Philips TV P5 processor adds AI: Recently, Philips TV P5 processor adds AI to its new models. Why? The importance of good image processing is difficult to overestimate. Why? We watch more and more different images, from our old DVD collection, via live TV to streaming services and game consoles, or the holiday videos that we shoot ourselves on our smartphone. All that material has very different characteristics. It is recorded with different cameras, uses different resolutions, frame rates, is saved with different video codecs and bit rates, and varies greatly in image quality. To show all those different sources as well as possible on your television screen, an enormous amount of know-how is required, and an impressive portion of computing power.
The P5 in five words
The know-how comes from the bright minds of Philips and their many years of experience. The computing power, which is provided by the in-house developed P5 processor. It owes its name to the five pillars of image quality that the processor optimizes: source signal, sharpness, color, contrast, and motion sharpness.
Evolution of the P5 processor
Philips introduced the P5 processor in 2017. Since then, a more powerful version with more functionalities and improved image quality has been released year after year. The second generation P5 brought us Perfect Natural Reality, Philips’ own technique to give all your SDR content a nice HDR look. The third generation P5 in 2019 brought improvements in detail reproduction and noise reduction, and added support for Dolby Vision HDR.
AI drives the decisions
In this new generation P5, the P5 AI Perfect Picture Engine, Philips uses artificial intelligence (AI) to optimize edits for each image. How does that work? Over the past 30 years, Philips has developed a database of millions of videos that it uses to assess image quality. Based on that database, Philips trained a neural network using machine learning by having it analyze those millions of videos. In essence, you give the neural network numerous examples with the desired answer, for example: this image is a landscape or this image contains a face. After the training, the network itself can provide the answer for images it has never seen before. In this way, the neural network classifies each image in one of five categories in real time: nature, face, movement, dark or other.
The processor then uses that classification to better match the work in each of the five pillars of the P5 (source enhancement, sharpness, color, contrast, and motion) to the type of image detected. The result is a more natural and realistic image.
Even smarter: P5 processor adds AI – Intelligent Dual Picture Engine
The OLED + 935 series features a groundbreaking audio solution from Bowers & Wilkins. In addition, Philips has equipped this top model with an even smarter P5. In addition to its new AI functions, it received four additional improvements and new features. These are ‘AI Smart Bit Enhancement’, ‘AI Machine Learn Sharpness’, ‘Improved² Perfect Natural Reality’ and a unique anti burn-in technique. This combination earned the device an EISA award.
P5: Smart detail
Sharpness and detail, these are properties that viewers invariably put forward as one of the most important factors for a realistic image. Traditional sharpness enhancements apply an algorithm to the entire image. But that can be better. AI Machine Learn Sharpness uses the power of the neural network to make custom adjustments locally within each image. For example, it can distinguish between areas of the image with a complex texture, areas where noise is visible or recognize faces. Each of those parts is given an adapted treatment to accentuate detail. For example, the algorithm does not have to compromise on the entire image, but uses the optimal approach for each part of the image.
Perfect color transitions
Streaming is becoming increasingly popular, YouTube, Netflix and other streaming services are an increasing part of your viewing feed. Streaming services always make a compromise between image quality and bitrate (how much data per second a video uses). And when the bitrate drops too low, you can sometimes see very clear errors in the image. Soft color gradients such as an impressive sunset then show visible color bands, instead of a smooth seamless color transition.
To address that problem, the TV P5 processor adds AI for Smart Bit Enhancement 2.0. A combination of new hardware and software, accurately detects problem areas and neatly removes the color bands. Thanks to this smart, localized approach, parts of the image where a lot of detail is visible remain untouched.
Impressive HDR picture with all your sources
Perfect Natural Reality is a unique Philips asset. This exclusive technology gives all your SDR images, for example from live TV, your old Blu-ray movies or game consoles, a beautiful HDR look with deeper contrast, intense white detail and richer colors.
The fourth generation P5 processor is now equipped with Improved² Perfect Natural Reality. The processor uses a new algorithm to detect and improve highlights. They therefore give an even more realistic impression. They also accentuate the natural sharpness of the image and thus give the image extra depth. That makes the image so real that it almost becomes tangible.
Renewed anti burn-in technology
Channel logos or the user interface of some games sometimes remain in the same place for a long time. After very long use, this creates the risk of burn-in. That is why Philips has developed a new anti burn-in solution. An advanced detection algorithm divides the image into 32,400 zones, and keeps a close eye on whether the image changes in each zone or not. For example, it finds very accurately the logos and static parts of the image. By reducing the intensity of the image only for those parts, burn-in is avoided without affecting the image quality.