YouTuber Denis Shiryaev has enhanced and restored original footage of a trip in New York, in 1911. For 8 minutes, you travel back in time to see how the New York City looked like for almost 110 years ago!
Travel back in time: New York City, 1911, in 4K
In 1911, the Swedish company Svenska Biografteatern captured the streets of New York City during a trip to America. The original footage in black-and-white is a historic gem that already had viral success. To make this unique footage even more interesting, YouTuber Denis Shiryaev has made it his specialty to upgrade and colorize old videos to bring them back to life.
New York in 1911
The population in New York City was 4.7 million people in 1910 (8.6 million in 2017). The Statue of Liberty had been on-site for 25 years (since 1886), Ellis Island was the gateway for immigrants coming to the U.S. (1892 to 1954) and there would still be 18 years before the Wall Street Crash of 1929. In 1911, International Women’s Day was celebrated for the first time In history, Chevrolet officially entered the automobile market to compete with the Ford Model and a big fire at Triangle Shirtwaist Company factory killed 146 workers. The 1910s was a decade of many race riots and the decade in which the U.S. first became a world power.
The team of Swedish cameramen working for the company Svenska Biografteatern was sent around the globe to take pictures and videos of famous, well-known locations. Apart from New York City, the workers traveled to Paris, Venice, Niagara Falls, and other world-famous places.
How to reproduce old videos
Denis Shiryaev is behind a number of upscaled classic short films. He restores and upscales the historic footage by using neural networks, and has also upscaled it to 4K and increased the frame rate to 60 frames per second before finally colorizing it. According to Bored Panda, Denis has spent 4 days working on this piece. His process includes manual stabilization in After Effect, footage ‘clearing’ via 4 different neural networks – first one for clearing from digital artifacts, the second one is ‘DeOldify’ used for colorization, the third one a custom version of DAIN for FPS boosting, and finally the fourth a custom version of ESRGAN trained on his own dataset. The merged result is the final video.