When Samsung unveiled the Galaxy S5, some people (yes, including me) were a bit disappointing by the display. It seems that it's basically the display used in the GS4, only slightly larger (5.1" vs 4.99"). But it turns out that Samsung does include a couple of new display related technologies in the GS5.
According to PC Magazine, the GS5 includes a custom image chip that dynamically adjusts the color gamut and contrast based on ambient light. This makes the colors much better under different lighting conditions. And it's bound to make DisplayMate's Raymond Soneira really happy - he's been long talking about how OLED's wide color gamut can be used in ambient light conditions to improve the image quality.
The GS5 also has a new software feature - when the battery is low (10%), it switches to an ultra power-saving mode which changes the screen to monochrome, lowers the brightness and closes all "non essential" features. In this low-power mode, the GS5 can last up to 24 hours in standby on that 10% battery.
So we didn't get a WQHD (2560 x 1440) AMOLED or a flexible OLED screen, but the GS5 display may still prove to be much better than the one used on the GS4. We'll have to wait to read some reviews to see if this new chip really works.
Comments
I don't think this is a new sensor. It's a new chip that takes the data from the sensor and calculates the best color balance. It seems like a pretty simple calculation, and i'm not sure why it isn't done in software. Perhaps it needs to be in real time, or it saves power that way or else it's indeed a rather heavy calculation which would slowed down the general CPU.
Ah ok, I misunderstood the idea.
In that case it might make sense: While the calculation should be pretty straight forward it is a process that will be running constantly. Over time the power consumption adds up and as they are fighting for every mA it might indeed make sense to use a dedicated hardware for this.
You're right in guessing that power and performance are the main benenfits of doing this type of simple calculation via specialized hardware.
To modify the framebuffer each frame, although a simple calculation, would require quite a fair bit of work for the CPU or GPU, and would require quite a lot of bandwidth as the framebuffer is read, modified, and written back to memory. These operations would significantly increase power consumption as they would be required every frame and cost performance. Additionally, in software the CPU/GPU workload would also have a direct effect on the performance! A very small controller that sits between the SoC and the display can make these modifications extremely quickly and at extremely low power and which operates regardless of the CPU/GPU workload.
This strategy is often used for mobile video as well. Custom circuits or chips are used to decode popular video codecs, which can be done at very low power and freqency compared to doing it in software. Without these solutions, decoding high definition video would be much more computationally expensive and power hungry. DSPs (like Snapdragon's Hexagon unit) are programmable and also well suited for these tasks, and extensions like ARMs NEON can do this as well. But custom hardware generally has better performance and more predictable power consumption. It is common to see dedicated H.264 hardware and in some cases VP8 hardware on a mobile SoC. The implementations take up very little die-area, they run at low clocks, and consume very little power.
One of the aspects of AMOLED screens that I love, is that they consume power proportional to the number of pixels that are lit and how bright they are. It's great that Samsung is exploiting this and has included a specialized "power saving mode" that makes the screen mostly black to conserve power. They are also probably putting limits on the number of usable core and the clocks of the CPU as well when in this mode.
It would also be great to see an always-on lock-screen that displayed time/notifications/etc. With really thin, grey fonts and a small area of the screen utilized, it may be possible to do this and still get great battery life. The benefit is not needing to hit a button to check the time. Additionally, the Snapdragon has some low power processing modes that may help in this regard as well.
This "custom image chip" is a bit of a headscratcher for me. Not the function (which is great), but the requirement for a new sensor. Didn't the older Galaxy S models already feature RGB sensors in combination with an ambient light sensor? And shouldn't this already be enough to collect all the sensor data needed for such adjustments?
Seems to me all they would need is the neccessary software feedback-control. Why would they need a competely new sensor for this?
Or did I miss something about how exactly this is supposed to work?