I am creating a directshow filter which's purpose is to take 3 input pins and create a video which shows alternately vidoe from the first source, the second source and the third source, in a fixed time internal.
So if i have three webcam connected to my filter, i want the final video for example to show 5 seconds of the first cam, five seconds of the second cam, and so on...
I have tried two approaches:
Approach one
I use a class TimeManager. This class has a function isItPinsTurn(pinname). This functions returns true or false regarding if the pin is supposed to send sample to the output. To do this the TimeManager creates a new thread which sleeps every x seconds.
After it slept it changes to the current active inputpin to the next.
The result is that every x seconds the isItPinSTurn(pinname) function returns another pin. This way every pin only seconds output to the outputpin when it is its turn, hence i get the desired videos with x intervalls between the input cam.
The problem with this approach
Sleep doesn't seem to work in directshow filters. I get a runtime error:
abort() has been called
Approach two
I use the samples GetMediaTime method and a buffer which keeps track of how much video samples in terms of its mediatime, has already been sent to the output pin. This is best illustrated with code:
void MyFilter::acceptFilterInput(LPCWSTR pinname, IMediaSample* sample)
{
mylogger->LogDebug("In acceptFIlterInput", L"D:\\TEMP\\yc.log");
if (wcscmp(pinname, this->currentInputPin) == 0)
{
outpin->Deliver(sample);
LONGLONG timestart;
LONGLONG timeend;
sample->GetTime(×tart, &timeend);
*mediaTimeBuffer += timeend - timestart;
if (*mediaTimeBuffer > this->MEDIATIME)
{
this->SetNextPinActive(pinname);
*mediaTimeBuffer = 0;
}
}
}
When the filter starts the currentInputPin is set to pin0 (the first). Calls to acceptFilterInput (which is called by the the input pins receie function) adjust the mediaTimeBUffer with the size of the MediaSample-MediaTime. If this buffer is higher than MEDIATIME (which can for example be 5 (seconds)), the buffer is set back to zero and the next pin is set active.
Problems with this approach
I am not even sure if CMediaSample->GetMediaTime returns the data i need, as it seems to return negative numbers, which doesn't seem to make much sense. I didn't find useful information about the return value of GetMediaTime on the web.
You are expected to block execution (incoming calls to IPin::Receive) on input streams so that other streams could catch up on their own streaming threads. You typically achieve this by either using wait/synchronization APIs and functions, or by holding references on media samples so that input peer would block on empty allocator waiting for a media sample (buffer) to get available.
Yes Sleep works well, although polling is the worst of possible options.
Approach two does not make sense for me because I don't see any real synchronization there: there is no execution blocking, and there is no making pin active. You cannot force data on the input pin, you only can wait to get called with new media sample. So you should block accepting data on one input stream/pin until you get data on another.
Some useful relevant information on multiplexing:
How to make a DirectShow Muxer Filter - Part 1
How to make a DirectShow Muxer Filter - Part 2
GDCL MPEG-4 Multiplexer - available in source, and can multiplex data from 2+ streams
Related
I am trying to control a robot using a template-based controller class written in c++. Essentially I have a UDP connection setup with the robot to receive the state of the robot and send new torque commands to the robot. I receive new observations at a higher frequency (say 2000Hz) and my controller takes about 1ms (1000Hz) to calculate new torque commands to send to the robot. The problem I am facing is that I don't want my main code to wait to send the old torque commands while my controller is still calculating new commands to send. From what I understand I can use Ubuntu with RT-Linux kernel, multi-thread the code so that my getTorques() method runs in a different thread, set priorities for the process, and use mutexes and locks to avoid data race between the 2 threads, but I was hoping to learn what the best strategies to write hard-realtime code for such a problem are.
// main.cpp
#include "CONTROLLER.h"
#include "llapi.h"
void main{
...
CONTROLLERclass obj;
...
double new_observation;
double u;
...
while(communicating){
get_newObs(new_observation); // Get new state of the robot (2000Hz)
obj.getTorques(new_observation, u); // Takes about 1ms to calculate new torques
send_newCommands(u); // Send the new torque commands to the robot
}
...
}
Thanks in advance!
Okay, so first of all, it sounds to me like you need to deal with the fact that you receive input at 2 KHz, but can only compute results at about 1 KHz.
