Marketers like large numbers, hang the physics of realising an image and the psychovisual complexities of rendering a scene into the visual cortex.
We saw all this with HD. Most European countries adopted 1080 interlaced HD, disregarding tests that showed 720p50 pictures looked better. We know that to avoid the artefacts of interlace vertical resolution had to be sacrificed, so a progressive scan picture with fewer lines looked sharper, but the number 720 was not that much more than the old SD standard of 625 lines. The detail that the SD picture had only 576 active lines was irrelevant, it’s all about numbers. On top of that the public were sold 1080P receivers, when broadcasts were interlaced, OK you could view 24P movies, but even today, 1080P remains overlooked, passed over in the numbers game.
The next conn is 4K, sounds like four times 1080, its just another detail that the latter is picture height, the former is width.
Viewing test after test has shown better pixels beat more pixels, but the CE guys like things simple, 4K presents the opportunity to sell at a better margin over the miserable profits to be made from HD screens. At the end of the day, its about profit, we all need to earn a living, so from studio to broadcaster to CE vendor, profit must take precedence over the niceties of picture quality and psychovisual realities.
What triggered off these recurring thoughts? I was at an ARRI event today, and the usual question came up, how does their 3.4K sensor stand up against a 4k sensor, when creating DCI or UHD pictures? ARRI cameras use a modest upsample of 1.2 times to get to 4K. This question ignores the fact that 4k single sensor camera, with the typical Bayer pattern colour-filter array has only 2048 x 1080 green photosites, and half that for red and blue. The de-Bayer algorithm is interpolating, effectively upsampling up to a 4096 x 2160 raster. If you want real 4k you need a camera like the Sony HDC-4300, a 3-chip camera with 4K sensors.
Of course it’s all numbers again, what really counts is what the pictures look like. What If the camera is considered as a black box, an opto-electric transfer engine? Then the number of photosites; the algorithms in the deBayering and the colour matrix; the size of the photosites; the dynamic range; all these factors contribute to the picture quality. What matters are how factors like textural reproduction and the sharpness of edges in the image are perceived? These depend on the lens and the camera together, so headline numbers like the number of photosites are important, but they don’t tell the whole story.
There is no doubt that if you sit close to a large display, then the individual pixels are visible, and that 4K displays come into their own as screen diagonals exceed 70 or so inches. 4K is a real improvement in image reproduction quality, but it’s not the full story. Rec 2020 includes wider colour space, higher frame rate and increased bit depth. Companies, notably Dolby, are working on formats to increase dynamic range.
It has long been my suspicion that TV receivers are designed to deliver punchy images in the showroom, but as for prolonged viewing, well the picture “enhancements” of consumer gear just need turning off. As usual, it’s all about quantity not quality.
Related Editorial Content
For all of last year’s talk about distributing 4K television to home viewers, there has been little real progress on making it a reality. The only place 4K technology is being used regularly is in some sports production and i…
The recent spate of announcements from camera manufacturers about plans to market HD and 4K shooters capable of accepting B4-mount standard lenses has been welcomed by the industry and appears to have brought live UHD broadcast launches closer.