I am developing a application in c# by using Directshow.NET. I am using a virtual camera which will help to record desktop screen. So my graph is:
Virtual cam --->color space converter --->sample grabber ---> ASF writer.
While coding, I used a custom .prx which I generated by Windows Media Profile Editorand configured into IConfigAsfWriterusing WMCreateProfileManager.
In .prx file Mode is CBR, codec Windows Media Video 9 and frame rate 15fps with 759Kbps video bit rate, but still video looks so blurry. If I increase video bit rate like upto 5Mbps then this blurriness is not coming but increasing bit rate results into large file size (54 seconds of recording file size is 10MB).
I tried another graph using graphEdit virtual cam ---> AVI mux ---> File writer but this also generating large .avi file.
How can i record video without blur effect by keeping minimum file size, for eg. 1 minute of video size up to 2-3 MB?
Do i need to use any video compressor?
The quality depends on the codec you use, but also on the number of bits per pixel. You can calculate it this way:
bits/pixel = bitrate / (width * height * framerate)
(bitrate in bits/second and framerate is in frames/second)
So if you want to reduce the bitrate without getting blurry video, you also have to reduce the resolution or framerate. That way you keep the number of bits per pixel the same.
Related
I am using DirectShow in my application to capture video from webcams. I have issues while using cameras to preview and capture 1080P videos. Eg: HD Pro Webcam C910 camera of Logitech.
1080P video preview was very jerky and no HD clarity was observed. I could see that the enumerated device name was "USB Video Device"
Today we installed Logitech webcam software on these XP machines . In that application, we could see the 1080P video without any jerking. Also we recorded 1080P video in the Logitech application and saw them in high quality.
But when I test my application,
I can see that the enumerated device name has been changed to "Logitech Pro Webcam C910" instead of the "USB Video Device" as in the previous case.
The CPU eaten up by my application is 20%, but the process "SYSTEM" eats up 60%+ and the overall CPU revolves around 100%
Even though the video quality has been greatly improved, the jerks are still there, may be due to the 100% CPU.
When I closed my application, the high CPU utlizaton by "System" process goes away.
Regarding my application - It uses ICaptureGraphBuilder2::RenderStream to create Preview and Capture streams.
In Capture Stream, I connect Camera filter to NULL renderer with sample grabber as the intermediate filter.
In preview stream, I have
g_pBuild->RenderStream(&PIN_CATEGORY_PREVIEW,&MEDIATYPE_Video,cam,NULL,NULL);
Preview is displayed on a windows as specified using IVideoWindow interface. I use the following
g_vidWin->put_Owner((OAHWND)(HWND)hWnd);
g_vidWin->put_WindowStyle(WS_CHILD | WS_CLIPSIBLINGS);
g_vidWin->put_MessageDrain((OAHWND)hWnd);
I tried setting Frame rate to different values ( AvgTimePerFrame = 500000 ( 20 fps ) and 666667(15 fps) etc.
But all the trials, still give the same result. Clarity has become more, but some jerks still remain and CPU is almost 100% due to 60+ % utlilization by "System". When I close my video application, usage by "System" goes back to 1-2 %.
Any help on this is most welcome.
Thanks in advance,
Use IAMStreamConfig.SetFormat() to select the frame rate, dimensions, color space, and compression of the output streams (Capture and Preview) from a capture device.
Aside: The comment above of "It doesn't change the source filter's own rate" is completely wrong. The whole purpose of this interface is to define the output format and framerate of the captured video.
Use IAMStreamConfig.GetStreamCaps() to determine what frames rates, dimensions, color spaces, and compression formats are available. Most cameras provide a number different formats.
It sounds like the fundamental issue you're facing is that USB bandwidth (at least prior to USB3) can't sustain 30fps 1080P without compression. I'm most familiar with the Microsoft LifeCam Studio family of USB cameras, and these devices perform hardware compression to send the video over the wire, and then eat up a substantial fraction of your CPU on the receiving end converting the compressed video from Motion JPEG into a YUV format. Presumably the Logitech cameras work in a similar fashion.
The framerate that cameras produce is influenced by the additional workload of performing auto-focus, auto-color correction, and auto-exposure in software. Try disabling all these features on your camera if possible. In the era of Skype, camera software and hardware has become less attentive to maintaining a high framerate in favor of better image quality.
The DirectShow timing model for capture continues to work even if the camera can't produce frames at the requested rate as long as the camera indicates that frames are missing. It does this using "dropped frame" count field which rides along with each captured frame. The sum of the dropped frames plus the "real" frames must equal the requested frame rate set via IAMStreamConfig.SetFormat().
Using the LifeCam Studio on an I7 I have captured at 30fps 720p with preview, compressed to H.264 and written an .mp4 file to disk using around 30% of the CPU, but only if all the auto-focus/color/exposure settings on the camera are disabled.
I'm using two custom push filters to inject audio and video (uncompressed RGB) into a DirectShow graph. I'm making a video capture application, so I'd like to encode the frames as they come in and store them in a file.
Up until now, I've used the ASF Writer to encode the input to a WMV file, but it appears the renderer is too slow to process high resolution input (such as 1920x1200x32). At least, FillBuffer() seems to only be able to process around 6-15 FPS, which obviously isn't fast enough.
I've tried increasing the cBuffers count in DecideBufferSize(), but that only pushes the problem to a later point, of course.
What are my options to speed up the process? What's the right way to do live high res encoding via DirectShow? I eventually want to end up with a WMV video, but maybe that has to be a post-processing step.
You have great answers posted here to your question: High resolution capture and encoding too slow. The task is too complex for the CPU in your system, which is just not fast enough to perform realtime video encoding in the configuration you set it to work.
My application consists of displaying a large custom tree like structure to the user that can eventually grow to massive proportions like the dimensions listed in the question. I allow them to export the image with the following line of code tied to a button click event:
var image:ImageSnapshot = ImageSnapshot.captureImage(this, 72, new PNGEncoder(), false);
I've managed to export images close to the dimensions listed but around there it start to get the error message listed below after spinning for close to 15 seconds:
Error: Error #1000: The system is out of memory.
at flash.utils::ByteArray/writeBytes()
at mx.graphics::ImageSnapshot$/mergePixelRows()[E:\dev\4.x\frameworks\projects\framework\src\mx\graphics\ImageSnapshot.as:511]
at mx.graphics::ImageSnapshot$/captureAll()[E:\dev\4.x\frameworks\projects\framework\src\mx\graphics\ImageSnapshot.as:482]
at mx.graphics::ImageSnapshot$/captureImage()[E:\dev\4.x\frameworks\projects\framework\src\mx\graphics\ImageSnapshot.as:318]
at vertical/saveChart()[C:\devel\workspace\vertical\src\CustomObject.mxml:501]
at vertical/__saveImageBtn_click()[C:\devel\workspace\vertical\src\CustomObject.mxml:574]
Is the flashplayer plugin for my browser running out of memory? I noticed in my task manager it got up to about 1.2GB of memory usage(I have 4GB on my system). If that is the case is it possible to limit the memory usage for a given function like the ImageSnapshot.captureImage() call above?
Is there maybe a way to generate the component into 2 or 4 ImageSnapshot objects and piece them together afterward?
Any advice would be greatly appreciated.
I believe the latest Flash Player 11 has a new feature to solve this issue:
"Enhanced high resolution bitmap support — BitmapData objects are no longer limited to a maximum resolution of 16 megapixels (16,777,215 pixels), and maximum bitmap width/height is no longer limited to 8,191 pixels, enabling the development of apps that utilize very large bitmaps." from this PDF
If you are using BitmapData, it makes a difference which FlashPlayer you are targetting:
versions VS maximum bitmapsize
flashplayer -9 : 2880x2880 px
flashplayer 10 : 4096x4096 px
flashplayer 11 : unlimited
I don't know what you exactly are trying to do with this huge capture, but I would recommend using tiles. Break it down to chunks of relative small bitmaps. Create them separately, so you don't have to open/create that huge amount of data in your memory.
Anyway, it would be nice to know if it is possible to encode that big-ass sized image, without Error #1000 out of memory errors.
I am trying to save out a large image from flash using bitmapdata and the jpegencoder. I am looking into the limitations of this process and have noticed you can only set bitmapdata pizel width and height to a certain amount and this might be flexible with what you set the jpegencoder quality to (1-100).
Does anyone know what the specific limitations of these two things are? I'm basically trying to see just how large of an image I can save out (because I need to use the image exported for printing purposes, so I need it as high quality as possible).
I have read articles that say in fp 10 you can render up to something like 16,000 px. But I tried an image that is 3500 x 3500 and it timed out. So not sure if this is correct information.
The image size limit up to Flash Player 9 is 2880x2880, Flash 10 increased this limit to 4096x4096. This applies also for the Stage, Sprites and MovieClips.
The quality used for the JPGEncoder class does not circumvent this limitation as this is tied to the Flash core.
I'm using the FileReference class to upload flvs to a server.
Is it possible to check the flv length not size before allowing an upload?
Are you targeting Flash Player 10 alone or lower versions too? Because lower versions of Flash player (9 etc) do not allow the uploading SWF to read the contents of file (other than creationDate, creator (The Macintosh creator type of the file), modificationDate, name, size in bytes and type), so there is no way you are going to be able to do this on those players.
If you are targeting solely FP10 users, you can load the FLV into a ByteArray in your SWF and
Play it using an FLV player and read the duration property from the player. But I couldn't find an FLV player that takes a ByteArray as input - and after reading this thread in SO, it seems like that is not possible at all.
Parse the FLV file, and read the duration property from its metadata. The FLV file specification is open, but this isn't going to be easy.
Update to the comment:
Excerpts from the FLV file spec:
onMetaData
An FLV file can contain metadata with an “onMetaData” marker. Various stream properties
are available to a running ActionScript program via the NetStream.onMetaData property.
The available properties differ depending on the software used.
Common properties include:
duration: a DOUBLE indicating the total duration of the file in seconds
width: a DOUBLE indicating the width of the video in pixels
height: a DOUBLE indicating the height of the video in pixels
videodatarate: a DOUBLE indicating the video bit rate in kilobits per second
framerate: a DOUBLE indicating the number of frames per second
videocodecid: a DOUBLE indicating the video codec ID used in the file (see “Video
tags” on page 8 for available CodecID values)
audiosamplerate: a DOUBLE indicating the frequency at which the audio stream is
replayed
audiosamplesize: a DOUBLE indicating the resolution of a single audio sample
stereo: a BOOL indicating whether the data is stereo
audiocodecid: a DOUBLE indicating the audio codec ID used in the file (see “Audio
tags” on page 6 for available SoundFormat values)
filesize: a DOUBLE indicating the total size of the file in bytes
FLV file can contain metadata - it doesn't say it will contain metadata. It also says that available properties can vary based on the software used to create FLV. So I guess there is no guarantee (as per specs) that the duration property will be present. That said, duration is one of the basic properties of FLV and it would be safe to assume that any reasonable software would include it.
You can use Netstream.appendBytes to feed FileReference.data (after a call to browse, before a call to upload) to a NetStream used for playing a video. From there, the duration can be taken from the metadata, as described elsewhere on this thread. Note that at least Flash Player 10 is required for this approach.