Advice, developing a high end video compression codec on Shield Android TV for Camera Acquisition and HQ video.
Hi
Aims
I am researching doing a high end streamlined video compression/decompression codec that can be installed and registered under Android, and be available to third party camera, editing and video apps.* Shield seems like a good top end development target.* I am hoping it will be able to compress 4k+ video streams, with small file sizes and reduced processing overheads.
Even though it is meant to be more for high end camera acquisition on Android in general, it also has other uses on the web.
I am trying to find out general, and detailed, information to see what I need to address.* I'm a newbie to all this, from back in the days that C++ was new and untaught in my college. I'm going to have to reteach myself programming, but have a lot of knowledge on the design side due to previous work.
Codec Programming?
So basically, I need advice on broad programming info on programming and registering a codec on Android and gp-gpu use?
But with Android things seem a bit more complicated to get performance due to the way things are structured:
Backend Camera Streamlining?
Previous high level camera projects have failed due to the underlying restrictions of the android camera interface and customisations from phone to phone, but also Android's slow nature. This is an attempt to bypass this with a high performance codec section.* L and M, reportedly address the deficiencies somewhat, but for the codec I realise the data rate of video data coming in might be poor, and* I might have to write a back end to acquire the frames from the hardware to the codec quick enough, which I don't want to do, but if I can't get frame data delivery fast enough I will have to look at it. I want to use mainly the GPU or other processing units instead of the main processor, for power efficiency and speed, but realise nothing is simple. All that sort of stuff that you have to do because it was not done right in the first place. So, avoiding going through slower high end camera interfaces as much as possible. I understand it is all based on a standard Linux camera API. If the camera software does not have to be rewritten and it can deliver frame data at streamlined timely speeds to a codec, then I can avoid much of this. So, I probably need advice in these things too.
Backend Storage Streamlining?
Now, on the other side we have storage**Hopefully the data rate can be small enough to avoid issues, but that is unlikely on a 4k-8k frame and would need advice on this too.
JavaScript to Android, Android to JavaScript transportability?
I actually want to develop the core of it within JavaScript primarily, for transportable use on the web and Firefox OS, so will have to find out the best way to transfer it to Android for compilation? As I know next to nothing about these new languages, it will be an uphill learning curve. As I understand, JavaScript syntax is separate from Java, and not a even a logical subset, which makes life hard.
----------
Anyway, it is a shame we don't have a kick starter like funding scheme, to pay a good programmer to do most of the background stuff, and upgrade the Linux code and drivers, so anybody can use the new code with any codec and camera app combination. My main interest is my own codec, not all the other stuff, that is really fixing Android and Linux camera code, which would help everybody.
This is not an official project start, just implementation research.
If anybody knows of anybody that can contribute, please direct them here?
Thanks.
Stevio2 said:
Advice, developing a high end video compression codec on Shield Android TV for Camera Acquisition and HQ video.
Hi
Aims
I am researching doing a high end streamlined video compression/decompression codec that can be installed and registered under Android, and be available to third party camera, editing and video apps.* Shield seems like a good top end development target.* I am hoping it will be able to compress 4k+ video streams, with small file sizes and reduced processing overheads.
Even though it is meant to be more for high end camera acquisition on Android in general, it also has other uses on the web.
I am trying to find out general, and detailed, information to see what I need to address.* I'm a newbie to all this, from back in the days that C++ was new and untaught in my college. I'm going to have to reteach myself programming, but have a lot of knowledge on the design side due to previous work.
Codec Programming?
So basically, I need advice on broad programming info on programming and registering a codec on Android and gp-gpu use?
But with Android things seem a bit more complicated to get performance due to the way things are structured:
Backend Camera Streamlining?
Previous high level camera projects have failed due to the underlying restrictions of the android camera interface and customisations from phone to phone, but also Android's slow nature. This is an attempt to bypass this with a high performance codec section.* L and M, reportedly address the deficiencies somewhat, but for the codec I realise the data rate of video data coming in might be poor, and* I might have to write a back end to acquire the frames from the hardware to the codec quick enough, which I don't want to do, but if I can't get frame data delivery fast enough I will have to look at it. I want to use mainly the GPU or other processing units instead of the main processor, for power efficiency and speed, but realise nothing is simple. All that sort of stuff that you have to do because it was not done right in the first place. So, avoiding going through slower high end camera interfaces as much as possible. I understand it is all based on a standard Linux camera API. If the camera software does not have to be rewritten and it can deliver frame data at streamlined timely speeds to a codec, then I can avoid much of this. So, I probably need advice in these things too.
