Recherche avancée

Sur d’autres sites (8362)

• ffmpeg app using node occasionally crashes as file doesn't appear to be read correctly

  31 mai 2022, par Zabs

  I have an simple Node application that allows me to pass an AWS S3 URL link to a file (in this case video files). It uses the FFMPEG library to read the video file and return data like codecs, duration, bitrate etc..

  


  The script is called from PHP script which in turn send the data to the Node endpoint and passes the Amazon S3 URL to node. Sometimes for no obvious reasons the video file fails to return the expected values regarding container, codec, duration etc... and just returns '0'. But when I try the exact same file/request again it returns this data correctly e.g container:mp4

  


  I'm not sure but I think the script somehow needs the createWriteStream to be closed but I cannot be sure, the problem is the issue I have found doesn't happen all the time but sporadically so its hard to get to the issue when its difficult to replicate it.

  


  Any ideas ?

  


  router.post('/', async function(req, res) {
  const fileURL = new URL(req.body.file);
  var path = fileURL.pathname;
  path = 'tmp/'+path.substring(1);    // removes the initial / from the path

  let file = fs.createWriteStream(path);  // create the file locally
  const request = https.get(fileURL, function(response) {
    response.pipe(file);
  });
  
  // after file has saved
  file.on('finish', function () {
    var process = new ffmpeg(path);
    process.then(function (video) {
      let metadata = formatMetadata(video.metadata);

      res.send ({
        status: '200',
        data: metadata,
        errors: errors,
        response: 'success'
      });

    }, function (err) {
      console.warn('Error: ' + err);

      res.send ({
        status: '400',
        data: 'Something went wrong processing this video',
        response: 'fail',
      });
    });
  });

  file.on('error', function (err) {
    console.warn(err);
  });

});

function formatMetadata(metadata) {
  const data = {
    'video' : metadata.video,
    'audio' : metadata.audio,
    'duration' : metadata.duration
  };
  return data;
}


  


  // Expected output

  


  {"data":{"video":{"container":"mov","bitrate":400,"stream":0,"codec":"h264","resolution":{"w":1280,"h":720},"resolutionSquare":{"w":1280,"h":720},"aspect":{"x":16,"y":9,"string":"16:9","value":1.7777777777777777},"rotate":0,"fps":25,"pixelString":"1:1","pixel":1},"audio":{"codec":"aac","bitrate":"127","sample_rate":44100,"stream":0,"channels":{"raw":"stereo","value":2}},"duration":{"raw":"00:00:25.68","seconds":25}}


  


  // Actual output

  


  {"data":{"video":{"container":"","bitrate":0,"stream":0,"codec":"","resolution":{"w":0,"h":0},"resolutionSquare":{"w":0,"h":null},"aspect":{},"rotate":0,"fps":0,"pixelString":"","pixel":0},"audio":{"codec":"","bitrate":"","sample_rate":0,"stream":0,"channels":{"raw":"","value":""}},"duration":{"raw":"","seconds":0}}


  


  Note - this happens sporadically

  


• lavc/aarch64 : motion estimation functions in neon

  26 juin 2022, par Swinney, Jonathan

  lavc/aarch64 : motion estimation functions in neon
   - ff_pix_abs16_neon
  
   - ff_pix_abs16_xy2_neon
  In direct micro benchmarks of these ff functions verses their C implementations,
  
  these functions performed as follows on AWS Graviton 3.
  ff_pix_abs16_neon :
  
  pix_abs_0_0_c : 141.1
  
  pix_abs_0_0_neon : 19.6
  ff_pix_abs16_xy2_neon :
  
  pix_abs_0_3_c : 269.1
  
  pix_abs_0_3_neon : 39.3
  Tested with :
  
  ./tests/checkasm/checkasm —test=motion —bench —disable-linux-perf
  Signed-off-by : Jonathan Swinney <jswinney@amazon.com>
  
  Signed-off-by : Martin Storsjö <martin@martin.st>
  

  • [DH] libavcodec/aarch64/Makefile
  • [DH] libavcodec/aarch64/me_cmp_init_aarch64.c
  • [DH] libavcodec/aarch64/me_cmp_neon.S
  • [DH] libavcodec/me_cmp.c
  • [DH] libavcodec/me_cmp.h
  • [DH] tests/checkasm/Makefile
  • [DH] tests/checkasm/checkasm.c
  • [DH] tests/checkasm/checkasm.h
  • [DH] tests/checkasm/motion.c
  • [DH] tests/fate/checkasm.mak

• AWS EC2 and OpenCV with UDP consumer

  22 juillet 2021, par NoobZik

  I have two EC2

  


  

  • A That grabs the video feed from Amazon Kinesis Video to send it into via UDP to B

  


  • B the consumer which it will grab the UDP feed from the A

  


  


  While debugging my opencv not grabbing the UDP feed, I suspect something is wrong with ffpmeg, so I am going to make sure that ffplay can read packets sent from A.

  


  Now when I try to read udp packets with ffplay {adress-ip-with-port}I have the following error :

  


  Could not initialize SDL - No available video device
(Did you set the DISPLAY variable?)


  


  How do I fix this one since there is no display on EC2.

  


  Also if it can help, this is my code for the consumer part

  


   Create your views here.
from django.http import HttpResponse
from django.template import loader
from django.shortcuts import render
# from .models import Vehicule
import cv2
import threading
from django.views.decorators import gzip
from django.http import StreamingHttpResponse


@gzip.gzip_page
def webcam(request):
    try:
        cam = VideoCamera()
        return StreamingHttpResponse(gen(cam), content_type="multipart/x-mixed-replace;boundary=frame")
    except:
        pass
    return render(request, 'vebcam.html')

#capture video
class VideoCamera(object):
    def __init__(self):
        self.video = cv2.VideoCapture('udp://172.31.57.243:55055', cv2.CAP_FFMPEG)
        (self.grabbed, self.frame) = self.video.read()
        threading.Thread(target=self.update, args=()).start()

    def __del__(self):
        self.video.release()

    def get_frame(self):
        image = self.frame
        _, jpeg = cv2.imencode('.jpg', image)
        return jpeg.tobytes()

    def update(self):
        while True:
            (self.grabbed, self.frame) = self.video.read()

def gen(camera):
    while True:
        frame = camera.get_frame()
        yield (b'--frame\r\n'b'Content-Type: image/jpeg\r\n\r\n' + frame + b'\r\n\r\n')


  


  It runs fine on local but not on EC2

  


  This is the log that led me to check ffplay

  


  Traceback (most recent call last):
  File "/usr/lib/python3.8/wsgiref/handlers.py", line 138, in run
    self.finish_response()
  File "/usr/lib/python3.8/wsgiref/handlers.py", line 183, in finish_response
    for data in self.result:
  File "/home/ubuntu/.local/lib/python3.8/site-packages/django/utils/text.py", line 304, in compress_sequence
    for item in sequence:
  File "/home/ubuntu/site/mysite/mysite/views.py", line 42, in gen
    frame = camera.get_frame()
  File "/home/ubuntu/site/mysite/mysite/views.py", line 33, in get_frame
    _, jpeg = cv2.imencode('.jpg', image)
cv2.error: OpenCV(4.5.3) /tmp/pip-req-build-xw6jtoah/opencv/modules/imgcodecs/src/loadsave.cpp:978: error: (-215:Assertion failed) !image.empty() in function 'imencode'