While I dislike the implementation of dynamic contrast by backlight global dimming (dynamically adjusting the intensity of the background lighting) on my LCD monitor, I really think, that a more sophisticated solution could boost the image quality tremendously without or with at least negligible drawbacks.
What method will the Pimax 8K use?
I never use the dynamic contrast option of my monitor for three reasons:
Dark images are way too dark. The bright areas are not taken adequately into account.
The dynamic contrast adaption is way too slow (it takes about a second to adapt).
There is no way to configure the dynamic contrast manually.
I would prefer a setting, where the dynamic contrast only takes into account the natural adaption of the eye, so that – in a dark image like e.g. a cockpit in a space game – the screen is dimmed only to a degree that lets bright areas still look relatively bright.
In any case I think, that dynamic contrast is a must have with LCD displays for obvious reasons. I just want to be sure that it’s not overdone, or in best case manually configurable to a degree (e.g. different presets: low, medium, heavy).
First of all: The natural dilation and contraction of your pupils, that you mention, works in VR just like it does in real life, so it is for the most part (more on this later) neither neccessary, nor desireable to simulate it on top of the real thing, just as we do not need to simulate convergence in software (EDIT: …which some still believe). If the environment is dark, your eyes will adjust; Within the enclosed headmounted display there is no ambient light from the real world, unbalancing and overpowering things, the way there is with your monitor, or cell phone screen.
Second: You may have noticed that a great many games have this feature already, way back since Half Life 2: The Lost Coast. The games render internally at High Dynamic Range, but since few monitors are capable of such range, the imagery is tone mapped to fit them, which can be seen as brightness cranking up several levels, when you walk into the shadow beneath a portcullis. Quite often you’ll even see what is ether the algoritm overshooting, or deliberate simulation of the time it takes one’s eyes to adjust, when you walk out from a cave into daylight, and the “exposure” whites out momentarily, before stabilising.
Ideally we’d have HDR displays, and not need tone mapping at all, so that we could just let our eyes do their thing, but since we’re not there yet, we’re going to have to live for a while longer, with the tone mapping doing its second guessing us, probably to some degree getting into something of a jousting fight with the natural reflexes of our eyes, just like we may in the not too distant future find them, in the same manner, very much imperfectly using blur to simulate Depth of Field, on our fixed focus displays, assisted by eyetracking.
In Elite Dangerous, there is currently a much maligned adding of rather ugly ambient light, when you fly in over the dark side of planet that one can land on. The developers chose to appease those players who want to easily see where they are going, but they have taken a lot of flak for it, from those who prefer the proper darkness. :7
Hopefully the revised lighting system that is scheduled for the game, toward the end of the year, will sort this out… :7
Thanks jojon for your reply, but I was talking about something different: the dynamic contrast feature of LCD display hardware that lowers the intensity of the background lighting dynamically.
For example, if the displayed image is only a grey box on black background, instead of having the background lighting at 100% and the grey pixels at 50% intensity, the background lighting is set to 50% and the grey pixels to 100%. This way the black will be much darker while the grey rectangle appears the same (= much higher contrast ratio over time). Another advantage is less power usage / less heat dissipation.
Ok, I think I’m following you now; Not the old adjusting final output brightness and contrast, so that a screen doesn’t come across just a black slab in daylight, and a blinding floodlight at night, but tuning (histogram allowing) the backlight and transmissive layer in order to maintain the same intended output contrast, only with a better result on account of better blacks through less lighbleed, such as can be seen in some television sets, some of which even have their backlight divided up into zones, allowing for one part of the screen to still have something bright, whilst a dark corner is toned down.
I must say I find the “dynamic contrast” nomenclature somewhat misleading; Sure, the factors change, but not the product, other than fidelity/energydrain-wise. Anyway: Now I know. :7
I could absolutely enjoy some of that! Proper blacks remains a worry with LCD panels.
Especially if we had the display receiving and utilising HDR input, making use of that effective extra depth resolution when it is dark, and where there is typically more colour banding.
[quote=“jojon, post:4, topic:5481”]
I must say I find the “dynamic contrast” nomenclature somewhat misleading; Sure, the factors change, but not the product, other than fidelity/energydrain-wise.[/quote]
The “dynamic contrast ratio” (100.000.000:1 ) marketing label is a joke, of course.
But the basic idea is good. It just doesn’t work well on my monitor for the stated reasons. It’s actually very annoying and distracting instead of boosting the image quality. In dark scenes it ruins the image completely. It’s such a shame, because with a little fine-tuning the results would be great in any case.
While dynamic contrast sounds like a good idea, most actual implementations have issues. Ideally, every pixel in an image needs to be scanned (to determine its brightness), the entire image needs to have its brightness scaled (to transform the “grey box” in the 3rd post to white), and then the LEDs need to be dimmed appropriately.
