2 * copyright (c) 2015 Rick Kern <kernrj@gmail.com>
4 * This file is part of FFmpeg.
6 * FFmpeg is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2.1 of the License, or (at your option) any later version.
11 * FFmpeg is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with FFmpeg; if not, write to the Free Software
18 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
21 #include <VideoToolbox/VideoToolbox.h>
22 #include <CoreVideo/CoreVideo.h>
23 #include <CoreMedia/CoreMedia.h>
24 #include <TargetConditionals.h>
25 #include <Availability.h>
27 #include "libavutil/opt.h"
28 #include "libavutil/avassert.h"
29 #include "libavutil/avstring.h"
30 #include "libavcodec/avcodec.h"
31 #include "libavutil/pixdesc.h"
39 #if !HAVE_KCMVIDEOCODECTYPE_HEVC
40 enum { kCMVideoCodecType_HEVC = 'hvc1' };
43 #if !HAVE_KCVPIXELFORMATTYPE_420YPCBCR10BIPLANARVIDEORANGE
44 enum { kCVPixelFormatType_420YpCbCr10BiPlanarFullRange = 'xf20' };
45 enum { kCVPixelFormatType_420YpCbCr10BiPlanarVideoRange = 'x420' };
48 typedef OSStatus (*getParameterSetAtIndex)(CMFormatDescriptionRef videoDesc,
49 size_t parameterSetIndex,
50 const uint8_t **parameterSetPointerOut,
51 size_t *parameterSetSizeOut,
52 size_t *parameterSetCountOut,
53 int *NALUnitHeaderLengthOut);
55 //These symbols may not be present
57 CFStringRef kCVImageBufferColorPrimaries_ITU_R_2020;
58 CFStringRef kCVImageBufferTransferFunction_ITU_R_2020;
59 CFStringRef kCVImageBufferYCbCrMatrix_ITU_R_2020;
61 CFStringRef kVTCompressionPropertyKey_H264EntropyMode;
62 CFStringRef kVTH264EntropyMode_CAVLC;
63 CFStringRef kVTH264EntropyMode_CABAC;
65 CFStringRef kVTProfileLevel_H264_Baseline_4_0;
66 CFStringRef kVTProfileLevel_H264_Baseline_4_2;
67 CFStringRef kVTProfileLevel_H264_Baseline_5_0;
68 CFStringRef kVTProfileLevel_H264_Baseline_5_1;
69 CFStringRef kVTProfileLevel_H264_Baseline_5_2;
70 CFStringRef kVTProfileLevel_H264_Baseline_AutoLevel;
71 CFStringRef kVTProfileLevel_H264_Main_4_2;
72 CFStringRef kVTProfileLevel_H264_Main_5_1;
73 CFStringRef kVTProfileLevel_H264_Main_5_2;
74 CFStringRef kVTProfileLevel_H264_Main_AutoLevel;
75 CFStringRef kVTProfileLevel_H264_High_3_0;
76 CFStringRef kVTProfileLevel_H264_High_3_1;
77 CFStringRef kVTProfileLevel_H264_High_3_2;
78 CFStringRef kVTProfileLevel_H264_High_4_0;
79 CFStringRef kVTProfileLevel_H264_High_4_1;
80 CFStringRef kVTProfileLevel_H264_High_4_2;
81 CFStringRef kVTProfileLevel_H264_High_5_1;
82 CFStringRef kVTProfileLevel_H264_High_5_2;
83 CFStringRef kVTProfileLevel_H264_High_AutoLevel;
84 CFStringRef kVTProfileLevel_H264_Extended_5_0;
85 CFStringRef kVTProfileLevel_H264_Extended_AutoLevel;
87 CFStringRef kVTProfileLevel_HEVC_Main_AutoLevel;
88 CFStringRef kVTProfileLevel_HEVC_Main10_AutoLevel;
90 CFStringRef kVTCompressionPropertyKey_RealTime;
92 CFStringRef kVTVideoEncoderSpecification_EnableHardwareAcceleratedVideoEncoder;
93 CFStringRef kVTVideoEncoderSpecification_RequireHardwareAcceleratedVideoEncoder;
95 getParameterSetAtIndex CMVideoFormatDescriptionGetHEVCParameterSetAtIndex;
98 #define GET_SYM(symbol, defaultVal) \
100 CFStringRef* handle = (CFStringRef*)dlsym(RTLD_DEFAULT, #symbol); \
102 compat_keys.symbol = CFSTR(defaultVal); \
104 compat_keys.symbol = *handle; \
107 static pthread_once_t once_ctrl = PTHREAD_ONCE_INIT;
109 static void loadVTEncSymbols(){
110 compat_keys.CMVideoFormatDescriptionGetHEVCParameterSetAtIndex =
