HDR10 on TV

When is a 10-bit panel a 10-bit panel? If it can accept 10 bits per color and can represent it, right? So if you can differentiate 1024 different brightness per color.

But this can be achieved in two different ways. Either one has a 10-bit D / A converter, which converts these 10 bits directly into an analog value, which is then applied to the TFT transistor. This would be a "real" 10-bit panel. Or by using slightly less color grades, and then very quickly switching between adjacent color values. For example, a "real" 8-bit panel, ie a low-noise 8-bit D / A converter, can be used, and the four remaining two bits (ie four brightness levels) can be realized by this fast switching back and forth. This method is commonly called frame rate control (FRC).

How does this work in detail? In any case, it is only a matter of producing inter-tones that lie between two adjacent colors. There are 10-bit data, ie values ​​between 0 and 1023. The 8-bit panel shows the upper 8 bits and the lower two bits must now be realized by fast switching of the color. If these two bits are both 0, the lower of the two colors is displayed. If these two bits have the value 01, 3/4 of the time the lower color value and 1/4 of the time the next higher color value is displayed. If these two bits have the value 10, half the time is displayed the lower color value and half the time the higher color value. And if the two bits have the value 11, the lower color value is displayed for 1/4 of the time and the higher color value for 3/4 of the time. If you do this fast enough, the human eye really sees the mixed color.

Let's take an example. The 10-bit value is 55, which means binary 0000110111. The upper 8 bits now go directly to the panel. Thus, the 8-bit panel represents the 8-bit color value 13, specifically any value between 13 and 14. The two lower bits are 11, so the 8-bit panel must be 1/4 the time The value 13 and for 3/4 of the time the value 14. The visible result is then the correspondingly weighted mean value, that is to say the brightness, which corresponds to the value 13.75, ie, 55/4. Perfect.

Question: If you switch back and forth, does not that flicker?

Answer: Yes and now. First, it is only about very fine color nuances, so the mixing color between two technically possible colors. Since the differences are anyway very small and even a minimal flicker would not be noticed at a certain distance. But usually such a TFT also represents a capacitance, i. E. The switching operations can not happen as quickly as possible. And even if the TFT transistor is still fast enough, the liquid crystals of the LCD have a certain inertia at the latest, so that actually a mean value is set and it does not actually flicker. In addition, this flicker with a 100Hz panel and then synonymous with 100Hz happens, so you should not see therefore any more of it.

Conclusion: In practice, this theoretical flickering is no longer visible.

There are, however, both variants of panels on the market, ie panels which accept 10 bits and directly convert via D / A converters, and panels accept the 10 bits and represent these internally via FRC. Both are however officially under the designation 10-bit panel.

What if the TV really only has an 8-bit panel? This does not have to mean anything at all, because just as a FRC panel can do this fast switching back and forth internally, of course, the picture processor of the TV this switching back and forth can already pre-calculate and these values ​​then fixed and ready the 8- Bit panel. If, as in the example, the 10-bit value 55 is to be displayed, the TV can send the value 13 to every fourth frame and the value 14 to the panel. And he also reached an average value of 13.75. Again, the higher the panel fre- quency, the finer the FRC will be.

Therefore, at least in theory * no * significant difference in the visible result is to be expected, whether panel with 8 or 10 bits is relatively no matter, as long as it is cleanly solved and with the 8-bit panel is really indicated by FRC.

In principle, one could even see an advantage in it when the picture processor of the TV makes. Because in the calculation of the picture are done so many calculations. The entire image settings must be counted on the image, such as brightness, screen, color, gamma, etc. In order to avoid any visible rounding errors, this calculation usually takes place within the image processor with even higher accuracy. 12 or even 14 bits.

And it does not mind to generate by FRC even finer differences. If you show the other color for 1/8 of the time and 7/8 of the time, then you have three additional bits implemented. Speaking through a FRC, which comes from the TV, this could ratchet still more bit information from the image calculation to the final picture, and would not already cut off with "only" 10 bits.

In this respect, I would not look so strongly at the bit depth of the panel. Global statements are not possible anyway. And that is the reason why e.g. Sony on its devices this information does not even provide. This value says only very conditionally what. It just depends on how well it is implemented with a concrete model. A well-made 8-bit control with FRC can be better than a badly made 10-bit control without FRC. The question is then rather, whether the Bildgine a good FRC can make or not.

Clearly, if a 10-bit panel is used, then you can be half-sure that at least the panel manufacturer has his internal FRC under control, if he realizes it so at all. In this respect, a higher number of bits i.a. Already an indicator for a good presentation. But it does not necessarily have to be worse if fewer bits are present.