That’s a fair bit of work, which would surely add some lag.
In addition, I would worry that inherently dark games, like Elite Dangerous would have strange lightness fluctuation, for example, when a bright star or planet suddenly comes into view.
This technology is rather trivial when compared to upscaling 1440p to 4k.
Nowadays almost every LCD monitor / TV has such a feature.
I assume that Pimax 8k will use it too and would like to have more information about their specific implementation.
[quote=“Hyperion, post:7, topic:5481, full:true”]This technology is rather trivial when compared to upscaling 1440p to 4k.
Nowadays almost every LCD monitor / TV has such a feature.I assume that Pimax 8k will use it too and would like to have more information about their specific implementation.[/quote]
What is your source for “trivial” comparison? Upscaling can be done with a simple box filter and doesn’t require changing the values of all pixels (to increase their brightness, once the LEDs are dimmed). I think dynamic contrast is significantly more work for the chip.
TVs can introduce some lag without people noticing. Too much lag induces VR sickness, so this is very important.
Pimax is using a very simple (limited) scaler chip; I would not assume Pimax 8K will include this. In fact, I hope they do not (or allow a way to turn it off) because of potential VR sickness issues.
I was hoping for a good upscaler. Changing the luminance of the pixels is a simple task that can even be done by the GPU (if synced correctly with the backlight dimmer).
Lag is irrelevant, because it’s not time critical. Since the dynamic contrast must be evened out over time anyways (to avoid flicker) you can just use the brightness level equotations of previous frames (so there is no added latency).
But you are right, if they are not able to get a good solution, it’s better to not implement it at all.
It would be just a pity to have such a nice headset, but which is not able to achieve its optimal performance, because it lacks some simple technique which could have improved visual quality, power consumption / heat dissipation and longevity considerably.
[quote=“Hyperion, post:9, topic:5481”]Changing the luminance of the pixels is a simple task that can even be done by the GPU (if synced correctly with the backlight dimmer).
Lag is irrelevant, because it’s not time critical. Since the dynamic contrast must be evened out over time anyways (to avoid flicker) you can just use the brightness level equotations of previous frames (so there is no added latency). [/quote]
While it is a relatively simple task, it’s not free. There are bandwidth considerations, which much be accounted for, which will add latency (lag).
Off-loading the work to the GPU makes sense, but the GPU will already be overloaded, due to the need of generating wide-angle content for both eyes. Using the brightness value of the previous frame will lead to “overshooting” as mentioned in the 2nd post.
70 dimming zones is nowhere near the effect of an oled display , The bloom On such a display is not viable for VR , it would basically intensify the god ray effect.
If you are interested in the topic, here is a source:
Backlight Local Dimming Algorithm for High Contrast LCD-TV
There must be a temporal filter anyways (as said: to prevent flickering), therefore the previous frames must be taken into account in any case. That means the brightness assessment is not time-critical (= it doesn’t matter if it’s lagging some frames behind).
[quote=“neal_white_iii, post:10, topic:5481”]
Using the brightness value of the previous frame will lead to “overshooting” as mentioned in the 2nd post.[/quote]
[quote=“jojon, post:2, topic:5481”]
Quite often you’ll even see what is ether the algoritm overshooting, or deliberate simulation of the time it takes one’s eyes to adjust, when you walk out from a cave into daylight, and the “exposure” whites out momentarily, before stabilising.[/quote]
As you see the post you referred to mentions that this delay (“latency”) does actually make sense. It’s not a bug – it’s a feature!
I don’t think it’s our “job” to discuss problems that have already been solved.
I have never seen a local dimming display in reality. There should be no or at least not much bloom and definitely not one that would make the god rays (fresnel lense artifacts) worse.
In any case the image should be way better than without any dimming.
Not that I am asking for local dimming, that actually would be a little bit more tricky to implement.
Good post. I think Dynamic Contrast comes down to the speed of it all. If you can see any dynamic process working (and hear it in the case of projectors and a dynamic iris) then it is more distracting than enhancing, and indicates a cheap solution or an uncalibrated display but when it works well it is great from my own experience. I agree you need controls to adjust the amount of dynamic contrast applied depending on how susceptible you are to noticing it. Purists have an option to turn it off.
However, I feel calibration of the panels is an important part of this process working well or not. An uncalibrated display/projector/TV can kill any dynamic process thereafter, white scenes can blind you and dark scenes can lose detail or the flip side can happen and bright scenes are dull and flat where dark scenes are grey and saturated. Dynamic contrast only works well when the display is calibrated to whatever gamut you are using. In the case of LCD VR that would be sRGB. This is something I hope somebody can test on the Pimax 8K. I would love it to hit 100% for sRGB but anything over 95% should allow dynamic contrast to have a nice range to work with. We also need a way to have controls to re-calibrate VR displays as the HMD beds in.
) marketing label is a joke, of course.