HDR Nits Are a Lie: Embrace the Stop — Archival Works

Okay so what’s going on here? How is that Vizio PX65-G1 the same price as the Samsung Q70T, but nearly five times brighter? Why is LG’s flagship CX the same price as Samsung’s flagship Q90T, but nearly half as bright? And lastly, why is that Sony display so insanely expensive but it’s brightness only middle of road?

Obviously, there is more to a TV or monitor than just how bright it is. There are lots of other factors at play like color reproduction, sounds quality, and more. But still, you would think that brightness would correlate directly with price, especially because, as we established, it’s one of the key things that defines a screen’s dynamic range.

But that’s just it. Brightness is one of the key things that defines a screen’s dynamic range. The other thing is black level.

This is where stuff gets a little complicated. You see, most screens can’t show true black. In fact, OLED and plasma screens are the only screens that can show a true black image as really black. Try this out: pull up a black screen (credits in a movie are a good place to start) on any LCD screen. This could be your TV, phone, tablet, or another device. You’ll notice that the screen is still glowing a little bit. It isn’t true black. While technically, true black should always be 0 nits, in the real world, it rarely is. Because of this, dynamic range is actually worse on those TVs with a higher brightness black.

“Now surely the difference in black level can’t be very much, right?” You might ask. You might be thinking that even if the black level on the worst TV is, say, two nits, and on an OLED or plasma, it’s zero, that’s only a two nit difference. Some of the TV’s in that chart have brightness differences over 1000 nits model to model. So why does the black level actually matter when it’s such a tiny difference? Well, we’ll get there, but we need to stop thinking in terms of brightness and nits. We need to look at how camera folk measure dynamic range.

Camera folk don’t think about dynamic range as nits the way TV and monitor manufacturers do. Instead, they think of it in term of Stops. A stop is a halving or doubling of light. So, if you have two light bulbs and one is twice as bright as the other, you could say the brighter one is “one stop brighter” or the dimmer one is “one stop darker”. If the bulbs are 4 times different in brightness, they are actually only two stops apart: the darker one is the brightness of the brighter one cut in half, and then cut in half again. That makes it two stops dimmer.

For one last example, if the brighter bulb is 24 times brighter, it would be 4.5 stops brighter than the darker one. This is because if the dim bulb is one unit of brightness, we could double it to two units (that’s one stop), double it again to four units (that’s two stops), double it again to eight units (that’s three stops), double it again to 16 units (that’s four stops), and then we need to bump it up by 1.5 times to get it to 24 units. Since that last bump was only a 50% increase and not a 100% increase, it’s only a half stop, giving us a total of four and a half stops.

This is a really useful way to measure dynamic range because it’s a lot closer to how our eyes work. We don’t perceive brightness linearly; we perceive it logarithmically.

Digital cameras, like TVs and monitors, have a fixed dynamic range. That is to say, they have a limit of the brightest thing they can capture and the darkest thing. Anything beyond that limit will be maxed out and be turned into solid white or solid black and lose all of its details.

Now I can hear some of you thinking “but I can change a digital camera’s ISO! You can’t change a TV’s ISO! That makes all this comparison worthless”. First of all, if you don’t know what ISO means, don’t sweat it. You can safely skip this paragraph. For the rest of you: no. You actually can’t change a digital camera’s ISO. A digital camera sensor’s ISO is fixed. This fixed ISO value is called its Native ISO. Any changes you make to ISO in your camera’s settings are simply an estimate of what that ISO would look like by artificially brightening or darkening the image. This is analogous to push or pull processing of a film stock. You can process 800 ISO film as 1600, but that doesn’t make it 1600 ISO film stock. Don’t believe me? Try this out: use a digital camera that can shoot RAW and take a picture of something really bright, like a very bright light bulb (don’t take a picture of the sun because you’ll cook your sensor). Shoot wide open on your fastest lens with a kind of slow shutter speed. Obviously, the picture will be totally blown out. “But wait!” you say, “I shot RAW so I can lower the ISO in post!”. Try it. It’ll still be blown out, but it’ll be gray with no detail instead of white with no detail. Even if you set your ISO to a mythical 0.001 you still won’t get any detail. You might get more detail out of something that looks to be blown out on your display (say, a cloud), but that’s because the image wasn’t actually clipped to begin with. Your display is what was clipping. When you lower the ISO, you are simply dropping the brightness of the image down into your screens “window” of possible dynamic range. Your camera’s ISO settings are a lie.

But anyway, I digress. Camera dynamic range is static and unchanging, just like a screen’s. Therefore, we can think of a screen as a sort of camera in reverse: instead of absorbing light it emits it. This means that we can use stops to measure a screen’s dynamic range as if it were a camera.

Now, this is a little challenging to do because:

1. OLED and plasma screens can turn off individual pixels (a true, 0 nit black). Since zero times zero is still zero, we need to use the brightness of the darkest shade the screen can produce before turning all the way off as our starting point. The same goes for LED/LCD TVs that use local dimming to produce true black.

2. The black level in nits is not something normally measured and I don’t own the equipment necessary to measure it myself. Therefore, we must rely on theoretical numbers. With this in mind, we’ll be using the black level numbers put in place by the UHD Alliance. They state that LCD/LED TVs reach about .05 nits, and OLEDs can get down to 0.0005 nits.

So, let’s look at that same table from earlier, with the estimated black level and dynamic range in stops: