We have a issue that specific combinations of filters including VMR7 causes frame is not rendered right. We noticed it is only happened with certain GPU card with some of driver versions.
We try to make some workaround (with some overhead) only for the GPU. Is any way to know the underlying VGA card associated with the VMR7?
I've found the answers for my question.
There is monitor related information interface IVMRMonitorConfig which is query-able from VMR7 filter to ask about device information which is associated.
https://msdn.microsoft.com/en-us/library/windows/desktop/dd390488(v=vs.85).aspx
IVMRMonitorConfig::GetAvailableMonitors(
[out] VMRMONITORINFO *pInfo,
[in] DWORD dwMaxInfoArraySize,
[out] DWORD *pdwNumDevices
);
I can recognize the specific VGA card by the keyword in VMRMONITORINFO::szDevice or VMRMONITORINFO::szDescription string.
Related
I am trying to understand timer interrupts in Arduino.
I am finding a major difficulty in retrieving documentation.
Sources online refer to apparently "magical" constants (like TIMSK1 for example) but I am unable to find where they are defined.
Do they come from some sort of header file?
Is there some reference documentation anywhere?
You've already answered, but I'd like to fill in some gaps, perhaps.
Those symbols are not magical, they are the names of the processor and I/O registers and values for the bit-fields inside those registers. As you found, the processor datasheet explains the layout of the I/O registers. This is standard for all microcontroller vendors.
Here's the home page for the Mega328P: https://www.microchip.com/en-us/product/ATmega328p
since the datasheet PDF might change some day. Note that farther down the page there are many very helpful "application notes" that describe how to do things...
How does a C program, or an Arduino script, get the values of the registers and fields? From the vendor supplied header files. See here for Microchips "packs": http://packs.download.atmel.com
The compiler that comes with the Arduino IDE distribution has included the headers for the boards it supports. They're buried pretty deeply - we go to the datasheet to find the register and field names, not usually to the header files.
For the mega328, the header file is here (macos): ~/Arduino.app/Contents/Java/hardware/tools/avr/avr/include/avr/iom328p.h
I'm not sure why you said "(although not explained in full detail)". There's really not much to it - it just holds a few bits that enable interrupts when certain things happen in Timer/Counter 1.
From the 328P datasheet 01/2015, TIMSK1 is described on page 90 "All interrupts are individually masked with the Timer Interrupt Mask Register (TIMSK1)." The register layout is on page 112. And its position within the I/O register set on page 278. A programmer doesn't need more than that! :-)
(I wish I could attach a pdf... it has a big table of the registers, fields, interrupt vectors, DIP pins, ... all generated from that header file.)
After some search I found that in the the ATMega processor manual those constants are named (although not explained in full detail)
One possible location of the manual is:
https://ww1.microchip.com/downloads/en/DeviceDoc/Atmel-7810-Automotive-Microcontrollers-ATmega328P_Datasheet.pdf
When receiving an output from IMFTransform::ProcessOutput, calling GetUINT32(MFSampleExtension_CleanPoint) on the sample fails and returns MF_E_ATTRIBUTENOTFOUND only while using the AMDh264Encoder (NV12 in, H264 out), as such there are no keyframes in the final output video so it is corrupted.
What causes getting attribute MFSampleExtension_CleanPoint MF_E_ATTRIBUTENOTFOUND to fail, only on the AMDh264Encoder?
Video encoder MFTs are supplied by hardware vendors. AMD does "AMDh264Encoder" for their hardware and introduces it with their video drivers, in particular.
For this reason implementations from different vendors have slight differences, AMD guys decided to not set this attributes on produced media samples.
You should skip this gracefully and treat the attribute as optional.
Do we have to go through the Device tree bindings documentation of a linux kernel when you start working on it.
Is there no standard set of fields in the device tree which are followed by all distros/kernel sources?
Secondly I need some guidance regarding adding nodes for devices on gpio bus using device tree. I have already consulted http://devicetree.org/Device_Tree_Usage.
stackoverflow-query here should point you to documentation on device tree. And yes it is a good idea to go through the documentation before you dive into using it.
As for your gpio devices (I assume you already have a gpio controller in place in your dts/dtsi file in place), there should be plenty under arch/arc/boot/dts . Pick one :)!
Eg: gpio1_8 for mmc dts and gpio1 controller dtsi
Device Tree Binding for peripherals in an SoC:
As an example for the v5.1 kernel, here are the device tree bindings listed for various peripherals available on an SoC.
Link:
https://elixir.bootlin.com/linux/v5.1/source/Documentation/devicetree/bindings
Device Tree Binding for a particular peripheral in an SoC:
To explain a bit about the device tree binding for a particular peripheral let's take an example of an SPI for a very popular TI OMAP family.
Link:
https://elixir.bootlin.com/linux/v5.1/source/Documentation/devicetree/bindings/spi/omap-spi.txt
The text in this particular link introduces basically the key-value pairs. The 'key' is the device tree property and the 'value' is the possible place holder values for the corresponding 'key'. As an example, in the above link the "compatible" property, which holds one of the value as "ti,omap2-mcspi".
Another example is the "dma-names" property which holds txN, rxN .
Now, in the below link you can clearly see how these device tree properties are used in the real device trees:
https://elixir.bootlin.com/linux/v5.1/source/arch/arm/boot/dts/omap3.dtsi#L365
mcspi1: spi#48098000 {
compatible = "ti,omap2-mcspi";
reg = <0x48098000 0x100>;
...
}
The value "ti,omap2-mcspi" for the key "compatible" is one of the available value in accordance with the device tree binding document for omap-spi.txt (as seen in the second listed link).
So now based on SoC and the peripheral used, the device tree property can be written consulting the device tree binding document.
I have to configure my video camera display resolution before capturing and processing the data. Initially I did it as follows.
Created all necessary interfaces.
Added camera and renderer filters
Did RenderStream with Capture and Preview PIN Categories.
Then did the looping through AM_MEDIA_TYPE structures and setting the params.
This worked for a lot of cameras, but a few cameras failed. Then I changed the order of 3 and 4 given above. That is, I did the setting of params before the RenderStream. This time, the error cases went through, but a few On board cameras in SONY VAIO laptop etc seem to fail.
Now, my questions are
Which is the optimal and correct method of getting and setting AM_MEDIA_TYPE parameters and running the graph?
If there are different cameras, if I get an indication of which order is the best for a particular camera by going through the camera's DirectShow interfaces, that will also serve my purpose.
Please help me in this at the earliest,
Thanks and regards,
Shiju
IAMStreamConfig::SetFormat needs to be used to set capture format before the pin is connected and rendered. This way the downstream subchain of filters is built with proper media types.
FMLE = Flash live media encoder 3.0
i have posted this question on Adobe Forum, but not sure if they have people on that forum with programming experience.
I am a developer writing a video capture and audio capture device. The devices already work in other encoders. The devices are written in directshow. I am integrating with FMLE and encountered this problem.
The audio device doesnt have a usable volume bar in FMLE. The FMLE error is "The selected audio device "censored (company secret)" doesn't allow setting volume intensity. Disabling the volume slider control."
my audio device implements these interfaces along with the standard directshow filter interfaces
IBasicAudio
IAMAudioInputMixer
I put tracepoints in queryinterface and found FMLE query's for (my comments in comment string)
{IID_IUnknown}
{IID_IPersistPropertyBag}
{IID_IBaseFilter}
{IID_IAMOpenProgress}
{IID_IAMDeviceRemoval}
{IID_IMediaFilter}
{IID_IAMBufferNegotiation}
{IID_IAMStreamConfig}
{IID_IPin}
{IID_IReferenceClock}
{IID_IMediaSeeking}
{IID_IMediaPosition}
{IID_IVideoWindow} // WTF ?? query video window ?
{IID_IBasicAudio}
{2DD74950-A890-11D1-ABE8-00A0C905F375} // i think this is async stream,
What am i missing ? FMLE doesnt use IAMAudioInputMixer ?
Anyone know the exact interface which FMLE uses for Volume intensity ? . .I assumed it was IBasicAudio, but it doesnt seem to call any methods in there.
Answer provided by Ram Gupta of adobe forum.
"FMLE does not query for CLSID_AudioInputMixerProperties interface.
FMLE enumerates all the pin of audio source filter(using EnumPins) and then it extracts each pin info using QueryPinInfo Function.
FMLE searches for the audio filter Pin whose direction is PINDIR_INPUT(using QueryPinInfo) and then it queries for IAMAudioInputMixer interface to set the volume level.
Could you pls chk if the following functions are properly implemented
-->get_enable: it should set its parameter value to true.
-->put_MixLevel
-->QueryPinInfo:"
This solution did work. My problem was that because i never declared an input pin (since i dont have any directshow related input).