Obsessed with technology?
Subscribe to the latest tech news as well as exciting promotions from us and our partners!
By subscribing, you indicate that you have read & understood the SPH's Privacy Policy and PDPA Statement.
News
News Categories

Samsung’s latest QLED 8K TVs use deep learning to improve 8K upscaling quality

By Ng Chong Seng - on 8 Jan 2020, 2:01pm

Samsung’s latest QLED 8K TVs use deep learning to improve 8K upscaling quality

First published on Jan 6, 2020 (SGT).

Samsung Q950TS QLED 8K TV. (Image: Samsung.)

A dramatic, bezel-free design for the Q950TS aside, Samsung’s 2020 QLED 8K TVs are better than their 2019 predecessors in several areas. For one, an improved Direct Full Array tech (for full-array local dimming) has increased the brightness of highlights by as much as 20% without using more power. This is possible because each zone’s current can now be controlled. Peak brightness remains at 4,000 nits, with full-fill at 500 nits.

The Quantum Processor 8K also gets an upgrade and now has an artificial neutral network that does deep learning to restore fine details in complex images during upscaling. This deep learning process, which doesn’t rely on any preloaded classifications because it can self-learn and create its own algorithms, will work with the existing machine learning tech that looks out for the other picture elements during upscaling, including texture detail, noise reduction and edge definition. Samsung stresses that its 8K AI upscaling is done pixel by pixel, rather than globally across the entire scene.

Also new is an Adaptive Picture feature that automatically re-adjusts the picture’s brightness and shadow levels based on the brightness of the room. Also available on the new QLED 4K TVs, there are four brightness and shadows settings to choose from. Other nitty-gritty details that probably only enthusiasts will care about include better motion error correction, 5:5 frame repeat support, reduced stutter and per-frame active tone mapping for HDR.

There are many other features that I will talk about in other articles, but briefly, the more obvious user-facing ones enabled by Quantum.AI (that's the umbrella term Samsung uses to group all its QLED TVs' AI features under) include a refined Smart Hub UI with a new Universal Guide that recommends OTT content based on your taste, Object Tracking Sound that creates surround sound using the TV's onboard speakers, Multi View multi-screen feature that lets you put your phone's screen on the TV and adjust the size and volume of each screen separately and Tap View for one-tap screen mirroring.

Also coming to this year's QLED TVs is support for multiple voice assistants, including Samsung Bixby, Amazon Alexa and Google Assistant. There's also a new Digital Butler app that lets you see and control all connected devices (including Wi-Fi, Bluetooth and IR devices) and the ability to send photos from the phone to the TV gallery with only a few taps.

Also introduced today are a couple of features which benefits may not be immediately apparent. For example, AI ScaleNet is a technique that enables 8K video transmissions to be streamed on less-than-ideal network conditions (in short: it compresses content for transmission and unpacks it when it reaches the TV). Samsung is working with Amazon Prime Video on this feature. AV1 codec support is built-in now, too. Suffice to say, Samsung hopes 8K streaming will take off soon.

In the gaming department, the TVs continue to support automatic Game Mode switching, AMD FreeSync/VRR for tear-free gaming, Dynamic Black Equaliser for improved visibility in dark scenes, Game Motion Plus motion interpolation and AI-powered surround sound. Input lag performance gets a good boost, too — dropping from the current 15.4ms down to 10ms (less than 1 frame at 50/60Hz).

Availability and pricing details for Samsung’s 2020 QLED TVs haven’t been announced yet — I’ll update this article when I hear more.