[FFmpeg] RGBA and YUV storage methods

RGBA

The data is stored in RGBARGBARGBA... format, and the starting address is frame->data[0]. In fact, in addition to RGBA, there are BGRA, AGBR, ARGB, GBRA. For details, please refer to the source code ffmpeg-4.3.1/libavfilter/drawutils.c

Traverse access to RGBA pixel frames

for (int y = 0; y < frame->height; y++) {
    
    
    row[0] = frame->data[0] + y * frame->linesize[0];
    for (int x = 0; x < frame->width; x++) {
    
    
        row[0][(x << 2) + 0]; //R
        row[0][(x << 2) + 1]; //G
        row[0][(x << 2) + 2]; //B
        row[0][(x << 2) + 3]; //A
    }
}

HOLE

The four types of data Y, U, V, and A are stored separately, and the starting addresses are frame->data[0], frame->data[1], frame->data[2], frame->data[3].

Traverse access to YUV pixel frame

const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(frame->format);
int hsub = desc->log2_chroma_w;
int vsub = desc->log2_chroma_h;
int have_alpha = desc->flags & AV_PIX_FMT_FLAG_ALPHA;

if (have_alpha) {
    
    
    for (int y = 0; y < frame->height; y++) {
    
    
        row[0] = frame->data[0] + y * frame->linesize[0];
        row[1] = frame->data[1] + (y >> vsub) * frame->linesize[1];
        row[2] = frame->data[2] + (y >> vsub) * frame->linesize[2]
        row[3] = frame->data[3] + y * frame->linesize[3];

        for (int x = 0; x < frame->width; x++) {
    
    
            row[0][x        ]; // Y
            row[1][x >> hsub]; // U
            row[2][x >> hsub]; // V
            row[3][x        ]; // A
        }
    }
} else {
    
    
    for (int y = 0; y < frame->height; y++) {
    
    
        row[0] = frame->data[0] + y * frame->linesize[0];
        row[1] = frame->data[1] + (y >> vsub) * frame->linesize[1];
        row[2] = frame->data[2] + (y >> vsub) * frame->linesize[2];

        for (int x = 0; x < frame->width; x++) {
    
    
            row[0][x        ]; // Y
            row[1][x >> hsub]; // U
            row[2][x >> hsub]; // V
        }
    }
}