[Libav-user] How to Convert AV_PIX_FMT_RGB24 to AV_PIX_FMT_YUV420P

[Hector Alonso]

There's not a better type, it depends on the whole system.

For instance, that's a part of a video capture utility. I recieve the
"ptBuffer", or the source image, in YUV422 (which is a very common pixel
format in capture cards and video cameras). I do the rescale as well as the
pixel format conversion (using pixel shaders) in the GPU, this was a quick
test I made...

You have to identify your pixel format and your data structure and convert
it to the desired one.

A part from the pixel format (RGB, BGR, RGBA, ...) you need to know whether
it's planar (RRR....GGGG....BBBB) or linear (RGBRGBRGB....). The pixel
format descriptors also include this information (represented as a "P" at
the end), you should check the documentation.

In your case:
AV_PIX_FMT_RGB24 ---> RGBRGBRGB...
AV_PIX_FMT_YUV420P ----> YYYYYY.....UUUUUUUU........VVVVVVVVV.......
If you use my function to make the conversion it allocates the structure
for you. Check the linesizes of the resulting picture for a better
understanding of what's going on.


On Fri, Feb 15, 2013 at 12:16 AM, Chris Share <cpsmusic at yahoo.com> wrote:

> Thanks for the sample code.
>
> I have a couple of further questions:
>
> 1. If the input data is in a vector of unsigned chars (RGBRGB...), what is
> the best pixel format to use?
>
> 2. The input data is represented by the following method parameter: void*
> ptInData - I'm not clear as to how the vector is converted to an
> appropriate type - should it be an array of chars (RGBRGB...) or does the
> data need to be arranged differently (RRR...GGG...BBB...)?
>
> Cheers,
>
> Chris
>
> ________________________________
> From: Hector Alonso <halonso at vpod.tv>
> To: "This list is about using libavcodec, libavformat, libavutil,
> libavdevice and libavfilter." <libav-user at ffmpeg.org>
> Sent: Thursday, 14 February 2013 8:38 PM
> Subject: Re: [Libav-user] How to Convert AV_PIX_FMT_RGB24 to
> AV_PIX_FMT_YUV420P
>
>
> Hi Chris,
>
> You can implement some kind of function like:
>
> // dependencies into C++
> extern"C"{
> #include<libavformat/avformat.h>
> #include<libswscale/swscale.h>
> #include<libavutil/pixdesc.h>
> #include<libavutil/samplefmt.h>
> #include<libavutil/intreadwrite.h>
> }
>
> /**
> *Thisfunctionconvertsanimage(uncompressedbuffer)fromaninput
> *specifiedpixelformattoanoutputspecifiedpixelformatandbuffer.
> *dependenciesLibAV.
> *@param[in]eInFormatinputimagePixelformat at seeLibaAVPixelFormatenumtype.
> *@param[in]iInWidthinputimagewidth
> *@param[in]iInHeightinputimageheight
> *@param[in]ptInDatainputimagebuffer
> *@param[in]eOutFormatoutputimagePixelformat at seeLibaAVPixelFormatenumtype.
> *@param[in]iOutWidthoutputimagewidth
> *@param[in]iOutHeightoutputimageheight
> *@param[out]srcoutputimagebuffer
> *@returntrueifeverythingwascorrect,falseotherwise)
> **/
>
> boolConvertImage1(PixelFormateInFormat,intiInWidth,intiInHeight,void*ptInData,PixelFormateOutFormat,intiOutWidth,intiOutHeight,AVPicture*src)
> {
> SwsContext*ptImgConvertCtx;// Frame conversion context
>
> AVPictureptPictureIn;
> uint8_t*ptBufferIn;
>
> //Initialize convert context
> //------------------
> ptImgConvertCtx=sws_getContext(iInWidth,iInHeight,eInFormat,// (source
> format)
> iOutWidth,iOutHeight,eOutFormat,// (dest format)
> SWS_BICUBIC,NULL,NULL,NULL);
>
> // Init input frame:
> //------------------
> // Allocate an AVFrame structure
>
>
> // Determine required buffer size and allocate buffer
> // int iNumBytesIn=avpicture_get_size(eInFormat, iInWidth,iInHeight);
>
> ptBufferIn=(uint8_t*)(ptInData);
>
> // Assign appropriate parts of buffer to image planes in pFrameOut
> avpicture_fill(&ptPictureIn,ptBufferIn,eInFormat,iInWidth,iInHeight);
>
> // Do conversion:
> //------------------
> intiRes=sws_scale(ptImgConvertCtx,
> ptPictureIn.data,//src
> ptPictureIn.linesize,
> 0,
> iInHeight,
> src->data,//dst
> src->linesize);
>
>
> //Free memory
> sws_freeContext(ptImgConvertCtx);
>
> //Check result:
> if(iRes==iOutHeight)
> returntrue;
>
> returnfalse;
> }
>
> And use it like this (in this example is used to downscale, but you can
> use it for rescaling and or pixel format conversions)
>
> // Downscale
> AVPicturesrc;
> intiWidth=m_ptCurrentVideoMode->getWidth()/DL_LOW_DEFINITION_DEN;
>
> intiHeight=(m_ptCurrentVideoMode->getHeight()*iWidth)/m_ptCurrentVideoMode->getWidth();
> iWidth=(floor(iWidth/2))*2;
> iHeight=(floor(iHeight/2))*2;
>
> avpicture_alloc(&src,AV_PIX_FMT_UYVY422,iWidth,iHeight);
>
> if(!ConvertImage1(AV_PIX_FMT_UYVY422,
> m_ptCurrentVideoMode->getWidth(),
> m_ptCurrentVideoMode->getHeight(),
> (void*)ptBuffer,
> AV_PIX_FMT_UYVY422,
> iWidth,iHeight,
> &src))
> {
> avpicture_free(&src);
> return;
> }
>
> // ... whatever ...
>
> // Free aux frame
> avpicture_free(&src);
> I hope it helps!
>
>
> On Thu, Feb 14, 2013 at 10:00 AM, Chris Share <cpsmusic at yahoo.com> wrote:
>
> Hi,
> >
> >I'm currently trying to implement file export for the open source
> animation program Pencil2D. This involves converting RGB (0 - 255) image
> data to a suitable movie format.
> >
> >The examples in the source tree have been very helpful however I still
> have some questions:
> >
> >The scaling_video.c is close to what I need however the conversion is the
> opposite of what I want. What I'm not clear about is how to change the
> "fill_yuv_image" function to something like "fill_rgb_image". How does the
> RGB data get written into the "uint8_t *data[4]"? Is it written
> consecutively (all R values get written to data[0], all G values to
> data[1], etc.)?
> >
> >Cheers,
> >
> >Chris
> >_______________________________________________
> >Libav-user mailing list
> >Libav-user at ffmpeg.org
> >http://ffmpeg.org/mailman/listinfo/libav-user
> >
>
> _______________________________________________
> Libav-user mailing list
> Libav-user at ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/libav-user
> _______________________________________________
> Libav-user mailing list
> Libav-user at ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/libav-user
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://ffmpeg.org/pipermail/libav-user/attachments/20130215/5fdb8877/attachment.html>