Spread the love

The top ’90s animated series shaped a whole generation and aired between 1990 and 1999. These cartoons were frequently aimed at children but some of them (shows like Family Guy, South Park, etc.) also were made for an older generation. With that being said, those of you who were born in the 1990s undoubtedly spent many hours as children watching shows like Doug Funnie, Rugrats, and The Simpsons.

An intriguing project named “Saturday school” was produced by artist Daniel Voshart using Artbreeder and Photoshop. He made fifty fake school photos showing popular cartoon characters who aged like they were at the end of high school/university in the ’90s.

The artist also told Bored Panda quite a few notable things regarding the project, “I know there has been a movement against AI art recently due to the non-consensual datasets being used. “Saturday School” was made before the rise of Midjouney, DALL-E Stable Diffusion, etc. No typing in prompts. Each face took about a day to make. This project was made possible with Artbreeder and their Flickr Faces HD Dataset where ‘Only images under permissive licenses were collected’. As well as Photoshop.”

More info: voshart.medium.com | voshart.com | Etsy

#1 Ashley Spinelli From Recess

Image credits: voshart

We also asked the author of the series to explain to us how he came up with the idea for “Saturday school”.

“I am far from the first to imagine cartoon characters ‘as real people’ but when a coworker told me Spongebob was Black, it seemed like a curious and fun challenge. That was the start of things.

When I decided to do a series of portraits, I thought about old-school graduation posters and how fun it would be to have them all attend the same imaginary school.”

#2 Misty Waterflower From Pokemon

Image credits: voshart

#3 Jane Lane From Daria

Image credits: voshart

We were also wondering what sort of tools were required to create these portraits.

“The tools I used were Photoshop and Artbreeder.

Artbreeder is a tool that allows you to make modifications to portraits (gender, age, ethnicity, etc.). If it recognizes the face, you can then make changes (the technical term is move through latent space). One needs to first prepare the flat cartoon to be realistic enough to be recognized by the software. Attached is a GIF process of how I made Superintendent Chalmers from the Simpsons. He didn’t make the cut, though!

Sometimes it was a celebrity as a starting point. Arthur the Aardvark is based on a young John Legend.”

#4 Daria Morgendorffer From Daria

Image credits: voshart

#5 Marge Bouvier From Simpsons

Image credits: voshart

Daniel also shared with us how long it takes for him to make one portrait from this series. “Each character took about a day but it typically ranged between 4-12 hours. Spongebob was an outlier and took many days with several disastrous attempts.”

To choose the characters he wanted to work on he told us that he actually made a list. “My partner and I made a really long list of the most memorable characters from cartoons we could remember. Non-Canadians might not appreciate the references to Reboot but I would be betraying my childhood if I changed them for more obvious fan favourites.”

#6 Ned Flanders From Simpsons

Image credits: voshart

#7 Philip Fry From Futurama

Image credits: voshart

AI art recently had taken over the internet, with people being both opposing and supportive of it, so we wanted to know how the author of the project felt about that.

“I think generative text-to-image AI is going to be a great tool for artists but we are at a weird point in history where, in the rush to be first, developers used datasets without consideration of the artist. Maybe they didn’t realise… or think that artists wouldn’t notice. I used them briefly until I had a Soylent Green moment.

I’ve noticed the biggest advocates of unrestricted text-to-image AI are non-artists. They haven’t developed an appreciation for the community that they base their current work on.

There is also the legal aspect: I can’t trust an AI’s output to make something transformative or original. It would be great if it could provide citations for the final output but it can’t.”