Sunday, May 20, 2018

Razer Phone XDA Display Analysis: A Great Start for 120Hz Displays on Android

When contemplating who'd be a major player in the Android smartphone business, the gaming hardware giant Razer probably doesn't come to mind. While they have yet to establish themselves as a reliable smartphone provider, Razer's first attempt did not at all seem like it was their first time dabbling into Android, likely because much of their engineering team came from Nextbit. Razer leveraged their status in gaming hardware to appeal to those who game, and those who game hold high refresh rate monitors in high regard. So Razer put one on a smartphone.


Technology

The Razer Phone boasts a fluid 5.7-inch 120Hz IGZO-IPS display with 2560×1440 pixels in a 16:9 aspect ratio, with each pixel arranged in a typical striped RGB subpixel pattern, a concept we're sure Razer is very familiar with.

With its resolution and subpixel pattern at its screen size, the display of the Razer Phone appears among the sharpest with unresolvable pixels when viewed further than 6.7 inches, which is much closer than typical smartphone viewing distances, for normal 20/20 vision. However, the display is not ideal for virtual reality (VR)  use (nor is it Daydream-certified) as its RGB stripe subpixel pattern results in a pronounced screen-door effect; Diamond PenTile is the desirable subpixel pattern for VR at the same resolution due to its smoothing characteristic.

The Qualcomm Snapdragon 835 improves upon the display processing unit compared to its predecessors, which now supports native 10-bit color depth and native wide color gamut. Razer implements these additions with Netflix HDR support and with automatic color management, which was introduced to Android in 8.0. The 835 also introduces Qualcomm's own dynamic refresh rate solution, named Q-Sync, similar to NVidia's G-Sync and AMD's FreeSync, which are technologies that match the display refresh rate with the active GPU rendering framerate.

The 120Hz display, which Razer brands as "UltraMotion", results in a much more fluid user experience within the system UI and with supported games and media. Razer is not the first company to include a high refresh rate display on a phone: Sharp introduced their Sharp Aquos Crystal smartphone in 2014, which not only debuted as the first production smartphone with a high refresh rate 120Hz display, but also as one of, if not, the first to begin the "bezel-less" phone trend. Uncoincidentally, the Razer Phone display was also sourced from Sharp. However, Razer Phone does not follow the bezel-less trend and proudly embezzles the device with possibly the best speakers on a smartphone. The Razer Phone also supports a dynamic refresh rate, implemented through Qualcomm's Q-Sync, which synchronizes the refresh rate of the display to the frame rate of the on-screen content, down to 30fps. The dynamic refresh rate allows the Razer Phone to render content smoother than other competitors' displays without a dynamic refresh rate, even at the same content frame rate. For example, if an app drops frames during a flick or an animation, the dynamic refresh rate can adapt to the lagging frame rate to reduce the appearance of frame stutter, which is caused when the active frame rate does not divide wholly into the display refresh rate.

The "UltraMotion" display is made practical with Razer's use of IGZO thin-film transistors, the significance of which is their remarkably low power leakage. The low power leakage allow the transistors to hold their charge longer when being driven than other thin-film transistors, such as the more-commonly used LTPS thin-film transistor found in most modern high-end smartphone LCDs. Since the transistors can maintain their charge longer, they can afford to "skip" some of the drive periods on static content without causing visual artifacts. Theoretically, this saves power by not needing to drive the transistors 120 times a second if the on-screen content doesn't require it, and it allows for the display to be explicitly set to a certain refresh rate.

Razer also employs their own content-adaptive backlight control (CABC) solution in their kernel, which saves battery on devices with LCDs by rendering on-screen color tones with a dimmer backlight, but with higher pixel color intensities, to deliver a perceptually identical image with lower display power consumption.

In their latest Android 8.1 update, the Razer Phone is a new player—and the only other player at the time of this writing that we are aware of, besides Google's Pixel phones—in supporting automatic color management, which was introduced to AOSP in Android 8.0 Oreo. Automatic color management is absolutely fundamental to functional color accuracy, and without it, the color accuracy of the different display profiles of a device (e.g. Samsung's AMOLED Cinema, AMOLED Photo display profiles) become mostly insignificant and impractical except for in a few niche scenarios. Automatic color management puts these dormant calibrations to proper use by applying them when viewing content that calls for the appropriate color space.


