This was measured using the battery status indicator in Android, with periods of ten-twenty minutes and the results were extrapolated. This is not at all an accurate way to measure power consumption. You need to use multimeter and measure real current going to the screen. Or for "real world" battery life results, you'd need to let the battery recharge and then drain all the way, not let it run for 20 minutes, then use a very inaccurate battery status indicator and finally extrapolate.
xda-developer forums seem to be full of tips and tricks that make your Android device battery last longer or the device work faster, but a lot of that stuff is just crap. This article was better than average, but the results are still very convincing.
What has always puzzled me is that I thought Google made stock Android pretty dark-themed because they wanted to help manufacturers that used AMOLED screens, specifically Samsung, who makes like 95% of AMOLED's - and yet Samsung has chosen to use a very light-themed Touchwiz skin. It makes no sense. You'd think they would be the first to want to take advantage of a darker theme with their Super AMOLED flagship devices with large screens.
Maybe, having the real data on power usage from the components instead of a script reading a non linear, corrected and unreliable indicator, they know it doesn't matter as much as XDA thinks.
there seems to be a lot of confusion over the efficiency of amoled vs lcd.
In lcd, the power consumed directly depends upon the brightness, basically LED's provide the back-light light source and the colors you see on the screen are the representative of the kind of filters the light from the LCD had to cross to reach the surface. All lcd's are in this way passive displays.
In an AMOLED display, there are three sets of actual LED's per pixel visible on screen. Each of the led making up a pixel can be individually be turned on or off, resulting in a display color, resulting in the infinite contrast and brighter colors. The current technology restricts the brightness of each of those tiny led's resulting in a lower brightness, and also the individual tiny led's brightness varies over the life of the panel. this is the biggest shortcoming for an AMOLED.
When you see a picture of light on an AMOLED screen, it is actually true that a physical light is being prepared in front of your eyes on the screen by adding the three colors in proportion. this is not an easy task and so explains why lcd's as of now display truer colors. This truly is a modern technical achievement.
OLEDs have larger color gamut, but larger is not always better.
“The Color Gamut is not only much larger than the Standard Color Gamut, which leads to distorted and exaggerated colors, but the Color Gamut is quite lopsided, with Green being a lot more saturated than Red or Blue, which adds a Green color caste to many images. Samsung has not bothered to correct or calibrate their display colors to bring them into closer agreement with the Standard sRGB / Rec.709 Color Gamut, so many images appear over saturated and gaudy.” – http://www.displaymate.com/Smartphone_ShootOut_2.htm
what i meant to explain was, this.. the LED's fade over time. hence their color changes over time. and making two LED's display the same color and making them time equally is a very challenging job, some thing which current day technology is unable to do. yes, led's have a bigger color gamut, but the accuracy is definitely to b improved. gamut is the range, while accuracy is the exact color required i.e precision
Spoilers: a very impressive +3 hours of screen time by using dark themes instead of bright ones on SAMOLED.
Minor tangent: I'm not sure how he defends his contention that SAMOLED > LCD by noting that he didn't test anything but SAMOLED in the comments. I realize he wasn't testing anything else and I'm perfectly willing to believe that SAMOLED is more efficient but he still claimed X and when asked for evidence punted.
Sadly, the Android Modding community is usually short on throughly testing findings in a scientific way that goes beyond anecdotal evidence. When you inquire about such things or comment, you either get indifference or at worse, hostility from those that start to think you're better than others for wanting to find out the truth from statements. Some users start to confuse curosity and questioning with outright hostile interrogation and lay on comments like "if you don't like what this guy is doing, then don't use it or read it." Sure there are trolls that really do cause problems, but those that honestly want real answers or to help get thrown in with them at times.
Granted that it's not all like this as some groups out there and developers really do care about such things (like Cyanogenmod). However, that's typically not the mode of operation. For example, kernel modders that think it is a good idea to disable fsync by default. On top of that, they generally do not give an option to enable it for end users. Result was lots of data corruption when people pulled out their batteries.
This is just my observations from being heavily into the modding community for around 2+ years helping to administer one of the larger modding forums. I try to push for factual information as much as I can and to lead by example when possible through any code or mods I might publish, but the community is pretty set in its ways still. I do think my efforts aren't in vain when I get the occasional positive feedback from users, so it keeps me going. However, there are just too many non-developers mixed in with a minority of actual developers and end users cannot seem to discern between them.
>I simply timed how long it took the battery to drain 5% with the screen always on at various brightness levels and while viewing an all white or all black 720p image fullscreen in QuickPic.
This is absolutely best case scenario, ignoring all other battery factors.
Best case scenarios for time, but honestly thats not what is important in these stats. Rather its the % difference between the two. The total life of a cell phone will always be different between devices/users, the % difference solely from the bright/dark difference should stay generally the same. Assuming most of the time is spent going through the UI(which is doubtful tho).
did he run it multiple times? was the battery recharged between each test? how did he make sure the 5% were actually always the same 5% and comparable?
Amongst OLED and LCD displays with everything else (size, resolution, etc.) OLEDs intrinsically consume a lot more power than LCDs. Using dark themes help for OLEDs for reasons already stated here, and from what I have seen working on the display industry for several years, the power numbers quoted by OLED display manufactures often presume a dark theme when comparing power to LCDs.
I remember these findings being refuted, and I thought it was because it took the same amount of power for an LCD screen to show a black pixel instead of a white pixel.
I will say, given these findings from OP, I hope these types of displays become more common.
It takes more energy for an LCD to show a black pixel. It is not much compared to the backlight, though. On one of my laptops solid black uses 0.5 watts more than solid white.
Lowering the brightness? Sure. Switching to a dark theme? Not so much.
If you are using a LCD the color of your pixels doesn't matter. Ok, it might matter a little bit but not much and it's not as easy as saying that white uses more power than black. It might even be the other way around.
In an LCD the color is determined by a liquid cristal filter that can be turned on or off and dimmed. The filter itself needs hardly any power, however. Not compared to the actual light that shines through that filter.
The light comes from LEDs or CFLs (I guess in principle you can use any light source but those two have useful form factors and need less power than, say, incandescent lightbulbs) behind that filter. And that light is always on. Some TVs have local dimming to make the blacks blacker in certain regions of the image, but computer monitors don't.
So by using a dark theme, all you influence is the filter, not the actual lights. As you might know, covering up a lightbulb so you don't see it's light anymore doesn't magically make the lightbulb consume no more power.
AMOLED displays work differently. There the pixels themselves light up or don't. They are like a matrix of thousands of minuscule lightbulbs you can individually turn on or off.
While there's definitely a difference, most likely because of dynamic contrast which might be negligible once you lower the brightness, the reason I switch to a dark theme is because low brightness white looks awful. Especially in a dark room. The monitor I'm using draws around 120 watts so lowering the brightness is a must.
xda-developer forums seem to be full of tips and tricks that make your Android device battery last longer or the device work faster, but a lot of that stuff is just crap. This article was better than average, but the results are still very convincing.