Backend Storage Streamlining?
Now, on the other side we have storage**Hopefully the data rate can be small enough to avoid issues, but that is unlikely on a 4k-8k frame and would need advice on this too.
JavaScript to Android, Android to JavaScript transportability?
I actually want to develop the core of it within JavaScript primarily, for transportable use on the web and Firefox OS, so will have to find out the best way to transfer it to Android for compilation? As I know next to nothing about these new languages, it will be an uphill learning curve. As I understand, JavaScript syntax is separate from Java, and not a even a logical subset, which makes life hard.
----------
Anyway, it is a shame we don't have a kick starter like funding scheme, to pay a good programmer to do most of the background stuff, and upgrade the Linux code and drivers, so anybody can use the new code with any codec and camera app combination. My main interest is my own codec, not all the other stuff, that is really fixing Android and Linux camera code, which would help everybody.
This is not an official project start, just implementation research.
Click to expand...
Click to collapse
I wish you all the luck in your endeavour, as this sounds really interesting, and different........
Saying that, i dont think your suppose to post anything in the dev thread that is'nt an actuall work, im just giving you a heads up, incase a moderator might come along............also i could be wrong, if this has changed recently
Your best bet i reckon, is to post in this thread
http://forum.xda-developers.com/general/general
Its the main general thread of the entire xda, so you'll have more eyeballs.........and maybe a better chance of getting a "start in the right direction" from someone knowledgeable
Ive also read many android technical question being asked at the "stackexchange" website, by devs working on their projects, so that might be another avenue to explore if your unlucky here
Anyways, wish you luck with this
Development Forums (ones with the word development in the title) - For Developers to post release threads e.g. ROMs and Kernels including modifications to kernels, bootloaders, ROMs, etc., as well as R&D development discussion threads designed with an end goal
Click to expand...
Click to collapse
Thanks. From forum discussion rules. I mistook this to mean development research discussion as well. If it actually should, then I'm happy for it to be moved to general.
Stevio2 said:
Thanks. From forum discussion rules. I mistook this to mean development research discussion as well. If it actually should, then I'm happy for it to be moved to general.
Click to expand...
Click to collapse
The Shield is based on the Nvidia X1 chip. Nvidia also just released the Jetson TX1 development board which is similar. If you register as a developer with nvidia (which is easy) you get access to all the dev docs (including video codec docs) for the TX1, which boots Ubuntu. That should be a good start.
Sounds good, it was an andriod development related question though (using shield hardware under android so it can be shared with different platforms, you just can do more on the shield hardware. Maybe there is a Linux overlap with andriod in codec support but I doubt it isva full story. I am interested in dealing with 8k content too. There is a way to do a 8k over HDMI 2, but muchntoo involved at this stage, the display has to also be modified or an adaptor made to interface to a future 8k interface.
I have just realised the shield might be good for touch table work (not so good on the software side as there are no established software base to work on). I located a new good cheap fine grained more transparent touch surface overlay technology a little while ago that is being used to do cheap touch tables in Asia. Using a 4 subpixel screen I can do a semi 8k display out of a 4k (though you can't directly access the white pixel through hdmi, which is useless). There is also now 6 color pixels. A firmware change might allow a display to sub pixel address. However, you can get panel frames without the internal section and get direct access to the internal panel interface (why hdcp is probably useless). Anyway, 8k would yeild 16k, a nice minimum for a 80 inch table, with OLED, or projector. Reprogramming a display to use display port/thuderbolt interfaces on a display would be more useful. I tried to negotiate access to a 16k projector chip once to connect up to a low powered processing array, but got nowhere. Henceforth I've been dealing with embedded machine code level concerns for decades off and on and let the newer high level language and OS stuff (like C# and Linux) go, due to health issues.
Another intetesting thing that can be done with a shield, is it can be hooked up to a camera head and rigged up to be a camera (or the next version). Problem is that USB 3 is useless compared to Thuderbolt 3 etc (though camera head computet interface standards take a while to catch up). My codec could be used for recording. We used to do this with PC's but the Shield offers a much better power consumption. There is Linux software around to do this, but the development board is half powered and expensive.
