10

I'm not a photographer, but I'm struggling to understand HDR technology as it relates to TVs and monitors, so I thought this would be a good place to ask.

First, I got a good understanding of the terms "color space" and "bit depth" by reading this excellent answer.

How does HDR technology relate to these terms? Is HDR simply a combination of a 10-bit color depth and really wide color space, or is there something more to it?

In the marketing language, certain areas of the screen (such as the rising sun) are said to be "brighter" than other areas of the screen on a HDR display. What does "brighter" actually mean here — does it just mean "farther to the edge of the color space"?

I currently have a DELL U2410 monitor, which, as I understand, has a 12-bit internal processor and an 8-bit panel with FRC dithering, and supports an Adobe RGB color space. Even though this is not true 10-bit, will it produce a result similar to HDR if I use a device which can output HDR? I currently set the color space to "sRGB" since Adobe RGB looks too oversaturated. Could this be because my graphics card does not support a 10-bit output?

EDIT: I am talking about "HDR" specifically as it relates to how the term is used in modern 4K TVs.

Improve this question
14
  • 1
    The problem with bit depth is that it has absolutely nothing to do with the size of a color space. You could have a very high bit depth in a very narrow color space where the steps between each gradation are very fine. You could also have a fairly low bit depth in a very wide color space where the steps between each gradation are much coarser.
    – Michael C
    Commented Jun 13, 2017 at 9:55
  • 1
    "Could this be because my graphics card does not support a 10-bit output?" It's probably much more likely your monitor's saturation is just turned up too high. Most screens come from the factory way too bright, way too saturated, and way too cool (high color temperature). Have you calibrated and profiled your system with a colorimeter?
    – Michael C
    Commented Jun 13, 2017 at 9:59
  • 1
    HDR as is normally used in still photography is not the same thing as HDR when used to describe 4K TVs. The reason we have more content containing the expression 'HDR' is probably because it's been around still photography for about 165 years and is a relatively new term with regard to video screens.
    – Michael C
    Commented Jun 13, 2017 at 10:26
  • 1
    brightness and color space are orthogonal concepts.
    – ths
    Commented Jun 13, 2017 at 11:38
  • 1
    Try to break this question down. It is so full of incorrect assumptions, I cannot begin to formulate an answer! HDR needs far more than 10-bits to be usable and bit-depth does not imply a high dynamic-range. My monitors are calibrated to 14-bit LUTs and have a reasonably low 400:1 maximum contrast.
    – Itai
    Commented Jun 13, 2017 at 22:24

5 Answers 5

Reset to default
7

TV technology and video encoding is in principle outside the scope of this site. Still, as photographers we are presumably interested in new display standards that promise both more dynamic range and wider color gamut. HDR computer monitors are coming, so it probably won't be long before photographers too will be interested in HDR displays.

HDR by itself just means High Dynamic Range. Dynamic range is the brightness ratio between the brightest and the darkest parts. "High" is relative to what the sensor can capture or to what a standard display can show.

HDR video is one of several standards for video encoding. They have names like HDR10, Dolby Vision, and Hybrid Log-Gamma. They all support higher bit depth (10-12 bits), wider color gamut (Rec. 2020 and/or DCI-P3), higher dynamic range, and 4k resolution. They are not compatible with each other, but some devices support more than one HDR standard.

"HDR TV" doesn't necessarily mean much, except that it will accept a HDR video signal and display an image. It does not necessarily mean that it can display the wider gamut and higher dynamic range, just that it can read the signals and show something, even if what it shows is plain sRGB with a mundane dynamic range.

"UHD Premium" is a certification standard for displays that can actually show HDR. Requirements include:

  • minimum 10 bit color
  • minimum 90% of the DCI-P3 color gamut
  • use a HDR transfer function (The transfer function for sRGB is called 'gamma', it translates bit value to brightness value. HDR uses a different function so it can cover a larger dynamic range with a small increase in the number of bits)
  • either: Minimum 1,000 nits brightness and max 0.05 nits black level, contrast ratio 20,000:1 or better (for LCD displays)
  • or: Minimum 540 nits brightness and max 0.0005 nits black level, contrast ratio 1,080,000:1 or better (for OLED displays, they can't go as bright but have better black levels)

Standard displays are 200-400 nits and rarely go beyond 1,000:1 contrast ratio. So a "Premium" HDR display has significantly higher peak brightness as well as an order of magnitude better contrast.

