Recently, there is a big event in the film industry - the 4K restored version of the Italian classic film "The Legend of 1900" will be released in mainland China on November 15th.
Yes, that's right. The classic film that was released in 1998 is now returning with a new and stunning look after 20 years of 4K restoration!
The film was released in 1998 and immediately became popular worldwide, winning 22 global awards including the Golden Globe Award in the United States. It has a high rating of 9.2 on Douban, ranking 15th on the Douban Top 250 list, with over 920,000 ratings, including over 600,000 five-star ratings. The melodious piano music and the touching story of the sea have touched the hearts of every audience.
Challenges in Film Preservation
Film is not only entertaining but also a crucial cultural heritage that deserves protection. However, the preservation process is not that simple.
Digital film has only been around for 10 years. Before that, most films were made on celluloid, which has a preservation period of only 100 years. In addition, many old films from decades ago have faced challenges such as dust accumulation, mold growth, and improper storage over half a century, making their preservation a huge challenge. Therefore, the restoration of old films needs to be prioritized.
Many organizations, including the European Commission, have emphasized the challenges and importance of preserving old film cultural heritage. The European Commission pointed out that without suitable playback devices, people may not be able to watch old films.
The first challenge in restoring film is eliminating distortion. Classic films are usually shot in low resolution, and the original film tapes are significantly aged, filled with noise and cracks. Furthermore, the process of converting film to digital format often damages the film and degrades the image quality.
However, in the past, restoration mainly relied on manual and traditional techniques, which were costly and time-consuming. As a result, only a few classic films had the opportunity to be restored, and the commercial returns were often unable to cover the huge costs. Therefore, restoring old films often remained a matter of sentiment.
But with the continuous development of artificial intelligence (AI) technologies such as machine learning and image enhancement, the restoration of old films has found new possibilities. AI has greatly improved efficiency, reduced costs, and saved manpower, resulting in hundreds or even thousands of times of change.
During this year's Shanghai International Film Festival, the 4K restored version of "The Story of Sanmao" returned to the big screen, triggering a wave of nostalgia. This film is the first domestically produced feature film released after the establishment of New China, and its revival is mainly attributed to AI-based digital high-definition restoration technology.
In the recent National Day holiday, the main theme film "The Final Moment" included colorized 4K footage of the founding ceremony. This is the first time that the historical footage of the founding ceremony has appeared on the big screen in color and ultra-high definition after 70 years.
Due to the poor quality of the film after 70 years, it was a challenge to restore it to 4K resolution. However, with the help of 4K restoration technology, the film was rejuvenated, ensuring the reproduction of these precious original films on the big screen.
Challenges of 4K Restoration: Labor-intensive and Time-consuming
In recent years, a batch of old Hong Kong films, such as "A Chinese Odyssey," "A Better Tomorrow," and "Days of Being Wild," have been re-released in 4K restored versions, igniting waves of nostalgia.
For most people, film restoration work is still a mysterious field. In fact, behind the 4K restored version, there is a huge cost in terms of labor and time.
Currently, most digital films in China are 2K, with a resolution of 2048×1080. 4K films refer to films with a resolution of 4096×2160, which is the highest resolution for digital films.
A true "4K" digital film is shot with a "4K" digital camera, produced with a "4K" digital intermediate process, and distributed and screened with a "4K" digital projector. However, some "4K" films are shot on 35mm film and then converted to "4K" digital format.
The first step in "4K" restoration is to digitize the film by scanning it into a sequence of frame images with a resolution of 4096×3112. Then, the film is restored through the "4K" digital intermediate process.
The image file size of a "4K" restored film is about 6TB, and many steps in the restoration process require real-time processing. Therefore, advanced image processing equipment is needed to meet the restoration requirements.
A person working in film restoration said that restoring old films is a complex process. Over time, the film gradually deteriorates, becomes moldy, and various dirt spots appear on the film. Restorers need to manually repair each frame with dirt spots, which is a repetitive and tedious process. This requires the restorer's patience and understanding of the film material, content, and color.
A restoration team requires at least 5 people, and each film needs to go through rough restoration, fine restoration, and color correction. The most critical part is the fine restoration, where each film is divided into segments of about 20 minutes, and each person is responsible for one segment. Restorers manually repair the video images frame by frame. For example, for a 90-minute film, there are approximately 120,000 to 140,000 frames. Under normal difficulty, each person can only restore 4-5 minutes per day. If the film is very difficult to restore, the team will have a "consultation" to propose comprehensive restoration ideas. Sometimes, one person can only restore 30 seconds per day.
