This article originally appeared on republicworld.com
By Kaushal Lahad
Satyajit Ray’s 1952 film Pather Panchali is one of the most unique projects and in the time of lockdown this film has got an upgrade as a professor from the University of Maryland made a colour version of the film. It was reported that a professor in his time in quarantine used artificial intelligence to add colour to some scenes from the Satyajit Ray Classic.
It was reported that the 2-minute long video of Pather Panchali was recreated ad uploaded on social media on May 14, 2020. It was reported that a professor by the name Aniket Bera did this as an experiment. Aniket has been a professor at the University of Maryland since 2018 and it was reported that this was done as a part of an experiment. it was also mentioned that a port of professors works with old footages but he chose this film as it is very close to his heart.
Aniket Bera born and brought up in Delhi did his PhD from the University of North Carolina. He uploaded this video on his YouTube and in the caption of the video he wrote “This footage has been automatically upscaled to 60 fps, 4K (Ultra High Definition) and digitally colourized using deep neural networks (DAIN, ESRGAN, DeOldify). This is purely an academic experiment and can be used for old films but the best cinematic version is the 2015 Criterion Restoration by the Academy of Motion Picture Arts and Sciences (Oscars).”
He then described some things about Pather Panchali and wrote “Pather Panchali (Translation: Song of the Little Road) is a 1955 Indian Bengali-language drama film written and directed by Satyajit Ray. It is frequently featured in lists of the greatest films ever made. If you would like to see more of such AI film experiments, please leave a comment below. Please note that the colourization is done using AI based on a huge training dataset, there is no “ground truth” data available from the era. In the future, it would be great to see more such AI restoration of Indian movies”.
https://www.youtube.com/watch?v=WwwzVSNbg9E&feature=emb_title