Performance Summary

One of the common shortcomings of LCDs is demonstrated immediately at the initial boot sequence, and that is its generally-poor black levels and contrast. The boot animation is composed of a black background that exhibits very visible backlighting. The contrast ratio of the Razer Phone display looks quite ordinary—that is to say, not particularly impressive, especially if coming from an OLED display.

Greeted by the device setup interface, the white point calibration of the display is noticeably cold. Colder white points are a common aesthetic calibration choice to make a display look more fresh, as opposed to warmer white points that tend to be likened to dirtied, aged white surfaces, such as yellowing teeth, yellowing paint, rusting metal, dirty porcelain, etc. Personally, I am not a fan of how cold the white point is calibrated on the Razer Phone; I interpret cold white point calibrations to this degree as looking too "digital", and reminiscent of many older, cheaper displays that are usually calibrated very cold. However, the human visual system is fascinating and can actually adapt to different white balances, given enough time for our cones to adjust. After a while, the white point is tolerable, but the higher amplitude of blue light from the colder color temperature can still cause more strain to the eye.

Starting with the Razer Phone's Android 8.1 update, the default color profile is set to "Boosted", which targets the sRGB color space, with slightly increased saturation. However, this comes with several concerns (which will be covered in detail later on) and I do not advocate for its use. In short, the colors on the "Boosted" color profile are slightly oversaturated with perceptual incongruencies and clipping on blue color mixtures. Razer should reassess its implementation or stick with its "Natural" color profile as the default color profile, which is actually calibrated quite well. The "Natural" color profile still takes on the colder white point, but it still pleasantly reproduces sRGB and P3 content. Colors are saturated nicely with color tones that are very well-lit to the standard gamma of 2.2, and color hues are adequate after chromatic adaptation of the white point. The color profile is also color-managed, which means that content of other color spaces (like P3) should appear correctly in this profile, if the app supports it. The "Vivid" color profile maps all colors, regardless of color space information, to the P3 color space, which is a good option for those that don't mind sacrificing color accuracy for punchier colors all around.

The maximum brightness of the Razer Phone display is an absolute disappointment. It is dimmer than any modern flagship smartphone, and even dimmer than most modern budget smartphones. This is confounding, as one of the key characteristics of IGZO thin-film transistors is their transparency, which allows more of the backlight to pass through. Electron mobility, refresh rate, and brightness should all be unrelated factors on their own—in fact, the higher refresh rate should make the display seem brighter at the same drive voltage due to the faster modulation. The brightness, along with black levels, ultimately comes down to panel quality, in which Razer most likely cut (pricey) corners in backlight technology to introduce their still-fantastic 120Hz QHD display.

The display power is also slightly disconcerting. Considering that the Razer Phone display utilizes an IGZO backplane that consists of transistors more translucent than those found in LTPS displays, the Razer Phone has worse display power efficiency than the iPhone 7 LTPS LCD. The dynamic refresh rate, however, does save a marginal amount of display power in addition to the power savings from the fewer frames that the CPU or GPU needs to render.


Methodology

To obtain quantitative color data from the display, we stage device-specific input test patterns on the display and measure the resulting emission from the display using an i1Pro 2 spectrophotometer. The test patterns and device settings we use are corrected for various display characteristics and potential software implementations that can alter our desired measurements. Many other sites' display analyses do not properly account for them, and consequently their data is inaccurate.

We measure the grayscale in steps of 5%, from 0% (black) to 100% (white). We report the perceptual color error of white, along with the average correlated color temperature of the display. From the readings, we also derive the perceptual display gamma using a least-squares fit on the experimental gamma values of each step. This gamma value is more meaningful and true-to-experience than those that report the gamma reading from display calibration software like CalMan, which averages the experimental gamma of each step instead for calibration data.

