Last edited by a moderator:
*spin-off* Personal Aspect Ratio Issues
- Thread starter zed
- Start date
D
Deleted member 13524
Guest
since my phones res (1080x1920) is greater than the videos quality eg 640x286 Im not missing out in any detail since 1080>640
picture size is governed by distance from eyes
I asked this here before but got no answers, how much does it save? Surely someones done the benchmarks
On an OLED display, it will save something. On an LCD, you'll save a tiny bit in transmission power and charge/discharge of pixels, but the LCD shutter themselves will be working just a hard to block that backlight behind them. So it's going to be in the noise.
Cheers but reguardless of display type theres gonna be a difference in number of pixels rendered in the film.
More pixels rendered = more GPU/CPU used = larger drain on battery.
I was wondering if this has been measured. Normally I can watch 4/5 films on a battery if I switch to holding the thing to horizontal (which I wont cause its a PITA for any great length of time) how many minutes less can I view 10minutes? 30 minutes?
Surely this has been benchmarked somewhere
More pixels rendered = more GPU/CPU used = larger drain on battery.
I was wondering if this has been measured. Normally I can watch 4/5 films on a battery if I switch to holding the thing to horizontal (which I wont cause its a PITA for any great length of time) how many minutes less can I view 10minutes? 30 minutes?
Surely this has been benchmarked somewhere
$100 saiz its measurable (the question is how much) my bank account is .....
How did they survive back in the old days with VHS (similar resolution, well actually a bit better than VHS) and smaller screens relative to area of FOV.
When you're in a bus/plane whatever youre not gonna have the best viewing experience thus have to compromise anyways
The effort for HW decoding is identical. The effort for scaling to the right factor is slightly different, the heaviest cost probably being memory bandwidth (but only if the scaling result goes back to memory, if not then the difference is even less.) The effort to send pixels to the LCD depends on the interface. If it's an interface that has some kind of encoding to ensure zero DC or some kind of copy protection, then the black pixel won't help you, but I don't think that's the case a lot. So you save a little bit there.
The black pixel don't need to be recharged as much.
The video decoding and the backlight will probably overwhelm everything else.
The black pixel don't need to be recharged as much.
The video decoding and the backlight will probably overwhelm everything else.
$100 saiz its measurable (the question is how much) my bank account is .....
How did they survive back in the old days with VHS (similar resolution, well actually a bit better than VHS) and smaller screens relative to area of FOV.
When you're in a bus/plane whatever youre not gonna have the best viewing experience thus have to compromise anyways
Well, VHS was pretty terrible, but that was all we had. Now, we don't have to endure that anymore. In a bus I'd rather read a book and wait until I'm in front of a decent 1080p display to watch movies, but I guess not everyone cares.
>>The effort for HW decoding is identical.<<
I assume so yes
but WRT the other stuff you're saying, on my PC say I render into a 640x480 sized viewport and then let the hardware rescale this to eg 1024x768 vs 1536x1152 I know my GFX chip works harder(higher load) rescaling to the larger size, Im pretty sure its the same with mobile devices right? My question is how much difference
I assume so yes
but WRT the other stuff you're saying, on my PC say I render into a 640x480 sized viewport and then let the hardware rescale this to eg 1024x768 vs 1536x1152 I know my GFX chip works harder(higher load) rescaling to the larger size, Im pretty sure its the same with mobile devices right? My question is how much difference
I get motion sick very quick trying to read in a moving vehicle (planes are OK) watching films is OK in moving vehicles though.
I assume this is common for motion sickness, i.e. reading is a lot worse?
So you dont watch stuff on youtube etc where the majority of content aint 1080p (& if its 1080p its typically encoded crap) but you said yourself watching films on tablets may be some ppl's most used thing, but surely thats an inferior experience than at home in front of there huge tv with surround sound
Scaling is a pretty trivial operation:
- you need something that feeds pixels
- a number of line buffers that are each as wide as min(outgoing image width, incoming image width) and as many as the number of vertical taps-1
- a matrix of multipliers connected to a set of coefficients
That's a bit of math (say 3x5x5 MADDs per pixel), but these days really not a big deal. At 1920x1080 and 60Hz, that's only 9G MADD/s, and those are not floating point but 8 or 10 bits wide color component inputs and maybe 6 bits for the coefficients.
That's nothing compared to the FP32 shader power of GPUs in mobile SOCs so I can't imagine this scaling consume a whole lot. Measurable, yes, if you're power meter is accurate to 10mW or so. But how many minutes of a difference is that going to give when the backlight is on?
- you need something that feeds pixels
- a number of line buffers that are each as wide as min(outgoing image width, incoming image width) and as many as the number of vertical taps-1
- a matrix of multipliers connected to a set of coefficients
That's a bit of math (say 3x5x5 MADDs per pixel), but these days really not a big deal. At 1920x1080 and 60Hz, that's only 9G MADD/s, and those are not floating point but 8 or 10 bits wide color component inputs and maybe 6 bits for the coefficients.
That's nothing compared to the FP32 shader power of GPUs in mobile SOCs so I can't imagine this scaling consume a whole lot. Measurable, yes, if you're power meter is accurate to 10mW or so. But how many minutes of a difference is that going to give when the backlight is on?
I don't know about the specific cases mentioned earlier in the posts but this paper suggests that the backlight (on a tested android device)
Given that a dedicated HW video decoder probably uses much less power (e.g. IMG's which decodes "H.264HP 1080P30 video in under 15mw (40LP)") it seems that the backlight is the real glutton in the system.
In the context of this discussion, I look at the video decoder as everything that's resolution independent, so everything up to where there's a need for a scaler. Not only the HW decoder itself, but also the power needed to feed the decoder with data (CPU managing the whole process, reading flash, streaming, storing data in memory, decryption, ...)
That makes the scaler, resolution dependent, part even more irrelevant compared to everything else.
That makes the scaler, resolution dependent, part even more irrelevant compared to everything else.