Bump
Well, when I said bump, I didn't really mean to move it to a third subforum
Seriously, I want to do a less than 20mbit/s 8k visually lossless codec. But at the moment I'm waiting to get checked out for dementia, which explains a lot about the last few decades and my decreasing amount I can do (beta amyloid in particular builds up for 10-20 years with low grade symptoms before it gets seriouse enough that it is can be picked up on older scanning, by then it has permanent problems. Apart from other types of dementia). At this stage I can't do much much of the time.
Anyway, as the thread has skipped to a second forum in two days, any more short cut advice is welcome.
Hello everyone,
let's make it as short as possible. As the title says, there are plenty of Polaroid tablets with not-bad specs ( one core arm cpu clocked at 1Ghz or more) that need some of the Devs magic. For unknown reasons, the manufacturing comapny, the very famous Polaroid released the MIDC range and left it without any upgrades and I am sure their forums is full with threads begging them to give our old but much loved devices upgrades to newer android versions or just updates to fix bugs and add some nowadays necessary features.
And here I am, the unelected spokesman of all Polaroid MIDC tablets owners, looking for some attention and care from our beloved XDA-devs to do something in regard of this cause.
Thanks in advance. I am totally willing to pay Devs for this cause and want to try Google now on my device please.
My device: Polaroid MIDC970PR001
CPU : Cortex Allwinner A10 1.01 GHz 1xARM Cortex-A8
GPU : Mali-400 MP
Screen : 8.02 inches 1024x768
RAM : 814 MB
Storage : 1GB internal sorage + 6GB
System: the very old Android 4.0.4 ( who in the world still has android 4.0.4 on a tablet).
ballakanzri said:
Hello everyone,
let's make it as short as possible. As the title says, there are plenty of Polaroid tablets with not-bad specs ( one core arm cpu clocked at 1Ghz or more) that need some of the Devs magic. For unknown reasons, the manufacturing comapny, the very famous Polaroid released the MIDC range and left it without any upgrades and I am sure their forums is full with threads begging them to give our old but much loved devices upgrades to newer android versions or just updates to fix bugs and add some nowadays necessary features.
And here I am, the unelected spokesman of all Polaroid MIDC tablets owners, looking for some attention and care from our beloved XDA-devs to do something in regard of this cause.
Thanks in advance. I am totally willing to pay Devs for this cause and want to try Google now on my device please.
My device: Polaroid MIDC970PR001
CPU : Cortex Allwinner A10 1.01 GHz 1xARM Cortex-A8
GPU : Mali-400 MP
Screen : 8.02 inches 1024x768
RAM : 814 MB
Storage : 1GB internal sorage + 6GB
System: the very old Android 4.0.4 ( who in the world still has android 4.0.4 on a tablet).
Click to expand...
Click to collapse
I got ics on my pmid10c. Polaroid strung me along from when i bought it new in 2011 until this year about kernel source.
First, "the device is too new for releasing kernel source."
Later "we're referring that question to engineering who'll get back to you."
Never heard back after any of my many many phone calls and emails, until the last time when they actually called me back.
"The device isn't supported anymore, the factory that built it is closed, and we don't have access to the source code now."
Worthless manufacturer.
I talked with another developer here on xda about kernel source, the sunxi kernels build and boot but lack support for the extra hardware like touchscreen and sound and gps etc.
Find me kernel source and maybe i can help out.
bigsupersquid said:
I got ics on my pmid10c. Polaroid strung me along from when i bought it new in 2011 until this year about kernel source.
First, "the device is too new for releasing kernel source."
Later "we're referring that question to engineering who'll get back to you."
Never heard back after any of my many many phone calls and emails, until the last time when they actually called me back.
"The device isn't supported anymore, the factory that built it is closed, and we don't have access to the source code now."
Worthless manufacturer.
I talked with another developer here on xda about kernel source, the sunxi kernels build and boot but lack support for the extra hardware like touchscreen and sound and gps etc.
Find me kernel source and maybe i can help out.
Click to expand...
Click to collapse
It's Ok, mine is dead now! it won't start no matter what I do to it.
Thanks a lot for your interest.
Editing with case closed:
HibikiTaisuna said:
The Moto Z2 Play uses a Snapdragon 626 which is based on the Cortex-A53 (same goes for the Snapdragon 625 in the Moto Z Play). This architecture isn't exploitable by Spectre (https://developer.arm.com/support/security-update). Meltdown is Intel-only so far.
Click to expand...
Click to collapse
Guess we dodged a bullet there!