Brighter highlights, higher contrast ratios, and a different "bits to brightness" encoding/decoding all combine to enable the "high dynamic range" part of a HDR TV.

Color space and bit depth is only related to HDR in the sense that they are part of the same standards. Yes, you need more bits to represent a higher dynamic range without banding. But the larger color space is a different decision, it's more a matter of including enough improvements to make a new video standard worth the trouble. (4k is also part of the standard, and 4k is not related to HDR either.)

"The sun looks brighter on a HDR display" means that a) the display emits more light in the brightest parts of the scene, and b) it may look even brighter because the rest of the scene can be dimmer without losing detail.

So it's a real improvement; the wider color gamut and higher dynamic range can contribute to more vivid photos and video.

The caveat is that TVs/monitors may be sold as "HDR compatible" without any actual HDR capability in the display. The content needs to be made specifically for HDR to make use of the new capabilities. And for computer monitors, you need a graphics card that supports HDR output.

Improve this answer
1

HDR doesn't relate to color space at all, as it stands for "High Dynamic Range", ie. brightness, not color, contrast.

Traditional TV signals had a rather low contrast ratio, as old TV sets had a contrast of only 30:1. Modern displays can produce a lot more, 1000:1 or more with dynamic backlight (local dimming).

"HDR" displays are just a marketing term for displays with a very high contrast ratio.

But using this high contrast will require a higher bit depth signal or you'll have only a few steps spread out over the whole range which may introduce a kind of banding just like in the color domain.

You can't squeeze 1000:1 into an 8 bit brightness signal, so you need a higher bit depth. 10 bit can encode 1024 steps, which still isn't that high really, but better than 256.

Calling those displays "HDR" is, from a photography perspective, still laughable, as HDR photos are encoded in 32 bit floating point values with near infinite steps, but the dynamic range is higher than that of a traditional CRT TV or the standard NTSC/PAL signal. (Note that you need an "HDR" signal source to actually enjoy that capacity.)

Improve this answer
8
  • 1
    How do colors on a color gamut relate to the brightness of those colors? Are colors closer to the edge of a gamut (e.g. Adobe RGB) necessarily brighter than the ones closer to the middle? Would it be correct to say that without a sufficient contrast ratio, the display cannot accurately show colors closer to the edge of a color gamut?
    – Vikram
    Commented Jun 13, 2017 at 12:03
  • When I say "closer to the edge of a gamut", I am specifically referring to how the gamut is represented on a diagram like this: mosaicdesignservices.com/webgraphics/presentations/2007-02/…
    – Vikram
    Commented Jun 13, 2017 at 12:06
  • 1
    color gamut has nothing at all to do with brightness. it's a different dimension.
    – ths
    Commented Jun 13, 2017 at 12:08
  • it's like volume and frequency of a sound.
    – ths
    Commented Jun 13, 2017 at 12:10
  • Okay that makes sense. So I guess modern HDR 4K TVs are a combination of three things: wide color gamut, 10-bit color depth and a high contrast ratio.
    – Vikram
    Commented Jun 13, 2017 at 12:20
1

HDR in the modern TV sense refers to one of several different HDR specifications for color. To understand what High Dynamic Range means, we first need to understand what dynamic range is.

Dynamic Range is the difference between the highest and lowest values in a signal or display. A high dynamic range TV generally means that there is a large amount of brightness levels between the darkest and lightest possible portions of the display that can be shown at the same time.

The HDR specifications specify more than just the max and min brightness levels that must be able to be displayed though. They also specify that a relatively high bit depth must be possible. This allows for a video to have a greater number of levels of brightness within it, which also means that more subtle variations in color can be created and thus better use of the increased range of brightness that can be produced. The higher the dynamic range, the more change in color between each value at a particular bit depth, but you can have fairly low dynamic range displays or fairly high dynamic range displays that still have a high bit depth.

There is also requirements for contrast ratio. Contrast ratio is similar to dynamic range, but not exactly the same thing. Contrast ratio is the relationship, in absolute light level, between the darkest and brightest values the display can simultaneously display. It is dependent both on how bright the display can get and how well it can block light. The contrast ratio determines how well a bright and dark portion of a scene can co-exist at the same time. Static contrast ratio is the meaningful value as it is what the display can really do. Dynamic doesn't mean a whole lot because it depends on being able to do adjustments that aren't at a pixel level and don't really produce an ideal image quality.