The colors that we target for our test patterns are derived from DisplayMate's absolute color accuracy plots, which are spaced roughly evenly throughout the CIE 1976 chromaticity scale, making them good targets to assess the complete color reproduction capabilities of a display.

We will primarily be using the color difference measurement CIEDE2000 (shortened to ΔE), compensated for luminance error, as a metric for chromatic accuracy. CIEDE2000 is the industry standard color difference metric proposed by the International Commission on Illumination (CIE) that best describes perceptually-uniform differences between color. Other color difference metrics exist as well, such as the color difference Δu′v′ on the CIE 1976 chromaticity scale, but these metrics are inferior in perceptual uniformity when assessing for visual noticeability, as the threshold for visual noticeability between measured colors and target colors can vary wildly. For example, a color difference Δu′v′ of 0.010 is not visually noticeable for blue, but the same measured color difference for yellow is noticeable at a glance.

CIEDE2000 normally considers luminance error in its computation, since luminance is a necessary component to completely describe color. Including luminance error in ΔE is helpful for calibrating a display to a specific brightness, but its aggregate value should not be used for assessing display performance; for that, chromaticity and luminance should be measured independently. This is because the human visual system interprets chromaticity and luminance separately.

In general, when the measured color difference ΔE is above 3.0, the color difference can be visually noticed at a glance. When the measured color difference ΔE is between 1.0 and 2.3, the difference in color can only be noticed in diagnostic conditions (e.g. when the measured color and target color appear right next to the other on the display being measured), otherwise the color difference is not visually noticeable and appears accurate. A measured color difference ΔE of 1.0 or less is said to be imperceptible, and the measured color appears indistinguishable from the target color even when adjacent to it.

Display power consumption is measured by the slope of the linear regression between device battery drain and display brightness. Battery drain is observed and averaged over three minutes at 20% steps of brightness, and trialled multiple times, while minimizing external sources of battery drain. To measure the display power consumption difference due to refresh rate, we measure device power drain at the different refresh rates instead.


Brightness

Our display brightness comparison charts compares the maximum display brightness of the Razer Phone relative to other smartphone displays that we have measured. The labels for the horizontal axis on the bottom of the chart represent the multipliers for the difference in perceived brightness relative to the Razer Phone display, which we fixed at "1×". The values are logarithmically scaled according to Steven's Power Law using the exponent for the perceived brightness of a point source, scaled proportionally to the maximum brightness of the Razer Phone display. This is done because the human eye has a logarithmic response to perceived brightness. Other charts that presents brightness values on a linear scale do not properly represent the difference in perceived brightness of the displays.

Razer Phone display brightness comparison chart: 100% APL

Razer Phone display brightness comparison chart: 50% APL

Razer most likely had to cut costs somewhere to be able to package an affordable QHD, wide-gamut high dynamic refresh rate display in a smartphone, and unfortunately that cut was most likely in the backlight. Increasing the brightness of a display is very cost-inefficient, as the increase in perceived brightness runs into some serious diminishing returns. This is because the perceived brightness of a display scales logarithmically. For example, doubling the backlight emission from 400 cd/m² to 800 cd/m² doesn't double the perceived brightness of the display, but only increases it by about 25%. The manufacturer has to pay for double the emission, while it perceptually only increases it by a quarter, and furthermore, it still requires double the power. If corners had to be cut, the backlight would be the reasonable place to start.

Measured with our spectrophotometer, the Razer Phone display reaches a maximum brightness of 415 cd/m² displaying a full-white canvas. This is very dim for a smartphone LCD in this generation. Flagship LCDs are usually much brighter than OLED displays at 100% APL, but in our measurements the Razer Phone display is even dimmer than all our OLED displays at 100% APL, except for the Google Pixel XL. The Pixel XL, however, pulls ahead in brightness at 50% APL, at which the Razer Phone is marginally dimmer than the rest. Because of its dim maximum brightness, the Razer Phone display isn't fit for comfortable outdoor viewing. This truly seems to fulfill the "gaming phone" niche, which has no business not being indoors.


