Global EditionASIA 中文双语Français
China Daily Global / 2020-12 / 18 / Page005

AI tech brings color to old films from Beijing

China Daily Global | Updated: 2020-12-18 00:00

People dressed in traditional attire bowing to each other in greeting, traveling on rickshaws and praying at a temple ... what life was like for Beijing residents a century ago was captured in rare black-and-white footage.

Using artificial intelligence technology, Hu Wengu, a 29-year-old artist and game developer, colorized and sharpened the monochromatic and blurred footage, giving modern folk an intriguing glimpse of life in the capital long ago.

"I was born and raised in Beijing. It has been my dream to restore and colorize old black-and-white film about this ancient city," Hu said.

He started using AI technology to develop games late last year. During the process, he developed a deeper understanding of its wide applications, such as generating artificial faces, composing music and colorizing grainy old films.

Film colorization has existed for decades, but the traditional handcrafted process is tedious, slow and expensive. Advances in AI can convert monochromatic images into vibrant color and turn blurred footage into clear video.

It usually takes Hu up to two weeks to restore a five-minute film with the help of several pieces of AI-powered computer software. For example, a coloring AI tool, based on machine learning algorithms, can build an understanding of how objects are usually colored and paint the old colorless films with hues.

With AI technology, low-resolution images in films can also be automatically upscaled into high-resolution video versions.

Although AI technology improves the efficiency of traditional film restoration techniques, technology is not a cure-all for artistic works like movies. Manual efforts are still required when AI fails to produce authentic color or fluid pictures.

For Hu, one of the highlights of film restoration is exploring history with his followers online and allowing more people to understand the value of old clips.

In a piece of black-and-white footage of Beijing life in 1929, a barber shaving a man's head wrapped a white towel around the man's neck and gave him a tray to catch the falling hair. Minutes later, the customer patted his shaved head with a big smile, exclaiming,"That's good!"

If not for Hu, the clips would still sleep silently in a university in the United States. Now the restored footage has drawn millions of views on social media platforms, and people can see the vibrant hues and vivid looks on the faces of their forebears.

"The barber in the film said he came from Baodi. I had never heard of the place. Then one of my followers told me that it is a district of Tianjin, near Beijing," Hu said. "It is interesting to explore history with so many people."

In August, he uploaded a colorized documentary about a Shanghai fashion show in 1929. A descendant of one of the models in the film contacted him and shared the story of that era with him.

"Some overseas Chinese in Malaysia and Singapore also contacted me, telling me that my restored clips reminded them of their childhood. It gave me a sense of accomplishment," Hu said.

With AI technology, Hu has restored many pieces of historical footage filmed in cities including Beijing, Jinan, Shandong province, and Shanghai. He plans to give face-lifts to more old, grainy films of different Chinese cities.

He also hopes programmers can improve AI technology to make it trace colors more accurately and authentically. For Hu, the value of restoring old films lies in creating an opportunity for modern Chinese to "travel back to the past" and communicate with their ancestors through a screen.

Most Viewed

Top
BACK TO THE TOP
English
Copyright 1995 - . All rights reserved. The content (including but not limited to text, photo, multimedia information, etc) published in this site belongs to China Daily Information Co (CDIC). Without written authorization from CDIC, such content shall not be republished or used in any form. Note: Browsers with 1024*768 or higher resolution are suggested for this site.
License for publishing multimedia online 0108263

Registration Number: 130349
FOLLOW US