Based on that, you're apparently going to have to discard roughly half the inputs, or else somehow (in a way that makes sense for your application) quickly combine the inputs that have arrived since the last time you processed the inputs.
But as the code is structured right now, you're going to fetch and process older and older inputs, so even though you're producing outputs at ~1 KHz, those outputs are constantly being based on older and older data.
For the moment, let's assume you want to receive inputs as fast as you can, and when you're ready to do so, you process the most recent input you've received, produce an output based on that input, and repeat.
In that case, you'd probably end up with something on this general order (using C++ threads and atomics for the moment):
std::atomic<double> new_observation;
std::thread receiver = [&] {
double d;
get_newObs(d);
new_observation = d;
};
std::thread sender = [&] {
auto input = new_observation;
auto u = get_torques(input);
send_newCommands(u);
};
I've assumed that you'll always receive input faster than you can consume it, so the processing thread can always process whatever input is waiting, without receiving anything to indicate that the input has been updated since it was last processed. If that's wrong, things get a little more complex, but I'm not going to try to deal with that right now, since it sounds like it's unnecessary.
As far as the code itself goes, the only thing that may not be obvious is that instead of passing a reference to new_input to either of the existing functions, I've read new_input into variable local to the thread, then passed a reference to that.
What is the primary difference between SetTime and SetMediaTime?
Right now in my directshow livesource I calculate the time it like this
REFERENCE_TIME rtStart = m_rtLastSampleTime;
m_rtLastSampleTime += pVih->AvgTimePerFrame;
pms->SetTime(&rtStart, &m_rtLastSampleTime);
pms->SetSyncPoint(TRUE);
pms->SetDiscontinuity(rtStart <= 1);
This doesn't work with some encoders.
I've noticed that source that do work with these encoders set mediatime and they seem to jump up.
Media Times:
Optionally, the filter can also specify a media time for the sample. In a video stream, media time represents the frame number. In an audio stream, media time represents the sample number in the packet. For example, if each packet contains one second of 44.1 kilohertz (kHz) audio, the first packet has a media start time of zero and a media stop time of 44100. In a seekable stream, the media time is always relative to the start time of the stream. For example, suppose you seek to 2 seconds from the start of a 15-fps video stream. The first media sample after the seek has a time stamp of zero but a media time of 30.
Renderer and mux filters can use the media time to determine whether frames or samples have been dropped, by checking for gaps. However, filters are not required to set the media time. To set the media time on a sample, call the IMediaSample::SetMediaTime method.
I don't think it is actually used anywhere. SetTime, on the contrary, is important.
Can anyone help decipher the correct implementation of the libspotify get_audio_buffer_stats callback. Specifically, we are supposed to populate a sp_audio_buffer_stats buffer, consisting of samples and stutter?
According to the Docs:
int samples - Samples in buffer.
int stutter - Number of stutters (audio dropouts) since last query.
I'm wondering about "samples." What exactly is this referring to?
The music playback (audio_delivery) callback has a num_frames variable, but then you have the issue of audio format (channels and/or sample_rate).
Is it correct to set "samples" to total amount of "num_frames" currently in my buffer? Or do I need to run some math based on total "num_samples", "channels", and "sample_rate"
It should be the number of frames in your output buffer. I.e. int samples is slightly misnamed and should probably be called int frames instead.
I am continuing work on my DirectShow application and am just putting the finishing touches on it. What the program is doing is going through a video file in 1 second intervals and capturing from the samplegrabber the current buffer and processing it before moving on. However, I was noticing that I was getting duplicate images in my tests to which I found out that DirectShow has not incremented through the video in that 1 second interval fast enough. My question is if there is a way to check when DirectShow is ready for me to call the samplegrabber to get the current frame and to process it. At the moment I call sleep for 1 second but there has to be a better method. Thank you in advance for the help.
EDIT
I just tried running a check to see if the video's position is equal to the next position that I would like to grab and process. That decreased the number of duplicate frames but I still see them showing up in chunks.
I always let the DS framework handle the processing rate:
in the main application thread, configure the sample grabber callback and then when the callback is triggered, you get the media sample as well as sample time: at this point you can process the sample if the appropriate interval i.e. 1 second has elapsed.
What do you mean you call sleep for a second and from where (which thread) do you call it?
If you're doing this from inside the callback you are effectively blocking the DirectShow pipeline? Perhaps if you could explain your setup in more detail I could be more helpful.