Gamma

The gamma of a display determines the overall contrast and lightness of the colors on the screen. The industry standard gamma for most displays follows a power function of 2.20. Higher display gamma powers will result in higher image contrast and darker color mixtures, which the film industry is progressing towards, but smartphones are viewed in many different lighting conditions where higher gamma powers are not appropriate. Our gamma plot below is a log-log representation of a color's lightness as seen on the Razer Phone display vs. its associated input color: Higher than the Standard 2.20 line means the color tone appears brighter, and lower than the Standard 2.20 line means the color tone appears darker. The axes are scaled logarithmically since the human eye has a logarithmic response to perceived brightness.

Razer Phone gamma plot

The Razer Phone display gamma just straddles the 2.20 Standard line, which is reflected by the display's excellent color tone reproduction. Most modern IPS displays achieve similar levels of tonal accuracy, and while it would be much more impressive (and difficult) to see this achieved on an OLED panel, it is still commendable to see Razer land right on 2.20 for the resulting display gamma. The Razer Phone display also has an excellent static contrast ratio of 2071:1, which is at the higher end for smartphone LCDs.


Display Profiles

A device can come in a variety of different display profiles that can change the characteristics of the colors on the screen.

The Razer Phone comes with three color profiles: Natural, Boosted, and Vivid.

Razer Phone display profiles

The "Natural" color profile is color-managed and targets the good ol' sRGB color space. The white point is intentionally set colder than D65.

The "Boosted" color profile is set as the default on the Razer Phone. It is also color-managed, targets the sRGB color space, and has a colder white point, but it expands its gamut by 10% with respect to the CIE 1931 color space. Just as I mentioned in my Pixel 2 XL display analysis, this color profile comes with some caveats.

The first issue I'd like to point out is that the color space expansion of the "Boosted" color profile is relative to the CIE 1931 color space instead of the later CIE 1976 color space, which "represents the most uniform colour space for light sources recommended by the CIE." Although it is not perfect, using the CIE 1976 chromaticity scale as the reference for the expansion would yield a more perceptually-uniform increase in saturation.

Another issue with the "Boosted" color profile is that, on the Razer Phone, the red and green primary chromaticities are indeed expanded, but the blue primary chromaticity is identical to that in the "Natural" (and "Vivid") color profile. This could be a calibration oversight by Razer or a hardware limitation of the display, depending on the true native gamut of the panel. Even though the blue primary remains intact, the "Boosted" color profile still increases the saturation of all other blue color mixtures. This causes clipping for higher-saturation blue color mixtures, making them appear indistinguishable.

Close-up of blue color plots: "Boosted" colors (right) show slight color expansion, except for blue primary (tip) which does not change.

The "Vivid" color profile maps all color values to the P3 color space, and is not color managed. Like the other two color profiles, it also has a cold white point.


Color Temperature

The average color temperature of a display determines how warm or how cold the colors look on the screen, most noticeably on lighter colors. A white point with a correlated color temperature of 6504K is considered the standard illuminant for the color of white, and is necessary to target for accurate colors. Temperatures higher than 6504K are said to be cold, while temperatures lower than 6504K are warm. Regardless of the target color temperature of a display, ideally the color of white should remain consistent at various tones, which would appear as a straight line in our chart below.

Razer Phone color temperature chart

All the Razer Phone color profiles are much colder than the standard 6504K, each averaging to about 7500k. There is marginal variation in color temperature throughout the different intensities of white, ranging from about 7300k up to the white point at 7700K. Both these factors can greatly affect color accuracy, although chromatic adaptation can help the cold white point in appearing accurate. While we haven't yet measured that many smartphones, the Razer Phone display is the coldest we've measured among displays in what should be their "color-accurate" display mode. We will flesh this out more in the next section.

Display white point color temperature  reference chart

Display average color temperature reference chart


Color Accuracy

Our color accuracy plots provide readers a rough assessment of the color performance and calibration trends of a display. Shown below is the base for the color accuracy targets, plotted on the CIE 1976 chromaticity scale, with the circles representing the target colors.

