The Simpsons, Futurama and Family Guy: this is how they look with artificial intelligence and as if they were in the 80s

January 30, 2023

They create videos of what various animated series would look like as if they were 80s sitcoms. Food series have a different magic than those with real actors, but there is always the question of what those characters would be like if they existed in the flesh. The Internet gave the answer. Using artificial intelligence, they developed models of what these iconic productions would look like.

To give it a touch of greater nostalgia and taking advantage of the fact that they are part of the sitcom genre, they asked MidJourneyartificial intelligence that generates images, that creates models of the characters in live action format and as if they were part of a series from the 80s.

The series that have been used are four of the most famous: The Simpsons, Futurama, Family Guy Y Scooby Doo. One of the authors of these videos was the owner of the channel Youtube called Lyrical Realms and told how the creation process was.

“All the images come directly from Midjourney, but they did not come up using only the text indications, it was a combination of the use of already existing images and indications as well,” he said in an interview with PetaPixel.

To achieve the appropriate results, he relied on photoshop in two specific cases: to remove artifacts, such as black or red borders, and to separate the intro into layers and give a 3D effect.

In total, it took him five days to make only the Family Guy video, since artificial intelligence sometimes took too long to generate the images or in some cases had to discard content because it was not complete, such as some characters without glasses or size proportions.

“The engineering part was definitely the hardest part, I had to generate an insane amount of images before it started to see the light of day and I finally managed to generate the type of look I was looking for. Once the first character (Peter) was generated, the rest was a bit easier. The most difficult to generate were Cleveland and Quagmire,” said the creator of this idea.

In total, due to all the problems and delays, to consolidate the video he generated about 1,500 images. Although that does not save the content from having some flaws, such as the hands of the characters and that some vary greatly in their design, as is the case with Brian Griffinthe dog, who changes breed in several photos.

“Basically I used a whole weekend plus two days off from work. Out of those five days, I would say the majority of my day was spent on making the video. I was not always in front of the screen, since sometimes I had to wait a few minutes for all the images to be generated, but I had no other plans for those days other than to create the images and the video, ”he assured.

But in addition to what was done with Family Guyother programs went through this experiment and the results were surprising, both positively with high fidelity adaptation and negatively with changes far removed from reality.

An example is Marge Simpson, because her hair is no longer combed up to have a totally different style, although maintaining the blue color. Instead, she has great successes with the models of Lisa, Mr. Burns and Bart.

another series was futurama, which is one of the best results, the only outstanding failure is that of Leela who has two eyes and is not a cyclops. But the rest of the characters are quite similar and would be an adaptation according to the productions of the 80s.

Finally, Scooby Doo generated curious images, but in general it is noticeable that artificial intelligence had a strong inspiration in the live action movies of the franchise, although with some changes such as the race of Scooby.