________ original post_______
I guess by now almost everyone know about the huge design flaw that's affecting almost every processor available. The ARM inside our Z2Play may also be vulnerable.
With these exploits, malicious code can extract basically anything from our phones, from passwords, to certificates, files, you name it.
Aaaaaand we also know that the mitigation of these vulnerabilities depends on software updates and that carriers may take forever to do so.
So I think we can use this space to discuss/get up to date on this issue and get is solved ASAP.
I'll update this post with all the new information we gather along the way.
Links to understand the vulnerabilities (Meltdown is as simple as a JS code loaded within a webpage)
https://twitter.com/nicoleperlroth/status/948684376249962496
https://meltdownattack.com/
https://spectreattack.com/
The Moto Z2 Play uses a Snapdragon 626 which is based on the Cortex-A53 (same goes for the Snapdragon 625 in the Moto Z Play). This architecture isn't exploitable by Spectre (https://developer.arm.com/support/security-update). Meltdown is Intel-only so far.
Hello everyone, I am Dutch and new on this forum.
I'm in the happy possession of a Peugeot 508 from 2012.
Since the Here navigation leaves a lot to be desired and I would also have some other software in the car, I'm considering buying an Android Head Unit from AliExpress.
There's quite a few offers, so I have some questions. I hope that someone has experience with the systems or can give me advice.
Just some examples that I found: https://nl.aliexpress.com/item/4001205344015.html and https://nl.aliexpress.com/item/1005001582465605.html
They all claim to be 'plug and play' and units are available for various brands and types of car.
Are these systems substantially different from newer Android systems in current cars?
I have the idea that all brands offered are all equal internally and that only the interface differs. Is that right?
Can a new Android version be updated easily?
All are available with 4Gb OF RAM, 64Gb ROM, Hexa or Octa Core, PX6 CPU, DSP. Are there any things I need to look out for?
I have an iPhone so I would also like to buy a Carplay dongel. Does this work properly?
Are there any other brands that might fit in this car?
Thanks for the help.
Best regards,
Rob
The cheap "4K Android TV" series from Walmart, built by Funai but branded as Philips, is extremely affordable and seems to be unlocked. Has anyone had any luck getting these to run alternative firmware (Lineage?) or even stock Android TV?
Mine is extremely buggy (forgets the date/time, causing HBO and other streaming apps to crash, etc) and extremely slow. Many times apps never fully exit, eating up the limited RAM available. Newer apps refuse to install from the Play Store (even though I'm able to force install them by downloading the APK) which tells me there's some sort of poorly-coded identification going on.
I was able to figure out some hardware specs:
It reports the board/device as "freesia" and the hardware/platform as "maxim" - the quad-core ARM Cortex A53 chip runs at 1400mhz (rev r0p4) and it has 1.5GB RAM and 4GB internal storage. The built-in graphics report as Mali T820 (ARM) (vulkan integrated gpu 1358996kb) running at 1920x1080.
I've updated it as far as the factory updates will go to Android 9 Pie.
I'm new to doing much with Android but I write software for a living so pretty comfortable hacking this thing if people can give me a few pointers!
Rooting a Funai Philips Android TV
I've got a 65-inch 4K Funai/Philips Android TV. Is this TV rootable or flashable? The OS is so laggy on this device it makes it almost unbearable to use. Opening specific screens takes seconds because of all the quirky animations. I'd much rather...
forum.xda-developers.com
The above has users who have unlocked fastboot etc, and have achieved root through the OTA firmware
Thanks. Do you know what it would take to go beyond that? I asked on the Lineage subreddit but they basically have rules against talking about anything but official builds, so it's kind of a catch 22.
andsat said:
Thanks. Do you know what it would take to go beyond that? I asked on the Lineage subreddit but they basically have rules against talking about anything but official builds, so it's kind of a catch 22.
Click to expand...
Click to collapse
To be honest? ¯\_(ツ)_/¯
Lineage for TVs seems far fetched,
You’d need to find a person who’d spend their time to build a rom for your TV.
Don’t know much about ROM development
It's already been done for other devices. Here's one for Raspberry Pi https://konstakang.com/devices/rpi4/LineageOS18-ATV/
how does one get/obtain the ota firmware ? I still have an unupdated bootloader unlocked 75 inch model and would like to do some stuff with it. i have been waiting to connect it to the internet because i didn't want to be lock into or out of anything until i was ready.