Hi,
Several apps I have tried (maybe ~10% of them) only have portrait mode. This is a problem if you have a TV you can't rotate.
I wonder if anyone has any ideas of how to write, say, a background service that detects if it's displaying in portrait mode, and then shrinks the screen view and flips it 90degrees (so essentially you have a portrait screen that is now the correct way up, with black screen on either side). You can't tell it to display in true landscape because that's hard-coded into the app, so it is better to just rotate the portrait display. However this is a hard thing to do as the screen display is very low-level... but does anyone have any ideas of how to accomplish that?
I've done some searching, but of course Android has never really faced this problem before. Android has either been on a rotatable mobile device, or it's been on a GoogleTV where everything is landscape...
If so, that would open up more apps for Q-compatibility...
Hi, curious question.
I'm wondering why the Nexus7 doesn't allow you to slow scroll/pan down at single pixel steps. Always seems to be in 2-3 pixel jumps. I got a bad habit of scrolling web pages while reading from iOS but on my N7 I get these kinda anoying start/stop jerks.
Is this related to the virtual pixel resolution being lower than the actual screen resolution? And would changing it allow one to match the virtual pixels to the screen pixels and get smoother slow speed panning?
Anyone know if some custom ROMs optimize for this?
BR
POTS101 said:
Hi, curious question.
I'm wondering why the Nexus7 doesn't allow you to slow scroll/pan down at single pixel steps. Always seems to be in 2-3 pixel jumps. I got a bad habit of scrolling web pages while reading from iOS but on my N7 I get these kinda anoying start/stop jerks.
Is this related to the virtual pixel resolution being lower than the actual screen resolution? And would changing it allow one to match the virtual pixels to the screen pixels and get smoother slow speed panning?
Anyone know if some custom ROMs optimize for this?
BR
Click to expand...
Click to collapse
I have been struggling with this issue for days. Looked into kernel code hoping to find anything related to scrolling or touchscreen but nowhere to find such a thing. It has nothing to do with virtual pixel resolution. If you swipe your finger at a degree of between 45-90 (accapted as scroll up by the system), Y coordinates change by 3 pixels where as X coordinates can change by 1.. Therefore its not a limitation, it is kind of scroll sensitivity which is 3px.. I've seen other devices can scroll by 1 pixel.
I believe google has set this value to 3 because of rendering issues, since its gonna render the screen 3 times more if it is set to 1. Maybe it is the optimum sensitivity. Lower than 3 pixels may disrupt buttery smoothness but its just a guess..
I hope one dev can lead us the right piece of code where it is set to 3 and we can change it ourselves to test the best..
You guys can try the problem yourself actually by installing the "Multi-touch accuracy check" app from market, which shows the exact pixel of the touch..
the xy coordinates readout from the developer "show touches" setting shows the 3pixel increments too.
when panning something just a bit faster than dead slow then suddenly gets rid of the jumps and goes smoother as well.
could it be a battery saving thing that they just get rid of taxing the cpu/gpu when people like me do these constant slow scroll pans on webpages and books etc.?
POTS101 said:
the xy coordinates readout from the developer "show touches" setting shows the 3pixel increments too.
when panning something just a bit faster than dead slow then suddenly gets rid of the jumps and goes smoother as well.
could it be a battery saving thing that they just get rid of taxing the cpu/gpu when people like me do these constant slow scroll pans on webpages and books etc.?
Click to expand...
Click to collapse
2 more thoughts:
a) this relatively coarse sensitivity can lead sometimes (not often) lead to a +/-3 pixel uncertanty in a UI element and cause it to jittter - sometimes happens when finger pressure changes.
b) I had for a while thought that it may be linked to the size of the touch sensors on the display. Being a lower cost device I figured that the HW resolution may be lower than pixel count. But I'm pretty sure that you must be always hitting multiple sensors and the system should just interpolate to a finer resolution.
POTS101 said:
the xy coordinates readout from the developer "show touches" setting shows the 3pixel increments too.
when panning something just a bit faster than dead slow then suddenly gets rid of the jumps and goes smoother as well.
could it be a battery saving thing that they just get rid of taxing the cpu/gpu when people like me do these constant slow scroll pans on webpages and books etc.?
Click to expand...
