Recherche avancée

Sur d’autres sites (5381)

• OpenCL YUV420P black pixel

  30 avril 2021, par Albert Tinon

  I'm working on an FFmpeg OpenCL filter for converting GoPro Max .360 files in Google EAC projected files.
For doing that I need to "stack" the 2 input streams to a twice height output.
It is working very well at a realtime speed. (The working code is in comment)

  


  For going further I need to replace some pixels with some specific colors.
I wrote some macros for making RGB->YUV conversion. But I get only some green or pink pixel (only grey is OK).
this is my test code (with stacking in comment)

  


  #define Y(R,G,B) 0.299 * R + 0.587 * G + 0.114 * B
#define U(R,G,B) -0.147 * R - 0.289 * G + 0.436 * B
#define V(R,G,B) 0.615 * R - 0.515 * G - 0.100 * B
#define YUV(R,G,B) (float4)(Y(R,G,B),U(R,G,B),V(R,G,B),0)

__kernel void gopromax_stack(__write_only image2d_t dst,
                                __read_only  image2d_t gopromax_front,
                                __read_only  image2d_t gopromax_rear)
{
    const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
                               CLK_FILTER_NEAREST);
    
    float4 val;
    int2 dst_size = get_image_dim(dst);
    int2 loc = (int2)(get_global_id(0), get_global_id(1));
    int split_loc = dst_size.y/2;

        if (loc.y < split_loc)
        {
          // working code for stacking
          //  val = read_imagef(gopromax_front, sampler, (int2)(loc.x, loc.y));

          // testing to put grey (working)
            val = YUV(0.5f,0.5f,0.5f);
        }
        else
        {
          // working code for stacking
          //  val = read_imagef(gopromax_rear, sampler, (int2)(loc.x, loc.y-split_loc));

          // testing to put black (gives green !)
            val = YUV(0,0,0);
        }

    if ((loc.xcode>

  


  I tried many think I cannot succeed to generate black or anything except grey.
What did I make wrong ?
I supposed that my pixels are YUV because I specified yuv420p as the format in my filter :

  


  -filter_complex '[0:0]format=yuv420p,hwupload[a] , [0:4]format=yuv420p,hwupload[b], [a][b]gopromax_opencl, hwdownload,format=yuv420p'


  


  The source streams are in hevc / nv12.

  


  Thanks all for your help.

  


• imgutils : add function to clear an image to black

  22 juillet 2017, par wm4

  imgutils : add function to clear an image to black
  Black isn't always just memset(ptr, 0, size). Limited YUV in particular
  
  requires relatively non-obvious values, and filling a frame with
  
  repeating 0 bytes is disallowed in some contexts. With component sizes
  
  larger than 8 or packed YUV, this can become relatively complicated. So
  
  having a generic function for this seems helpful.
  In order to handle the complex cases in a generic way without destroying
  
  performance, this code attempts to compute a black pixel, and then uses
  
  that value to clear the image data quickly by using a function like
  
  memset.
  Common cases like yuv410p10 or rgba can't be handled with a simple
  
  memset, so there is some code to fill memory with 2/4/8 byte patterns.
  
  For the remaining cases, a generic slow fallback is used.
  Signed-off-by : Anton Khirnov <anton@khirnov.net>
  Merged from Libav commit 45df7adc1d9b7.
  

  • [DH] doc/APIchanges
  • [DH] libavutil/imgutils.c
  • [DH] libavutil/imgutils.h
  • [DH] libavutil/version.h

• Why does an "ffmpeg fail" occur when trying to export frames from my "data_src.mp4" file, while using DeepFaceLab ?

  28 août 2020, par Toby Ryberg

  I've been following this step-by-step videoin installing and using DeepFaceLab on my Macbook Pro, which uses OS Catalina. I've been stuck at the step which begins at the 4:42 mark, where he enters this code :

  


  python main.py videoed extract-video —input-file WORKSPACE/data_src.mp4 —output-dir WORKSPACE/data_src

  


  After entering the command, I'm prompted to enter an FPS value ; no problem. Next, it asks for a file output type, which is where I get tied up— as far as I know. It appears as though I can enter either png or jpeg, but receive the same error message, which looks like this :

  


  / !\ ffmpeg fail, job commandline :['ffmpeg', '-i', '/Users/tobyryberg/Documents/deepfakes/DeepFaceLab/WORKSPACE/data_src.mp4', '-pix_fmt', 'rgb24', '/Users/tobyryberg/Documents/deepfakes/DeepFaceLab/WORKSPACE/data_src/%5d.png']

  


  I have no idea what went wrong, as I BELIEVE that I have been doing everything right thus far. If this is at all necessary information, I've opted to use an 18+ minute long video for my source clip, which is an .mp4 file, without audio. If there's someone out there who has the patience to work with me and figure out what went wrong, I'll be forever grateful !