Recherche avancée

Sur d’autres sites (7884)

• Direct Show Video capture performance

  23 mai 2019, par Barry Andrews

  I originally started out looking for an example of how I can use FFMPEG in c++ builder to create an application to record from usb capture device and playback video because of apparrent poor performance

  I tried Mitov components, Datastead, FFMPEGVCL and winsoft camera which use directshow but their capture performance seemed poor.

  I need to capture 1920x1080 at up to 60fps into a compressed format and play this back later at both normal speed and slow speed.

  What I found was that DirectShow itself has a number of limitations which can be improved by adding things like FFMPEG, but ultimately PC hardware, in particular the HDD and processor limit capture capability.

  1920x1080 60fps is basically the upper end for DirectShow so you need to have best performing hardware in order to achieve this sort of performance @Spektre kindly gave me examples of DirectShow using the API direct which were good for comparison with the purchased components.

  Using this and comparing to the components I found that MITOV has a major issue with regards the larger video sizes and frame rates. Using this 1920x108 30fps and 60fps can be previewed but they have a massive delay between video feed and preview (5 or 6 seconds). The other components performed similar to the API direct method with only minor variations in performance. None were able to capture and record 1920x108 60fps with any sort of compression filter without large frame drops and very jerky preview.

• swscale/utils : add helper function to infer colorspace metadata

  28 novembre 2024, par Niklas Haas

  swscale/utils : add helper function to infer colorspace metadata
  Logic is loosely on equivalent decisions in libplacebo. The basic idea is to try
  
  and be a bit conservative by treating AVCOL_*_UNSPECIFIED as a no-op, unless the
  
  other primaries set are non-standard / wide-gamut or HDR. This helps avoid
  
  unintended or unexpected colorspace conversions, while forcing it in cases where
  
  we are almost certain it is needed. The major departure from libplacebo semantics
  
  is that we no default to a 1000:1 contrast ration for SDR displays, instead modelling
  
  them as idealized devices with an infinite contrast ratio.
  In either case, setting SWS_STRICT overrides this behavior in favor of always
  
  requiring explicit colorspace metadata.
  

  • [DH] libswscale/utils.c
  • [DH] libswscale/utils.h

• Multiframe Dicom file not readable by some viewers

  12 mai 2023, par PizzaPasNette

  I have developed a program in C++ that works in 2 major steps :

  


  


  • extract frames from a video with ffmpeg and convert them to jpeg
ffmpeg -i videofile.mp4 -q:v 1 -r desired_framerate tmp_04%d.jpeg

    


  


  • create a multiframe Dicom file from those jpeg images via DCMTK

    


    //Example for one image&#xA;&#xA;// Get the n-th image&#xA;QFile* imageFile = new QFile("Path/To/Image");&#xA;&#xA;// Read the image inside a byte array&#xA;QByteArray * ba = new QByteArray();&#xA;imageFile->open(QIODevice::ReadOnly);&#xA;*ba = imageFile->readAll();&#xA;&#xA;// Insert the byte array in the Dicom file&#xA;cond = jpegPixelItem->putUint8Array(reinterpret_cast<uint8>(ba->data()), ba->size());&#xA;&#xA;// Rinse and repeat&#xA;</uint8>

    


  


  


  It works well and can be displayed on most viewers but some complain about the Transfer Syntax being wrong.
As of now, I'm using the "JPEGLossless:Non-hierarchical-1stOrderPrediction" but It seems some viewers can't read my Dicom file unless I change the Transfer Syntax to "JPEG Baseline (Process 1)" which is a lossy Transfer Syntax.

  


  I assume the problem comes from the first step of the program with ffmpeg that might be creating lossy JPEG images but I'm not sure because I'm using the "best" q:v scale as I've seen in other posts on the subject.

  


  Therefore, my 2 main questions are :

  


  

  • why do some viewers have no problems displaying the dicom file whereas some others can't ?

  


  • is there any way to obtain lossless JPEG images from a ffmpeg command ?