111 (getParameterSetAtIndex)dlsym(
113 "CMVideoFormatDescriptionGetHEVCParameterSetAtIndex"
116 GET_SYM(kCVImageBufferColorPrimaries_ITU_R_2020, "ITU_R_2020");
117 GET_SYM(kCVImageBufferTransferFunction_ITU_R_2020, "ITU_R_2020");
118 GET_SYM(kCVImageBufferYCbCrMatrix_ITU_R_2020, "ITU_R_2020");
120 GET_SYM(kVTCompressionPropertyKey_H264EntropyMode, "H264EntropyMode");
121 GET_SYM(kVTH264EntropyMode_CAVLC, "CAVLC");
122 GET_SYM(kVTH264EntropyMode_CABAC, "CABAC");
124 GET_SYM(kVTProfileLevel_H264_Baseline_4_0, "H264_Baseline_4_0");
125 GET_SYM(kVTProfileLevel_H264_Baseline_4_2, "H264_Baseline_4_2");
126 GET_SYM(kVTProfileLevel_H264_Baseline_5_0, "H264_Baseline_5_0");
127 GET_SYM(kVTProfileLevel_H264_Baseline_5_1, "H264_Baseline_5_1");
128 GET_SYM(kVTProfileLevel_H264_Baseline_5_2, "H264_Baseline_5_2");
129 GET_SYM(kVTProfileLevel_H264_Baseline_AutoLevel, "H264_Baseline_AutoLevel");
130 GET_SYM(kVTProfileLevel_H264_Main_4_2, "H264_Main_4_2");
131 GET_SYM(kVTProfileLevel_H264_Main_5_1, "H264_Main_5_1");
132 GET_SYM(kVTProfileLevel_H264_Main_5_2, "H264_Main_5_2");
133 GET_SYM(kVTProfileLevel_H264_Main_AutoLevel, "H264_Main_AutoLevel");
134 GET_SYM(kVTProfileLevel_H264_High_3_0, "H264_High_3_0");
135 GET_SYM(kVTProfileLevel_H264_High_3_1, "H264_High_3_1");
136 GET_SYM(kVTProfileLevel_H264_High_3_2, "H264_High_3_2");
137 GET_SYM(kVTProfileLevel_H264_High_4_0, "H264_High_4_0");
138 GET_SYM(kVTProfileLevel_H264_High_4_1, "H264_High_4_1");
139 GET_SYM(kVTProfileLevel_H264_High_4_2, "H264_High_4_2");
140 GET_SYM(kVTProfileLevel_H264_High_5_1, "H264_High_5_1");
141 GET_SYM(kVTProfileLevel_H264_High_5_2, "H264_High_5_2");
142 GET_SYM(kVTProfileLevel_H264_High_AutoLevel, "H264_High_AutoLevel");
143 GET_SYM(kVTProfileLevel_H264_Extended_5_0, "H264_Extended_5_0");
144 GET_SYM(kVTProfileLevel_H264_Extended_AutoLevel, "H264_Extended_AutoLevel");
146 GET_SYM(kVTProfileLevel_HEVC_Main_AutoLevel, "HEVC_Main_AutoLevel");
147 GET_SYM(kVTProfileLevel_HEVC_Main10_AutoLevel, "HEVC_Main10_AutoLevel");
149 GET_SYM(kVTCompressionPropertyKey_RealTime, "RealTime");
151 GET_SYM(kVTVideoEncoderSpecification_EnableHardwareAcceleratedVideoEncoder,
152 "EnableHardwareAcceleratedVideoEncoder");
153 GET_SYM(kVTVideoEncoderSpecification_RequireHardwareAcceleratedVideoEncoder,
154 "RequireHardwareAcceleratedVideoEncoder");
157 typedef enum VT_H264Profile {
166 typedef enum VTH264Entropy{
172 typedef enum VT_HEVCProfile {
179 static const uint8_t start_code[] = { 0, 0, 0, 1 };
181 typedef struct ExtraSEI {
186 typedef struct BufNode {
187 CMSampleBufferRef cm_buffer;
189 struct BufNode* next;
193 typedef struct VTEncContext {
195 enum AVCodecID codec_id;
196 VTCompressionSessionRef session;
197 CFStringRef ycbcr_matrix;
198 CFStringRef color_primaries;
199 CFStringRef transfer_function;
200 getParameterSetAtIndex get_param_set_func;
202 pthread_mutex_t lock;
203 pthread_cond_t cv_sample_sent;
210 int64_t frame_ct_out;
220 int64_t frames_before;
221 int64_t frames_after;
228 bool warned_color_range;
230 /* can't be bool type since AVOption will access it as int */
234 static int vtenc_populate_extradata(AVCodecContext *avctx,
235 CMVideoCodecType codec_type,
236 CFStringRef profile_level,
237 CFNumberRef gamma_level,
238 CFDictionaryRef enc_info,
239 CFDictionaryRef pixel_buffer_info);
242 * NULL-safe release of *refPtr, and sets value to NULL.