Improve this answer
0

To comment two well-known statements here:

"You could also have a fairly low bit depth in a very wide color space where the steps between each gradation are much coarser."

"Color space and bit depth is only related to HDR in the sense that they are part of the same standards. Yes, you need more bits to represent a higher dynamic range without banding."

Both don't take dithering into account. In the absence of dither, the statements are correct but with dither applied, in theory, a smooth gradient would be even possible with only one bit, although it's extremely noisy when applying spatial dithering.

If one accepts a lower SNR, there won't be banding regardless of the bit depth.

If it is true that bit depth and colour space are not related (which makes sense as any colour space could exist in a purely analog domain as well without any bits to consider at all), then the logical conclusion would be: theoretical perfect dithering techniques assumed, a given bit depth - analog to audio - only limits the achievable signal to noise ratio and thereas the dynamic range as the noise will mask the lower end, however not the possible amount of colors, not any banding and no brightest whites or whatever.

Improve this answer
7
  • 1
    Sure, but then you are sacrificing resolution for color and it only works to a point because if you don't have enough pixels to dither over, it becomes noticeable to the eye. Also,. to the best of my knowledge, most displays aren't applying dithering on the fly to compensate for something like this, they are just displaying pixel values provided, so dithering doesn't make much of a practical difference in relation to things like the various HDR tv specs.
    – AJ Henderson ♦
    Commented Jul 24, 2017 at 13:45
  • @AJ Henderson: thanks for your reply. You are right that many practical implementations lead to effects, which the oversimplified and - from the signal theory's point of view incorrect - assumptions suggest. However, the fact alone that practical implementations are so often so bad, doesn't make the oversimplified assumptions correct vice-versa, either. I can't prove it yet either, but I doubt that dithering in fact reduces resolution as I think most people confuse resolution and signal-to-noise ratio and these are two different parameters determining the quality.
    – little-endian
    Commented Jul 27, 2017 at 2:55
  • One can beautifully test this on one's own by using the renderer "madVR" under Windows (recommended in conjunction with the player MPC, but it should work with others, too) which supports several great implementations of dither. Using "randomised dither", one can reduce the bit depth to even only 1-bit per color channel (leading to 8 distinct colors only in total) and still have all the colors and no banding.
    – little-endian
    Commented Jul 27, 2017 at 2:57
  • The image is very noisy (just like a bad analog TV signal back in the days) and certain detail is lost in the noise of course, but the nominal frequency response of the image (resolution) seems to remain unaffected which can be indirectly proven by the absence of introduced aliasing which would be to be expected if the real resolution frequency-wise would be lowered by the dithering. In the audio domain, it's also exactly the same. There dither definitely does NOT reduce or limit the frequency response in any way, but depending on the bit depth it will raise the overall noise floor.
    – little-endian
    Commented Jul 27, 2017 at 2:59
  • Video resolution equals audio frequency response. Hence from my understanding, it is correct to talk about a higher dynamic range and lower noise floor with a higher bit depth (one bit giving about 6dB of additional SNR), but not about "finer resolution of colors" or even a higher color space as all these assumptions are wrong in the audio domain and a video signal could always also be interpreted as a high frequency audio signal too so all parameters logically have to have a 1:1 expression (which they seem to have in practise too).
    – little-endian
    Commented Jul 27, 2017 at 3:00
0

HDR is high dynamic range. The older technology is SDR - standard dynamic range. When we talk about colorspaces, there are also colorspaces specifically designed for HDR like ICtCp. ICtCp is derived from IPT (a colorspace for SDR) by using a different function for lightness.

Standard colorspaces are usually poorly defined for highlights and dark tones. This means highlights would be most likely encoded as a small region around pure white (i.e. loose precision/data/information there). Similar with shadows.

So, HDR in the colorspace terms is: a different lightness function (HLG - hybrid log or PQ function) + better defined color information for highlights and shadows.

Using HDR colorspace is only relevant for editing/storing when you deal with SDR content (that is why you can recover shadows/highlights when editing RAW files). When you deal with HDR content - you would have to use a colorspace for HDR content.

A lot of research papers clearly say that SDR colorspaces (i.e. CIECAM02, Lab) were only tested for illuminant around 100 nits. Higher illuminant or extreme changes in lightness and SDR colorspaces enter an "uncharted" territory. They could work fine, but less optimal as colorspaces defined for HDR.

Improve this answer

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.