Reference sRGB color accuracy plots

The target color circles have a radius of 0.004, which is the distance of a just-noticeable color difference between two colors on the chart. Units of just-noticeable color differences are represented as white dots between the target color and the measured color, and one dot or more generally denotes a noticeable color difference. If there are no dots between a measured color and its target color, then the measured color can be safely assumed to appear accurate. If there are one or more white dots between the measured color and its target color, the measured color can still appear accurate depending on its color difference ΔE, which is a better indicator of visual noticeability than the Euclidean distances on the chart.

Razer Phone Natural Profile color accuracy plots: sRGB

Razer Phone Natural Profile color accuracy chart: sRGB

Razer Phone Natural Profile color accuracy plots: P3

Razer Phone Natural Profile color accuracy chart: P3

The Razer Phone display in its "Natural" color profile measures to be mostly inaccurate at a glance, with an average color difference ΔE = 2.8 for sRGB and an average color difference ΔE = 2.7 for P3, both of which are above the 2.3 threshold for accurate colors. The color error can most definitely be attributed to the intentional colder white point calibration. This is a disappointment for a color profile that is supposed to be accurate.

However, there are multiple external factors that can affect the perceived color accuracy of a display. One factor is the color of ambient lighting, which can affect the perceived white point of a display. For example, being in a room with warm tungsten lights can make an "accurate" 6504K white point appear colder than in typical indirect sunlight. However, even with these clashing color temperatures, the human visual system is incredible at correcting for differences in white point, and after spending some time looking at the display, it will be perceived as "perfect white" again (that is, until a more "fitting" white appears). This concept is known as chromatic adaptation, and can help the cold white point of the Razer Phone display to appear accurate in unfitting lighting conditions.

Razer Phone Natural Profile color accuracy plots: sRGB, corrected for white point

After applying a white point color transformation, the Razer Phone can appear perfectly accurate, with a theoretical color difference ΔE = 0.5 after white point correction. This also reveals underlying potential for the Razer Phone to properly calibrate their display, although calibration is not as simple as a color transformation.

Of course, having fine color accuracy after chromatic adaptation doesn't deserve much credit. Chromatic adaptation is an uncomfortable transition for the eye and the calibration ultimately still strays slightly too far from the standard. While the colder white point may have been a design intent, it's an odd choice to supply an otherwise-accurate color profile without providing a way to tweak the color temperature, which should be the minimum acceptable option when straying from the standard this far. The best option is still unique to the Apple devices, and that is their brilliant TrueTone dynamic color temperature solution, which adjusts the color temperature of the display according to the color of the ambient light.

One quirky find is that by searching for "temperature" in the Settings of the Razer Phone, we see an inactive "Cool color temperature" setting that is vestigial from Android N on the Nexus devices. Razer would benefit from having the opposite of this.

The color performance of the "Boosted" and "Vivid" color profiles are not important to analyze, since that is not the goal of their usage. The design flaw of the "Boosted" profile is covered in Display Profiles, in which I recommend not using it. Provided below are additional plots for the "Boosted" and "Vivid" modes along with the device reference charts for display color accuracy.

Display white point accuracy reference chart

Display color accuracy reference chart


Power Consumption

Since the Razer Phone display utilizes an IGZO backplane, we expect marginal power efficiency improvements over displays that use a LTPS backplane. Since this is our first analysis to include measurements for display power, we will use DisplayMate's iPhone 7 Display Analysis as reference for the power consumption of a LTPS LCD.

Measuring the two devices at their peak brightness, we found that the Razer Phone display consumes 1.18 watts, while DisplayMate reports the iPhone 7 display to consume 1.08 watts. The Razer Phone display consumes about 8.5% more power overall at their maximum brightness, but these values do not indicate the efficiency of the display, which is what we are interested in. The Razer Phone has a larger screen area that requires higher backlight emission than the iPhone 7 to reach the same uniform brightness. On the other hand, the iPhone 7 has a considerably-higher peak brightness. Normalizing these factors, the Razer Phone consumes 0.32 watts per candela while the iPhone 7 only consumes 0.29 watts per candela, making the iPhone 7 the more efficient panel by 9.4%. At the efficiency of the iPhone 7 display, it would only take 1.06 watts to power a display of the same screen area and peak brightness as the Razer Phone. Note that refresh rate is not considered in the wattages. This is a conflicting verdict, as we expected the IGZO display to be more efficient than the LTPS display. However, Apple is a veteran in the smartphone business and is exceptionally experienced with displays, so these results are not completely surprising.

