1
\$\begingroup\$

As far as I am aware,

  1. HDMI 2.1 does support 12-bit 4K 60fps, Also it doesn't use TMDS, rather FRL. Sends upto 48Gbps

  2. GPUs can do calculation in fp32 and from some reference, I think that it can send 16bpc (48-bit per pixel deep color) (Assuming that it is a high end Quadro GPU) (fp16 had the capability to send 12bpc 36-bit deep color, right?)

But there are monitors that don't support more than 10bpc 30-bit depth deep color

Why is that? Why do displays have limited bit-depth?

Can you give me some idea about the science behind this? Why does it cost more to produce higher bit-depth monitors?

Isn't it like, RGB pixels on the displays are emitting photons? Then how do different displays differ in terms of emitting photons?

\$\endgroup\$
  • 1
    \$\begingroup\$ Because it costs more and no-one would be able to differentiate that degree of color depth. \$\endgroup\$ – brhans Sep 22 at 19:54
  • 1
    \$\begingroup\$ Well, You can Really Differentiate 8-bit Grayscale Darker Areas. You can see that Banding, Sir. I would Really Appreciate it if you can give me slight idea about the Technology of Displays, Why it Costs More.... \$\endgroup\$ – user263557 Sep 22 at 19:56
  • 3
    \$\begingroup\$ @WesleyLee bad analogy to compare those audio parameters with those video parameters. 8-bpc color depth is OK for graphics but with visible dark artifacts in video, 10-bit colors is quite good and 12-bits is more than eye can possibly detect. 128 kbps MP3 is absolute garbage, compares to 6-bit colors with dithering in video world, while 320kbps MP3 is usually transparent compared to original. \$\endgroup\$ – Justme Sep 22 at 20:26
  • 3
    \$\begingroup\$ Because D/A converter is expensive. A few years ago most panels are actually 6-bit believe it or not. \$\endgroup\$ – user3528438 Sep 22 at 20:29
  • 2
    \$\begingroup\$ HDMI was invented for televisions, it's a consumer A/V interface which computers did not need. Computer monitors used DVI which did not support more than 8bpc and it was not until DisplayPort came and started to support more depth. Cheap consumer displays do use HDMI but not many features of what HDMI is capable of, they use simply the minimum featureset on top of DVI. Cheap means 8-bit panels. \$\endgroup\$ – Justme Sep 22 at 20:33
2
\$\begingroup\$

Can you give me some idea about the Science behind this? Why does it cost more to produce higher Bit-Depth Monitors? Isn't it Like, RGB pixels on the Displays are Emitting Photons? Then how Does different Displays differ in terms of Emitting Photons?

One of the things no one else has mentioned so far is the effect of gamma. You mentioned that you use 32 bit linear floating point, but the values that HDMI outputs are gamma corrected (spread nonuniformly in amplitude so that the difference between brightness levels gets larger as you approach maximum brightness), so if you want to show 10 bits, you actually need a panel that can change the number of photons it outputs by a lot more than just 1023 times.

That is a really tough thing to do. It means you have to have a pixel that can be very bright, and also very dim in a precise way. If you look into HDR video specifications, they either require very expensive panel technologies or use tricks like locally dimmable backlights. Even then, at a per pixel level, a lot of those technologies still might struggle to show you even 9 or 10 real bits.

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ Aren't gamma corrected Values sent as Floating Point Values? Doesn't that Require more bits to go through the HDMI and More Bit Supporting D/A Converter? I think i've read, To save 8bpc non-Gamma Corrected Colors we would need 12bpc Floating Point Precision \$\endgroup\$ – user263557 Sep 23 at 2:15
  • 2
    \$\begingroup\$ @AmitraksharBiszz No, floating point is not used. You're just sending a number between 0 to 255, or 1023, ... The actual information you're working with is integer, so no need for floating point. \$\endgroup\$ – user1850479 Sep 23 at 2:20
  • 1
    \$\begingroup\$ @AmitraksharBiszz No, absolutely not. For a gamma of 2.4 you take the number of photons and raise to the power of 0.45. That is very nonlinear. \$\endgroup\$ – user1850479 Sep 23 at 2:33
  • 1
    \$\begingroup\$ @AmitraksharBiszz Gamma correction is applied to images (for example, from a camera). Data in your computer is already gamma corrected, so inverse gamma correction (raise to power of 2.4) is applied to the data received over HDMI by your monitor. \$\endgroup\$ – user1850479 Sep 23 at 2:44
  • 1
    \$\begingroup\$ @AmitraksharBiszz, as asked, please cut out Starting Every Word With A Capital Letter In Your Comments. It's hard to read. I appreciate that English may not be your first language but you can copy the style you see the other posters using. Thanks. \$\endgroup\$ – TonyM Sep 23 at 10:01
3
\$\begingroup\$

The short answer is, cost. Higher bit-depths require more bandwidth at every point in the digital signal chain, from rendering unit, DRAM, link, all the way to the DACs that drive the panel.

Technology being what it is, as pipeline costs have come down, deeper color standards have been introduced, so panels have evolved along with the rest of the display ecosystem to offer both higher resolution and increased color depth. To enable this, HDMI 2.1 supports 8K and 48 bits per pixel now.

Looking further up the pipeline, cameras commonly support 12 bits, with 14 coming available (RAW modes). Editing systems already use 16 bits per component internally and quantize back down at the end when the work is finalized.

I should also mention the various HDR coding schemes that seek to improve display dynamic range in a more bit-efficient way than just adding more color bits.

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ Thanks, Sir. And Yeah, As An VFX Artist, i can say at Industry We Do use 16-bit. and also 32-bit floating-Point Linear Encoding..... It would be great if you could give me a small introduction to D/A Converts... \$\endgroup\$ – user263557 Sep 22 at 20:54
0
\$\begingroup\$
  1. Yes, true, but it also goes better than 12-bpc 4K 60Hz. It supports anything that goes up to 48 Gbps, and beyond that if DSC compression is used, the compressed stream is 48 Gbps so there is even more uncompressed bandwidth.

  2. Certainly sounds possible. Many devices should support 16 bpc Deep Color just as easily as 12 or 10 bpc.

And yes, there are displays that support 12bpc. In fact if a HDMI display device supports 10bpc link, it must support 12bpc link too. Whether it actually uses those bits to drive the panel is something we cannot know. Note that the panel native depth may be less that what is possible by using dithering, so it is certainly possible to show images that equal 12 bpc with only a 10bpc panel, so that should count as a 12bpc display.

10 bits with HDR is slightly above the curve where eye can discern the quantization levels, but 12 bits with HDR is below the curve where eye cannot see a difference between quantization levels any more. If you have a 10-bit HDR capable panel and can make a display with adjustable backlight, preferably several separately dimmable areas, and work out some signal processing magic to get HDR image quality equaling to 11 bits with dithering and backlight control, there is no real visible improvement the human eyes can see.

| improve this answer | |
\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy