Recherche avancée
Médias (1)
-
The Great Big Beautiful Tomorrow
28 octobre 2011, par
Mis à jour : Octobre 2011
Langue : English
Type : Texte
Autres articles (34)
-
Des sites réalisés avec MediaSPIP
2 mai 2011, parCette page présente quelques-uns des sites fonctionnant sous MediaSPIP.
Vous pouvez bien entendu ajouter le votre grâce au formulaire en bas de page. -
Support audio et vidéo HTML5
10 avril 2011MediaSPIP utilise les balises HTML5 video et audio pour la lecture de documents multimedia en profitant des dernières innovations du W3C supportées par les navigateurs modernes.
Pour les navigateurs plus anciens, le lecteur flash Flowplayer est utilisé.
Le lecteur HTML5 utilisé a été spécifiquement créé pour MediaSPIP : il est complètement modifiable graphiquement pour correspondre à un thème choisi.
Ces technologies permettent de distribuer vidéo et son à la fois sur des ordinateurs conventionnels (...) -
HTML5 audio and video support
13 avril 2011, parMediaSPIP uses HTML5 video and audio tags to play multimedia files, taking advantage of the latest W3C innovations supported by modern browsers.
The MediaSPIP player used has been created specifically for MediaSPIP and can be easily adapted to fit in with a specific theme.
For older browsers the Flowplayer flash fallback is used.
MediaSPIP allows for media playback on major mobile platforms with the above (...)
Sur d’autres sites (3254)
-
FFMPEG/NVDEC Fails When Under 7 Frames
13 août 2021, par Meme MachineI was looking the examples from NVIDIA's repository, specifically their Encoding and Decoding projects. I downloaded the desktop duplication project, which allows you to capture a certain number of frames from the desktop as raw h264. I also got AppDecode, which decodes and displays frames from an input file. I noticed that if I try and capture only a single frame, it fails to decode the input file.


Here is the output


C:\Users\Admin>C:\Users\Admin\source\repos\video-sdk-samples\Samples\x64.Debug\AppDecD3d -d3d 11 -i C:\Users\Admin\source\repos\video-sdk-samples\nvEncDXGIOutputDuplicationSample\x64\Debug\ddatest_0.h264
GPU in use: NVIDIA GeForce RTX 2080 Super with Max-Q Design
Display with D3D11.
[INFO ][17:59:47] Media format: raw H.264 video (h264)
Session Initialization Time: 39 ms
[INFO ][17:59:47] Video Input Information
 Codec : AVC/H.264
 Frame rate : 30000/1000 = 30 fps
 Sequence : Progressive
 Coded size : [1920, 1088]
 Display area : [0, 0, 1920, 1080]
 Chroma : YUV 420
 Bit depth : 8
Video Decoding Params:
 Num Surfaces : 20
 Crop : [0, 0, 0, 0]
 Resize : 1920x1088
 Deinterlace : Weave

Total frame decoded: 7
Session Deinitialization Time: 10 ms

C:\Users\Admin>C:\Users\Admin\source\repos\video-sdk-samples\Samples\x64.Debug\AppDecD3d -d3d 11 -i C:\Users\Admin\source\repos\video-sdk-samples\nvEncDXGIOutputDuplicationSample\x64\Debug\ddatest_0.h264
GPU in use: NVIDIA GeForce RTX 2080 Super with Max-Q Design
Display with D3D11.
[INFO ][17:59:54] Media format: raw H.264 video (h264)
[h264 @ 0000023B8AB5C3A0] decoding for stream 0 failed
Session Initialization Time: 42 ms
[INFO ][17:59:54] Video Input Information
 Codec : AVC/H.264
 Frame rate : 30000/1000 = 30 fps
 Sequence : Progressive
 Coded size : [1920, 1088]
 Display area : [0, 0, 1920, 1080]
 Chroma : YUV 420
 Bit depth : 8
Video Decoding Params:
 Num Surfaces : 20
 Crop : [0, 0, 0, 0]
 Resize : 1920x1088
 Deinterlace : Weave

Total frame decoded: 6
Session Deinitialization Time: 10 ms


I started from 10 frames and counted down to 6 where it eventually failed. It is important for me to know why this happens, because I plan to implement this decoder into my project, and will be feeding it single frames from a stream.


Oh, and also I noticed the coded size is 1088 by 1920 instead of 1080 according to the output log. Not sure why that is occurring or if it is relevant


-
CUDA_ERORR_INVALID_CONTEXT
15 août 2021, par Meme MachineI am making a desktop sharing application based off of these repositories from NVIDIA.


https://github.com/NVIDIA/video-sdk-samples/tree/master/nvEncDXGIOutputDuplicationSample


https://github.com/NVIDIA/video-sdk-samples/blob/master/Samples/AppDecode/AppDecD3D/


https://github.com/NVIDIA/video-sdk-samples/tree/master/Samples/AppDecode/AppDecMem


I intend to have a setup function that is called once when Remote Desktop is selected, and then a second function that actually displays the received frames which is called when a frame is received


The below functions are nearly identical to the main() and NvD3D() functions found in AppDecD3D and AppDecMem repositories


CUcontext cuContext = NULL; // maybe it has to do with this variable?

int setup()
{
 char szInFilePath[256] = "C:\\Users\\Admin\\Desktop\\test.h264";
 int iGpu = 0;
 int iD3d = 0;
 try
 {
 //ParseCommandLine(argc, argv, szInFilePath, NULL, iGpu, NULL, &iD3d);
 CheckInputFile(szInFilePath);

 ck(cuInit(0));
 int nGpu = 0;
 ck(cuDeviceGetCount(&nGpu));
 if (iGpu < 0 || iGpu >= nGpu)
 {
 std::ostringstream err;
 err << "GPU ordinal out of range. Should be within [" << 0 << ", " << nGpu - 1 << "]" << std::endl;
 throw std::invalid_argument(err.str());
 }
 CUdevice cuDevice = 0;
 ck(cuDeviceGet(&cuDevice, iGpu));
 char szDeviceName[80];
 ck(cuDeviceGetName(szDeviceName, sizeof(szDeviceName), cuDevice));
 std::cout << "GPU in use: " << szDeviceName << std::endl;

 ck(cuCtxCreate(&cuContext, CU_CTX_SCHED_BLOCKING_SYNC, cuDevice));
 //NvDecD3D<framepresenterd3d11>(szInFilePath);

 std::cout << "Display with D3D11." << std::endl;
 }
 catch (const std::exception& ex)
 {
 std::cout << ex.what();
 exit(1);
 }
 return 0;
}

template<class typename="std::enable_if<std::is_base_of<FramePresenterD3D," framepresentertype="framepresentertype">::value>>
int NvDecD3D(char* szInFilePath)
{
 FileDataProvider dp(szInFilePath);
 FFmpegDemuxer demuxer(&dp);
 NvDecoder dec(cuContext, demuxer.GetWidth(), demuxer.GetHeight(), true, FFmpeg2NvCodecId(demuxer.GetVideoCodec()));
 FramePresenterType presenter(cuContext, demuxer.GetWidth(), demuxer.GetHeight());
 CUdeviceptr dpFrame = 0;
 ck(cuMemAlloc(&dpFrame, demuxer.GetWidth() * demuxer.GetHeight() * 4));
 int nVideoBytes = 0, nFrameReturned = 0, nFrame = 0;
 uint8_t* pVideo = NULL, ** ppFrame;

 do
 {
 demuxer.Demux(&pVideo, &nVideoBytes);
 dec.Decode(pVideo, nVideoBytes, &ppFrame, &nFrameReturned);
 if (!nFrame && nFrameReturned)
 LOG(INFO) << dec.GetVideoInfo();

 for (int i = 0; i < nFrameReturned; i++)
 {
 if (dec.GetBitDepth() == 8)
 Nv12ToBgra32((uint8_t*)ppFrame[i], dec.GetWidth(), (uint8_t*)dpFrame, 4 * dec.GetWidth(), dec.GetWidth(), dec.GetHeight());
 else
 P016ToBgra32((uint8_t*)ppFrame[i], 2 * dec.GetWidth(), (uint8_t*)dpFrame, 4 * dec.GetWidth(), dec.GetWidth(), dec.GetHeight());
 presenter.PresentDeviceFrame((uint8_t*)dpFrame, demuxer.GetWidth() * 4);
 }
 nFrame += nFrameReturned;
 } while (nVideoBytes);
 ck(cuMemFree(dpFrame));
 std::cout << "Total frame decoded: " << nFrame << std::endl;
 return 0;
}
</class></framepresenterd3d11>

