How does 4K conversion actually work?
The intricate process of bringing old videos to modern display standards.
1The difference between 4K and 1080p
4K resolution offers 3,840 x 2,160 pixels, while 1080p provides only 1,920 x 1,080 pixels. Standard definition videos deliver 480 or 576 vertical lines. Cinematic 4K, called 'true 4K', reaches 4,096 x 2,160 resolution. These differences explain why the upscaling process is so complex.
2The core challenge of upscaling
Transitioning from 1080p to 4K requires creating three additional pixels for every original pixel, totaling four times more pixels. Performing this without quality loss is nearly impossible. Simply changing the resolution doesn't reveal hidden details in the video.
3Traditional upscaling methods
Interpolation estimates how pixels should appear when enlarging each frame. These estimations can make the image appear softer. Techniques like edge sharpening and noise reduction may be used, but they can produce unnatural results. Noise reduction, in particular, can make the image look artificial.
4The AI upscaling revolution
AI models are trained on examples of low and high-resolution images. Instead of guessing pixels, they predict what a sharper image should look like. However, AI upscaling isn't perfect and can sometimes create artifacts not seen in interpolation. Still, AI technology is rapidly advancing.
5A professional restoration case
The 1080p upgrade of Star Trek: The Next Generation used original 35mm film negatives with quality equivalent to 20 megapixels. The project utilized 25,000 original film reels, took over three years, and cost more than $12 million. This demonstrates how laborious upscaling can be.
6The importance of source material
The quality of original source files directly affects upscaling results. As shown in the Star Trek example, analog film negatives contain far more detail than digital masters. With low-resolution sources, all created pixels are guesses, potentially leading to unrealistic outcomes.
7Options for home users
Home users without professional resources can choose between traditional interpolation and AI-based tools. While interpolation analyzes pixel color and brightness, AI models make predictions based on training data. TVs' built-in upscaling features also offer varying success rates.
8The future of upscaling
Continuous improvement of AI models will enhance upscaling quality. However, no method can perfectly replicate original high-resolution recordings. Especially for old and interlaced videos, upscaling requires specialized techniques. These technologies are expected to become more accessible and effective in the near future.
Original article: Engadget — 4K'ya dönüştürme nasıl çalışır?