/// Callback that is triggered per media sample
/// Note this all happens in the DirectShow streaming thread!
STDMETHODIMP SampleCB( double dSampleTime, IMediaSample * pSample )
{
// check timestamp and if one second has elapsed process sample accordingly
// else do nothing
...
// Get byte pointer
BYTE* pbData(NULL);
HRESULT hr = pSample->GetPointer(&pbData);
if (FAILED(hr))
{
return hr;
}
...
}
P.S if you want to process the samples as fast as possible, you can set the sample timestamp to NULL in your callback.
// set time to NULL to allow for fast rendering since the
// the video renderer controls the rendering rate according
// to the timestamps
pSample->SetTime(NULL, NULL);
Try setting your graph timer to NULL. It will allow to:
process the file as fast as possible
will relieve you of the issues you have.
Of course, it won't work if you are rendering the file to screen at the same time.
Can someone explain how snd_pcm_writei
snd_pcm_sframes_t snd_pcm_writei(snd_pcm_t *pcm, const void *buffer,
snd_pcm_uframes_t size)
works?
I have used it like so:
for (int i = 0; i < 1; i++) {
f = snd_pcm_writei(handle, buffer, frames);
...
}
Full source code at http://pastebin.com/m2f28b578
Does this mean, that I shouldn't give snd_pcm_writei() the number of
all the frames in buffer, but only
sample_rate * latency = frames
?
So if I e.g. have:
sample_rate = 44100
latency = 0.5 [s]
all_frames = 100000
The number of frames that I should give to snd_pcm_writei() would be
sample_rate * latency = frames
44100*0.5 = 22050
and the number of iterations the for-loop should be?:
(int) 100000/22050 = 4; with frames=22050
and one extra, but only with
100000 mod 22050 = 11800
frames?
Is that how it works?
Louise
http://www.alsa-project.org/alsa-doc/alsa-lib/group___p_c_m.html#gf13067c0ebde29118ca05af76e5b17a9
frames should be the number of frames (samples) you want to write from the buffer. Your system's sound driver will start transferring those samples to the sound card right away, and they will be played at a constant rate.
The latency is introduced in several places. There's latency from the data buffered by the driver while waiting to be transferred to the card. There's at least one buffer full of data that's being transferred to the card at any given moment, and there's buffering on the application side, which is what you seem to be concerned about.
To reduce latency on the application side you need to write the smallest buffer that will work for you. If your application performs a DSP task, that's typically one window's worth of data.
There's no advantage in writing small buffers in a loop - just go ahead and write everything in one go - but there's an important point to understand: to minimize latency, your application should write to the driver no faster than the driver is writing data to the sound card, or you'll end up piling up more data and accumulating more and more latency.
For a design that makes producing data in lockstep with the sound driver relatively easy, look at jack (http://jackaudio.org/) which is based on registering a callback function with the sound playback engine. In fact, you're probably just better off using jack instead of trying to do it yourself if you're really concerned about latency.
I think the reason for the "premature" device closure is that you need to call snd_pcm_drain(handle); prior to snd_pcm_close(handle); to ensure that all data is played before the device is closed.
I did some testing to determine why snd_pcm_writei() didn't seem to work for me using several examples I found in the ALSA tutorials and what I concluded was that the simple examples were doing a snd_pcm_close () before the sound device could play the complete stream sent it to it.
I set the rate to 11025, used a 128 byte random buffer, and for looped snd_pcm_writei() for 11025/128 for each second of sound. Two seconds required 86*2 calls snd_pcm_write() to get two seconds of sound.
In order to give the device sufficient time to convert the data to audio, I put used a for loop after the snd_pcm_writei() loop to delay execution of the snd_pcm_close() function.
After testing, I had to conclude that the sample code didn't supply enough samples to overcome the device latency before the snd_pcm_close function was called which implies that the close function has less latency than the snd_pcm_write() function.
If the ALSA driver's start threshold is not set properly (if in your case it is about 2s), then you will need to call snd_pcm_start() to start the data rendering immediately after snd_pcm_writei().
Or you may set appropriate threshold in the SW params of ALSA device.
ref:
http://www.alsa-project.org/alsa-doc/alsa-lib/group___p_c_m.html
http://www.alsa-project.org/alsa-doc/alsa-lib/group___p_c_m___s_w___params.html