An AI machine-learning program was recently used to recreate book-accurate Harry Potter characters based on their written descriptions.
An artificial intelligence machine-learning program was recently used to recreate book-accurate Harry Potter characters based on their written descriptions. Released in 1997, J. K. Rowling's Harry Potter and the Sorcerer's Stone spawned six sequel books, eight main films, three spinoff films with two more planned, and a stage production of a sequel story, all in a multi-billion dollar multimedia franchise. The series has proved to one of the most popular in history with over 500 million copies sold and each book translated into 80 languages.
As with many book-to-film adaptations, characters from the original story sometimes do not look as they were described in the original source material. This is of course due to the fact that actors are cast more on just their looks, but also for their ability to capture the essence of a character and their skill as an actor. In the original book, Harry Potter was described as having green eyes, but in the films he had blue eyes due to Daniel Radcliffe being cast and a reported allergy to colored contacts. Hermione Granger was said to have large bushy hair and big front teeth. Although Emma Watson's portrayal did have slightly frizzy hair, it was nothing like her character's original description and nor did she have large teeth.
Related: Fantastic Beasts' Failures Makes A Cursed Child Movie More Likely
On Reddit, user lettuceown shared renderings depicting prominent Harry Potter characters created using an AI machine learning-based art program called Artbreeder. Using the descriptions of the characters from the novels, this was an attempt to create the most book-accurate versions of the characters. Check them out in the space below:
Click to see original post
Click to see original post
Click to see original post
The Harry Potter movies have been ingrained in popular culture for more than 20 years now, so it's interesting to look at these renderings and compare them to their live-action counterparts. Some are notably different; the Ron image bears very little resemblance to Rupert Grint, and Neville Longbottom has blonde hair as opposed to Matthew Lewis' darker color. Conversely, there are images that closely match the movie versions. Snape, Voldemort, and others fall into this category, illustrating the filmmakers tried to stick to the source material when possible. The Voldemort rendering in particular arguably could pass as concept art for Ralph Fiennes as the iconic Harry Potter villain.
AI machine-based learning programs, such as Artbreeder, gives artists the ability to imagine characters through the use of physical attribute sliders, similar to that of character creation tools found in many role-playing video games. Through the use of such programs, characters in books, that perhaps have yet to be adapted in any other visual media, can finally become realized to a degree of accuracy. Although characters like Ron Weasley and Hermione were excellently portrayed in the Harry Potter film series by Grint and Watson respectively, these images allow depictions that hold more true to the original well-beloved books.
Next: Perfect Harry Potter Theory Fixes 1 Major Albus Dumbledore Plot Hole
Grew up absorbing every book, movie, TV show I could get my eyes on. Every time my parents took me to the doctor, I would forgo the obligatory post-shot ice cream cone for a two hour vacation in the theater. I have a passion for storytelling and a stomach for pizza. Who am I: Vertigo, Panic Room, Jurassic Park, Lawrence of Arabia.