Click to collapse
When scrolled a little bit faster, it still increases by 3px , not 1.. I also assume that it is about battery saving but if I can find a way to decrease the step size, I'd love to try it by myself..
Also It can be related to the sensors , as you mentioned, but not likely.. It can catch 1px changes when swiped horizontally..
Any news on this?
As far as I can see, this is only related to touch screen; if you use a usb mouse, you can actually see the scroll by one pixel at time, and there is no graphical issues. With mouse all seems smoother...
At this point, I think It's an hardware issue of the touch screen, as the system can easly support 1px scrolling
aLPaSLaN07 said:
I have been struggling with this issue for days. Looked into kernel code hoping to find anything related to scrolling or touchscreen but nowhere to find such a thing. It has nothing to do with virtual pixel resolution. If you swipe your finger at a degree of between 45-90 (accapted as scroll up by the system), Y coordinates change by 3 pixels where as X coordinates can change by 1.. Therefore its not a limitation, it is kind of scroll sensitivity which is 3px.. I've seen other devices can scroll by 1 pixel.
I believe google has set this value to 3 because of rendering issues, since its gonna render the screen 3 times more if it is set to 1. Maybe it is the optimum sensitivity. Lower than 3 pixels may disrupt buttery smoothness but its just a guess..
I hope one dev can lead us the right piece of code where it is set to 3 and we can change it ourselves to test the best..
You guys can try the problem yourself actually by installing the "Multi-touch accuracy check" app from market, which shows the exact pixel of the touch..
Click to expand...
Click to collapse
My two cents...I tried this approach and when you move your finger with about 5 degree you can get 1 pixel change by Y and 3 pixels change by X. And if you move your finder with about 85 degree you get 1 pixel by X and 3 pixels by Y. It seems that hardware can handle the 1 pixel change by X or Y. It looks like the touch events are throttled by the system based on the distance between points.
There is no decent and satisfying answer to this yet .. I'm still waiting for a solution..
I can't find a reason about why this can't be possible. I want to know if there's a way to take landscape photos in portrait mode (holding the device vertical) on S3.
I was testing this program "Shot Control" it gives you some controllers in the screen, you have half of the screen to see the photo preview, considering the size of the S3 screen it was not a bad layout, but I'll love to have the same possibility holding the device in portrait (vertically) having the preview on top, controls down and taking pictures in landscape.
Is there a program to allow this?
I search over and I saw some solutions for iphone but pretty much nothing for us, here's a related topic:
http://forum.xda-developers.com/showthread.php?t=849118
Anyone?
I have device (RSH A400, not Tronsmart) which I loaded with Tronsmart 316k4 firmware. Almost everything works like it should with one exception - portrait oriented application.
When I try to run portrait oriented application (without any support for landscape) on this firmware, if I disallow application to change resolution from landscape to portrait, it runs in landscape with screen stretched to fill whole 1920x1080 and looks very bad and is literally unusable.
I have done some investigations (switching beetween different firmwares released in different times) and i now know that it is not portrait / landscape issue but aspect ratio issue. On older firmwares (as well on my factory firmware) if application needs to run in portrait mode (no alternative for landscape mode) but mode is locked to landscape and not changable, the device maintain proper aspect ratio by itself (screen height is set to 1080, width calculated and set to about 607.5, screen centered). It looks like app is working in portrait mode without rotating screen but in fact it is working in landscape with maintained aspect ratio.
I need to fix this. I gave 316k4 as example because it is quite popular firmware, but I had the same behaviour on Probox2 EX and few other, newer firmwares.
It can be some bad setting on kernel level, or something needs to be modified in build.prop or any other config file. I don't see any suspects in build.prop (checked the differences) so the change must be done somewhere else or this behaviour is nonfixable at all. I can't use portrait mode because screen is rotated for 90 degrees then and my monitor doesn't have pivot and rotating screen.
Do you guys know what to change in device configuration to fix this?
When I am playing a certain game (Dragalia Lost), I often notice performance issues when there are significant amounts of particles on screen. However, going into YouTube, starting a video, and allowing it to enter PIP mode and returning to Dragalia seems to alleviate a significant amount of stutter, returning it to a full 60-fps framerate. What does PIP force the phone to do that allows it to run smoother? Does it force the game to render things through a different means?