AI Empowers Restoration: Improving Quality and Speed
The cost of manual restoration is enormous. In recent years, in the field of film restoration, AI technology has emerged from scratch. Before AI starts "working," it needs to undergo a lot of learning to build models and inference algorithms. The learning materials mainly come from two aspects: comparing the images before and after manual restoration and artificially simulating old images by adding blur and noise. With these basic materials, super-resolution, denoising and sharpening, stain removal, and color enhancement have become the "basic skills" of AI-assisted old film restoration. The development of AI has brought new opportunities to old film restoration.
In the case of "The Legend of 1900," advanced image super-resolution technology was used, significantly improving the resolution and visual effects. The worn-out video images were restored to their original state when the film was shot, presenting the details vividly with clear, delicate, and pleasing light and shadow. How does this image super-resolution technology work, and how does AI enhance video resolution?
Image Super-Resolution Technology
Super-resolution is a technique that improves the resolution of an original image through hardware or software methods. The process of obtaining a high-resolution image from a series of low-resolution images is called super-resolution reconstruction. The core idea of super-resolution reconstruction is to exchange time bandwidth (obtaining a sequence of images of the same scene) for spatial resolution, realizing the transformation from temporal resolution to spatial resolution.
In simple terms, super-resolution transforms the image from the left side of the figure below to the right side. It provides more pixels and makes the image more detailed. From the perspective of signals, it supplements more high-frequency components in the original signal.
In addition to restoring and reconstructing video images of traditional film and television programs, this technology also has extensive applications in medical image analysis, biometric recognition, video surveillance, and security. For example, the police can use super-resolution technology to identify criminals in surveillance videos, and doctors can obtain clearer lesion images through super-resolution technology.
Multiple Video Websites Unearth "New Treasures"
In addition to traditional film restoration institutions exploring AI-based film restoration technology, some internet companies, such as iQiyi, Youku, and Baidu, have also joined the restoration of old films, becoming new forces in restoring old films with their technological advantages.
In 2006, it took dozens or even hundreds of days for a single studio to restore a 2-hour film. Nowadays, with Youku's heterogeneous computing platform, it only takes 1 hour to restore a classic film to the 4K level. Youku has also launched the "Classic Film Restoration Program" and opened the "HD Classics" section, planning to release over 1,000 high-quality films and classic TV series from home and abroad.
iQiyi has launched the "Global Classic Copy Restoration Program" and the "Classic TV Drama Digital Restoration Project," planning to restore and release hundreds of classic Chinese dramas and dozens of classic films. The iQiyi team has also developed ZoomAI technology, which can be used for old film restoration and 3D animation frame interpolation. Currently, it only takes 12 hours to restore and enhance a 2-hour film using ZoomAI, while manual restoration previously required a team of 10 people working for nearly 20 days.
In addition, MTI, the world's most famous film restoration software, has also introduced AI-based restoration features, such as automatic intelligent stabilization, automatic intelligent global and local anti-flicker, automatic intelligent scratch removal, and automatic intelligent dirt removal.
Film restoration itself has a significant cultural value. Old films are not only important footnotes in Chinese film history but also an indispensable part of world film history and precious cultural assets.
In the current Chinese film market, there seems to be an increasing focus on film restoration. Compared to other methods, using AI technology to restore films can save time and effort. It also strives to improve the success rate of cultural heritage protection, making old films more relevant and creating new commercial value and business models. In fact, the future market for film restoration has unlimited potential but is also facing a difficult beginning.
In conclusion, film restoration is a daunting task, but it is a commendable endeavor for both the industry and the market. It requires more people to recognize and join this cause.
Alphabet Q1 paid subscribers reached 350M, fueled by YouTube & Google One. YouTube ad revenue slightly missed at $9.88B amid Premium growth. Total Q1 revenue hit $109.9B, beating forecasts.
Anthropic releases Claude connectors for Adobe, Blender, and Ableton, enabling AI workflow automation and app control. It also sponsors the Blender Development Foundation.
Apple introduces an App Store model splitting 12-month plans into monthly installments. On iOS 26.4+, it lowers upfront costs without reducing total price, adding billing tracking.
Google launches Aadhaar digital IDs in Wallet for India, enabling secure QR/NFC verification. A phased partner rollout targets key sectors, while passport-based credentials expand globally.
YouTube tests "Ask YouTube," an AI tool blending text and video for complex queries. Initially for US Premium users, it supports interactive follow-ups, expanding Google's AI search.
Google grants US DoD classified AI access, joining OpenAI and xAI in defense deals amid Anthropic's dispute. Contractual safeguards against surveillance and lethal weapons remain unenforceable.
Mistral AI launches Workflows, a Temporal-powered orchestration platform for secure, Python-driven enterprise AI automation featuring human-in-the-loop approvals.