Notice the line
NvDecD3D<framepresenterd3d11>(szInFilePath);</framepresenterd3d11>? I plan to callNvDecD3D()when a frame is received. So, I commented out the call insetup()and moved it to my asio:async_read function. (see below)

void do_read_body()
 {
 readBuffer.reserve(_read_msg.ReadLength);
 _read_msg.Body = readBuffer.data();
 auto self(shared_from_this());
 asio::async_read(_socket,
 asio::buffer(_read_msg.Body, _read_msg.ReadLength),
 [this, self](std::error_code ec, std::size_t /*length*/)
 {
 if (!ec)
 {
 if (_read_msg.CmdId == 0x5)
 {
 std::cout << "Received a frame" << std::endl;

 NvDecD3D<framepresenterd3d11>(szInFilePath);
 }
 else
 {
 std::cout << std::string(_read_msg.Body, 0, _read_msg.ReadLength) << std::endl;
 }
 
 do_read_header();
 }
 else
 {
 _room.leave(shared_from_this());
 }
 });
 }
</framepresenterd3d11>

However, when I go to execute it, I get
CUDA_ERORR_INVALID_CONTEXTwhencuMemAlloc()is called. If I uncomment the call toNvDecD3D()insidesetup()and call it from there, it does not error however.

Do you have any idea what could be causing this problem ? Perhaps it is related to the ASIO.


-
Create drawbox with ffmpeg between specific seconds (without reencoding whole video - faster)
31 mars 2022, par protterMy plan was to put a transparent red box behind a video. This box should only be present from second 1-45.
But if the videos are 3 hours long, the process takes a long time although it only has to process 45 seconds.


My first attempt takes too long :


ffmpeg -i %1 -vf drawbox=0:9*ih/10:iw:ih/10:t=fill:color=red@0.5:enable='between(t,1,45)' "%~dp0transpred\%~n1%~x1

Then i tried splitting the video into two parts. put the box on the first video, and then put the two back together again.


ffmpeg -ss 00:00:00.0000 -i %1 -to 00:00:45.0000 -vf drawbox=0:9*ih/10:iw:ih/10:t=fill:color=red@0.5:enable='between(t,1,45)' "%~dp0transpred\%~n1A%~x1"

FFMpeg -ss 00:00:45.0000 -i %1 -c:v copy -c:a copy -avoid_negative_ts make_zero "%~dp0transpred\%~n1B%~x1"

But i don't even have to try to put these two together, because they are not separated exactly at the second. I have read that this is due to "timestamps" and the different video and audio streams.


Now I'm trying an approach to create a stream with the bar, and then overlay it with the finished video. I haven't quite managed that yet, and I don't know if it's faster.
Shortening the video is very fast.


EDIT (Added as a replacement for the comment later)


Thanks for your help I have almost done it with a slightly different approach. Unfortunately, the second part now always has no sound. No matter if I put A and B (B no sound) or B and A (A no sound) together.


- 

- First split with mkvmerge so i have no worrys about the keyframes and get the exact time

mkvmerge --split timestamps:00:00:45.100 A.MKV -o splitmkm.mkv - Then add the Bar (Black because of easier testing) :

ffmpeg -i splitmkm-001.mkv -vf drawbox=0:9*ih/10:iw:ih/10:t=fill BAR1.MKV - Merge (mkvmerge ends with error) :

ffmpeg -safe 0 -f concat -i list.txt -c copy output1.mkv








EDIT (Answer to kesh)


This was the error
Again, audio codec config's must match across all your concat files. Thedrawboxchanged the audio Codec from AC-3 to Vorbis.

the procedure is now :


- 

mkvtoolnix\mkvmerge --split timestamps:00:00:05.100 %1 -o A_splitmkm.mkvwith mkvmerge i have an exact split at the time, and i don't have to learn about keyframes.ffmpeg -i A_splitmkm-001.mkv -vf drawbox=0:9*ih/10:iw:ih/10:t=fill:color=red A_BARmkm.MKVcreate the Barffmpeg -i A_BARmkm.MKV -i A_splitmkm-001.mkv -map 0:v -map 1 -map -1:v -c copy A_BARwithAudio.mkvredo the step with the changed audio from drawboxffmpeg -safe 0 -f concat -i list.txt -map 0 -c copy A_output1.mkvmerge










Now everything works.
Thanks alot !


- First split with mkvmerge so i have no worrys about the keyframes and get the exact time

Syndiquer tout le site
