Friday, August 19, 2016

Carbon LP 5.1.1 for Galaxy Player 5 Update 2

I purchased my Samsung Galaxy Player 5.0 new at best buy in late 2011. Ever since then I've been working on it. Most of you might not remember this but I released the first custom ROM for this device. It was called "FOREIGNER ROM". It was little more that a stock 2.3.6 modified with what ever hack's I could figure out; patching and slicing stuff together. I was a complete n00b. Afaik, every custom rom that was not stock or modified stock has had some significant issues. 1, the camera has never worked right. 2, the magnetic sensor has never worked. With this release of Carbon LP 5.1.1, update 1 and this update, we take a couple of steps forward in having this device completely working

In ICS and JB the rear camera photo snap was not working at all. The dev's that brought up and worked on those ROM's did an excellent job. Bringing up a 3.0 kernel base on a kernel form a different device could not have been easy. But the camera photo snap did not work producing all green pictures. In development of KitKat, I found a way to give the rear camera photo snap some limit functionality. Basically I peeled an image off the preview window just as the front camera does but the resolution was limited to the preview window resolution; 800x840. Although the camera is still not working completely as it should, by adjusting kernel memory allocation for fimc and jpeg drivers, I was able to increase the resolution all the way up to 3MP the default for the US/INT versions of the YP-G70. I never expected to be able to do this; there are just so many moving parts to adjust to make this happen. However, although the images are 'adequate' they are still not what the image sensor is capable of and in addition, the exif data is incorrect; specifically the flash, exposure time, focal length,iso, white balance(?) and aperture fields are derived from incorrect data. From time to time I take a stab at this and every time I learn more so perhaps one day I will have this completely corrected.

I have recently had a few breakthroughs in my understanding of how sensors work in Android. Those breakthroughs have allowed me to write a hw sensors library for venturi based on the work I did for codinalte. I had to calibrate the magnetic sensor manually but as far as I can tell, the compass works perfectly. I used codinalte with the stock sensors lib to calibrate and use as a comparison for the compass. When laid flat and parallel to each other, they give very similar cardinal directions. I have not, however, tested the compass with any navigation apps or even google maps, I'm running without gapps. Performance seems greatly improved without them. I think that apps are supposed to adjust the cardinal degrees when the device orientation changes because at the hw lib level, there is know way to know if orientation is locked or not.

Please let me know if you experience some situation where the compass does not seem to behave as it should...

Update next to the build download