a lot to say.
But the controversy goes far beyond the choice of character, and has more to do with the technology that was used to enable their presence. The controversial technique of deepfake That’s still news, but which side of the debate should you be on?
Cruzcampo recovers La Faraona
Lola Flores returned to our lives on the occasion of what would have been her 98th birthday with an advertisement for Cruzcampo that quickly became trending topic. The spot in question, which you can see below, shows the singer addressing the camera with a message that invites us to be proud of our roots.
In With a lot of emphasis, the campaign slogan, Cruzcampo “brings back” the singer so that we can resume our accent (with a special nod to Andalusia, where the company and the artist come from), but not only in speaking, but in all aspects of our life.
According to its creators in a recent interview, the announcement was scheduled just before the pandemic, but this quasi-year “as a gift” allowed them to further perfect the technique that was used to achieve it.
What is the deepfake technique?
The technique we are talking about is the call deepfake, a word which results from the junction of the English terms “deep learning“and”false“. It’s about using artificial intelligence to create fake videos with incredible credibility.
In the case of Lola Flores’ ad, for example, it’s hard to believe it’s not about itself. Before knowing how it was done, surely more than one hadn’t thought that maybe it wasn’t the case.
It is true that his use of the word “accountability“It made us suspect that there was something behind it, but maybe not the hard work carried out for weeks by the teams of two companies in Barcelona: the advertising agency Ogilvy and the studio Metropolitana.
For the Cruzcampo spot, they recovered some 5,000 images from the Sevillian artist, which they had to optimize before putting them through the two programs of deepfake that they used to get the final product.
In the process, they got an actress similar to Lola Flores and created a 3D model of her that they would later work with to give the collected images the necessary gestures and movements to achieve such a believable result.
In addition to using the FaceSwap program to transfer the singer’s real face to that of the 3D model, the final touches to make the Lola Flores in the video look like the real Lola Flores in great detail were used Software DeepFaceLab.
Speak more technically, in the videos deepfake, the so-called RGA algorithms are used. In this system of two neural networks, a model is presented with photos or videos from which it creates false images, and another which is responsible for detecting the latter.
The process is repeated as many times as necessary until the second model cannot distinguish the counterfeit images. The more real images you have, the more convincing those created from scratch will be.
Clever, but controversial
Lola Flores’ daughters, Lolita and Rosario Flores, were involved in the production process of the ad. And it’s important to take this into account before you start debating whether this place should have been done or not.
Few but can be put to the assertive and stimulating message of Cruzcampo’s advertising, and no one can deny that the result is spectacular. There is no doubt that this is a triumph of technology and science.
No one better than her daughters to know what Lola Flores would have liked or not, but it is inevitable to wonder to what extent our heirs have the power to consent that to our image and our voice, fake videos of this type are create.
Spanish legislation does not yet consider the specific case of deepfakes, but they can be included in the image rights which, once you die, pass into the hands of your heirs. The same goes for the power to manage its mission for commercial purposes, as in the case of Cruzcampo advertising.
Also, the key to differentiating what is a crime from what is not would be in the reason for its use and the end result. And in the ad, it doesn’t appear that anyone could be offended by the use that was made of the image of Lola Flores, including Lola Flores.
The expert in digital law, Borja Adsuara, explained a few years ago where the problem lay with this artificial intelligence technique: “Unless you are an expert in these technologies, it is very easy for everyone, including a judge in a process in which they provide a test with this technology, they can force it ”.
Thus, the debate goes beyond whether or not it is legal or not, or should be or not, to use images of the deceased to create fake videos. Although we can ask the same if it is about living people. Another big problem lies in what makes this technology so incredible: its verisimilitude.
The fact that you are watching Lola Flores’ video and not questioning that she isn’t really the one in it means that the technique could also be used with similar quality results but with less innocuous intentions.
A few years ago, this Barack Obama video went viral. It is not halfway that we realize that the atrocities he says there, and that it is surprising that someone as politically correct as Obama publicly says, has been made to say by someone. ‘another.
The dangers of deepfakes they have become more evident in their uses in industries such as pornography. In 2017, pornographic videos of actresses such as Daily Ridley or Scarlett Johansson began to circulate, but they turned out to be fake. Soon after, DeepNude was born, an app that uses technology to create nudes from images of people in clothes. It closed a few days after its creation.
“The deepfakes it’s a dangerous way to spread hoaxes. But the rolls they show that advanced technology is not necessary, ”says Panda Security of the threat of this new type of technology.
Anyone Can Create rolls, because you only need one Software (which is also generally cheap and accessible) which creates the fake videos for us. So to create deepfakes more advanced editing knowledge is needed, social networks like TikTok already have filters which cause a lot of problems.
Beyond Lola Flores: the use of deepfake in entertainment
But not everything will be negative, and Lola Flores’ announcement also helps us remember how far we’ve come in areas such as artificial intelligence. We usually associate it with humanoid robots, but it can also be applied in art.
The best-known example is probably that of the films of Star wars. Harrison Ford returned to his youth with this technique, which was used on actor Alden Ehrenreich’s face to help explain Han Solo’s origins.
Something similar was done more recently in the Spanish series on HBO 30 pieces. Instead of having other actors to make the youth versions of the protagonists, the creators decided to use the deepfakes to rejuvenate Eduard Fernández and Manolo Solo.
And it is that all is not what it seems.