Moving on to the refresh rates, we calculated that the display consumes 0.003 watts per Hz, which results in expending 0.09 watts for 30Hz up to 0.36 watts for 120Hz. Recall that the Razer Phone display has a dynamic refresh rate, so for static images it is possible to save up to 0.27 watts, which is a respectable amount. Note that another bulk of the power drain/savings comes from the extra heavy-lifting done by the CPU and GPU to render the additional/fewer frames, which will not be tested for here.


Overview

Specification Razer Phone Notes
Display Type IGZO IPS LCD Acronyms
Display Refresh Rate 30Hz–120Hz Razer Phone has a dynamic high refresh rate
Display Size 5.0 inches by 2.8 inches

5.7 inches diagonally

Display Resolution 2560×1440 pixels RGB stripe subpixel pattern
Display Aspect Ratio 16:9
Pixel Density 515 pixels per inch Subpixel density is identical
Distance for Pixel Acuity <6.7 inches Distances for just-resolvable pixels with 20/20 vision. Typical smartphone viewing distance is about 12 inches
Peak Display Brightness 415 cd/m² Measured at 100% APL
Static Contrast Ratio 2071:1 Ratio of peak brightness to black level
Maximum Display Power 1.18 watts Display power for emission at peak brightness
Refresh Rate Power 0.09 watts for 30Hz/static image

0.18 watts for 60Hz

0.27 watts for 90Hz

0.32 watts for 120Hz

Power consumption for dynamic refresh rate
Display Power Efficiency 0.32 watts per candela Normalizes brightness and screen area

 

Specification Natural Boosted Vivid Notes
Gamma 2.20 2.19 2.21 Ideally between 2.20–2.40
Temperature of White 7670K

Colder by design

7684K

Colder by design

7702K

Colder by design

Standard is 6504K
Color Difference of White ΔE = 7.3 ΔE = 7.4 ΔE = 7.5 Ideally below 2.3
Average Correlated Color Temperature 7470K

Colder by design

7498K

Colder by design

7471K

Colder by design

Standard is 6504K
Average Color Difference ΔE = 2.8

for sRGB

ΔE = 2.7

for P3 color space

ΔE = 3.4

for sRGB

ΔE = 2.9

for P3 color space

ΔE = 3.2

for sRGB


Not color managed; oversaturated by design

Ideally below 2.3
Maximum Color Difference ΔE = 5.4

at 25% cyan

for sRGB

ΔE = 5.8

at 25% yellow

for P3

ΔE = 5.8

at 100% cyan-blue

for sRGB

ΔE = 5.2

at 25% cyan

for P3

ΔE = 5.4

at 25% cyan

For sRGB

Ideally below 5.0

Razer Phone Final Thoughts

For Razer's first smartphone, they show magnificent effort and seem extraordinarily involved, implementing some fundamental options and special feats that most OEMs have yet to touch on. The dynamic high refresh rate panel is an absolute joy to use, and paired with its smooth OS, the Razer Phone serves the most fluid-feeling interactive Android interface experience on a phone. However, most people that have set foot outdoors will find the maximum display brightness completely unacceptable. On top of its poor brightness performance, its display power performs relatively inefficiently for having transparent IGZO thin-film transistors, although it does save a decent amount of power on static content from its dynamic refresh rate.  The color performance is also not great, but it's not absolutely terrible. Lastly, the cold white point of the display is sure to throw off its users' circadian rhythm—in fact, that's probably why the Razer Phone display is calibrated that way: to keep them deprived of sleep, keeping gamers focused on every single one of those frames.


Visit the Razer Phone Forums on XDA



from xda-developers https://ift.tt/2rWlBhI
via IFTTT

No comments:

Post a Comment