我正在研究用于编码图像的X264 API。
到目前为止，我已经构建了X264库，以下代码片段显示了我的目标：
int frame_size;
x264_t* encoder;
x264_picture_t pic_in, pic_out;
x264_param_t x264Param;
int fps = 20;
int width = 1280;
int height = 720;
x264_nal_t* nals;
int i_nals;
x264_param_default_preset(&x264Param, "veryfast", "zerolatency");
x264Param.i_threads = 1;
x264Param.i_width = 1280;
x264Param.i_height = 720;
x264Param.i_fps_num = fps;
x264Param.i_fps_den = 1;
x264Param.i_keyint_max = fps;
x264Param.b_intra_refresh = 1;
x264Param.rc.i_rc_method = X264_RC_CRF;
x264Param.rc.f_rf_constant = 25;
x264Param.rc.f_rf_constant_max = 35;
x264Param.b_repeat_headers = 1;
x264Param.b_annexb = 1;
x264_param_apply_profile(&x264Param, "baseline");
encoder = x264_encoder_open(&x264Param);
x264_picture_alloc(&pic_in, X264_CSP_BGR, width, height);
/* How to fill in bitmap data? */
frame_size = x264_encoder_encode(encoder, &nals, &i_nals, &pic_in, &pic_out);
if (frame_size >= 0)
{
printf("OK\n");
}
所以我试图编码24位BGR位图图像。但是，x264头文件没有显示用于将位图图像写入编码器的任何API函数。这是怎么做到的？
修改
此代码段似乎有效。我将很感激审查和一些评论。感谢。
int frame_size;
int accum_frame_size;
x264_t* encoder;
x264_picture_t pic_in, pic_out;
x264_param_t x264Param;
int fps = 20;
int width = 1280;
int height = 720;
x264_nal_t* nals;
int i_nals;
int64_t frameCount = 0;
int k;
for (k = 0; k < (1280*3*720); k++)
{
bgr[k] = rand();
}
x264_param_default_preset(&x264Param, "veryfast", "zerolatency");
x264Param.i_threads = 1;
x264Param.i_width = 1280;
x264Param.i_height = 720;
x264Param.i_fps_num = fps;
x264Param.i_fps_den = 1;
x264Param.i_keyint_max = fps;
x264Param.b_intra_refresh = 1;
x264Param.rc.i_rc_method = X264_RC_CRF;
x264Param.i_csp = X264_CSP_BGR;
x264Param.rc.f_rf_constant = 25;
x264Param.rc.f_rf_constant_max = 35;
x264Param.b_repeat_headers = 1;
x264Param.b_annexb = 1;
x264_param_apply_profile(&x264Param, "baseline");
encoder = x264_encoder_open(&x264Param);
x264_picture_alloc(&pic_in, X264_CSP_BGR, width, height);
/* Load 24-bit BGR bitmap */
pic_in.img.i_csp = X264_CSP_BGR;
pic_in.img.i_plane = 1;
pic_in.img.i_stride[0] = 3 * 1280;
pic_in.img.plane[0] = bgr;
pic_in.i_pts = frameCount;
pic_in.i_type = X264_TYPE_AUTO;
pic_out.i_pts = frameCount;
/* Returns a frame size of 912 for first frame in this case */
frame_size = x264_encoder_encode(encoder, &nals, &i_nals, &pic_in, &pic_out);
printf("Decoder returned frame size = %d \n", frame_size);
printf("Decoder returned %d NAL units \n", i_nals);
if (frame_size >= 0)
{
int i;
int j;
accum_frame_size = 0;
for (i = 0; i < i_nals; i++)
{
printf("******************* NAL %d (%d bytes) *******************\n", i, nals[i].i_payload);
for (j = 0; j < nals[i].i_payload; j++)
{
if (j == 0) printf("First 10 bytes: ");
if (j < 10) printf("%02X |", nals[i].p_payload[j]);
accum_frame_size++;
}
printf("\n");
}
}
printf("Verified frame size = %d \n", accum_frame_size);
编辑＃2 编码器输出：
x264 [error]: baseline profile doesn't support 4:4:4
x264 [info]: using cpu capabilities: MMX2 SSE2Fast SSSE3 SSE4.2 AVX
x264 [info]: profile High 4:4:4 Predictive, level 3.1, 4:4:4 8-bit
Decoder returned frame size = 1467194
Decoder returned 4 NAL units
******************* NAL 0 (31 bytes) *******************
First 10 bytes: 00 |00 |00 |01 |67 |F4 |00 |1F |91 |89 |
******************* NAL 1 (8 bytes) *******************
First 10 bytes: 00 |00 |00 |01 |68 |EF |1F |2C |
******************* NAL 2 (595 bytes) *******************
First 10 bytes: 00 |00 |01 |06 |05 |FF |FF |4C |DC |45 |
******************* NAL 3 (1466560 bytes) *******************
First 10 bytes: 00 |00 |01 |65 |88 |82 |0A |FF |F5 |B0 |
Verified frame size = 1467194
不是每个NAL单元都应该以0x00 0x00 0x00 0x01开头吗？
szatmary ：感谢您的宝贵意见。所以你说每个NAL单元不一定以0,0,0,1开头。但是，我对你的答案有点不清楚。您是否暗示使用某种配置时，NAL单元将以0,0,0,1开头？如果是这样，那是哪种配置？我需要确保我在网络上传输到远程接收器的每个NAL单元都以0,0,0,1开头。在探索x264库之前，我使用了x264 exe和管道BMP数据，并从x264进程中编码出数据。然后，我解析了编码器输出，并通过查找0,0,0,1来查找NAL单位。如何使用x264库完成相同的操作？
关于libswscale：
我下载了ffmpeg源并在MINGW中运行configure和make。在完成该过程后，我找不到除了许多.exe文件之外的任何内容。如何构建可在Visual Studio项目中使用的实际静态库（.lib）？
答案 0 :(得分：1)
x264无法编码BGR。您必须将图像转换为YUV 4：2：0p。我建议使用ffmpeg项目中的libswscale来执行此操作。
编辑＃2
不，起始码至少有两个0x00字节，后跟一个0x01。但它可以有任何数字（但通常是两个或三个空字节）4字节版本用于检测串行传输中的字节相关，通常在SPS和PPS NALU之前。