244 static void vt_release_num(CFNumberRef* refPtr){
253 static void set_async_error(VTEncContext *vtctx, int err)
257 pthread_mutex_lock(&vtctx->lock);
259 vtctx->async_error = err;
261 info = vtctx->q_head;
262 vtctx->q_head = vtctx->q_tail = NULL;
265 BufNode *next = info->next;
266 CFRelease(info->cm_buffer);
271 pthread_mutex_unlock(&vtctx->lock);
274 static void clear_frame_queue(VTEncContext *vtctx)
276 set_async_error(vtctx, 0);
279 static int vtenc_q_pop(VTEncContext *vtctx, bool wait, CMSampleBufferRef *buf, ExtraSEI **sei)
283 pthread_mutex_lock(&vtctx->lock);
285 if (vtctx->async_error) {
286 pthread_mutex_unlock(&vtctx->lock);
287 return vtctx->async_error;
290 if (vtctx->flushing && vtctx->frame_ct_in == vtctx->frame_ct_out) {
293 pthread_mutex_unlock(&vtctx->lock);
297 while (!vtctx->q_head && !vtctx->async_error && wait && !vtctx->flushing) {
298 pthread_cond_wait(&vtctx->cv_sample_sent, &vtctx->lock);
301 if (!vtctx->q_head) {
302 pthread_mutex_unlock(&vtctx->lock);
307 info = vtctx->q_head;
308 vtctx->q_head = vtctx->q_head->next;
309 if (!vtctx->q_head) {
310 vtctx->q_tail = NULL;
313 vtctx->frame_ct_out++;
314 pthread_mutex_unlock(&vtctx->lock);
316 *buf = info->cm_buffer;
319 } else if (info->sei) {
320 if (info->sei->data) av_free(info->sei->data);
329 static void vtenc_q_push(VTEncContext *vtctx, CMSampleBufferRef buffer, ExtraSEI *sei)
331 BufNode *info = av_malloc(sizeof(BufNode));
333 set_async_error(vtctx, AVERROR(ENOMEM));
338 info->cm_buffer = buffer;
342 pthread_mutex_lock(&vtctx->lock);
343 pthread_cond_signal(&vtctx->cv_sample_sent);
345 if (!vtctx->q_head) {
346 vtctx->q_head = info;
348 vtctx->q_tail->next = info;
351 vtctx->q_tail = info;
353 pthread_mutex_unlock(&vtctx->lock);
356 static int count_nalus(size_t length_code_size,
357 CMSampleBufferRef sample_buffer,
364 size_t src_size = CMSampleBufferGetTotalSampleSize(sample_buffer);
365 CMBlockBufferRef block = CMSampleBufferGetDataBuffer(sample_buffer);
367 if (length_code_size > 4)
368 return AVERROR_INVALIDDATA;
370 while (offset < src_size) {
375 status = CMBlockBufferCopyDataBytes(block,
380 for (i = 0; i < length_code_size; i++) {
382 box_len |= size_buf[i];
385 curr_src_len = box_len + length_code_size;
386 offset += curr_src_len;
395 static CMVideoCodecType get_cm_codec_type(enum AVCodecID id)
398 case AV_CODEC_ID_H264: return kCMVideoCodecType_H264;
399 case AV_CODEC_ID_HEVC: return kCMVideoCodecType_HEVC;
405 * Get the parameter sets from a CMSampleBufferRef.
406 * @param dst If *dst isn't NULL, the parameters are copied into existing
407 * memory. *dst_size must be set accordingly when *dst != NULL.
408 * If *dst is NULL, it will be allocated.
409 * In all cases, *dst_size is set to the number of bytes used starting
412 static int get_params_size(
413 AVCodecContext *avctx,
414 CMVideoFormatDescriptionRef vid_fmt,
417 VTEncContext *vtctx = avctx->priv_data;
418 size_t total_size = 0;
420 int is_count_bad = 0;
423 status = vtctx->get_param_set_func(vid_fmt,
435 for (i = 0; i < ps_count || is_count_bad; i++) {
438 status = vtctx->get_param_set_func(vid_fmt,
446 * When ps_count is invalid, status != 0 ends the loop normally
447 * unless we didn't get any parameter sets.
449 if (i > 0 && is_count_bad) status = 0;
454 total_size += ps_size + sizeof(start_code);
458 av_log(avctx, AV_LOG_ERROR, "Error getting parameter set sizes: %d\n", status);
459 return AVERROR_EXTERNAL;
466 static int copy_param_sets(
467 AVCodecContext *avctx,
468 CMVideoFormatDescriptionRef vid_fmt,
472 VTEncContext *vtctx = avctx->priv_data;
474 int is_count_bad = 0;
479 status = vtctx->get_param_set_func(vid_fmt,
492 for (i = 0; i < ps_count || is_count_bad; i++) {
497 status = vtctx->get_param_set_func(vid_fmt,
504 if (i > 0 && is_count_bad) status = 0;
509 next_offset = offset + sizeof(start_code) + ps_size;
510 if (dst_size < next_offset) {
511 av_log(avctx, AV_LOG_ERROR, "Error: buffer too small for parameter sets.\n");
512 return AVERROR_BUFFER_TOO_SMALL;
515 memcpy(dst + offset, start_code, sizeof(start_code));
516 offset += sizeof(start_code);
518 memcpy(dst + offset, ps, ps_size);
519 offset = next_offset;
523 av_log(avctx, AV_LOG_ERROR, "Error getting parameter set data: %d\n", status);
524 return AVERROR_EXTERNAL;
530 static int set_extradata(AVCodecContext *avctx, CMSampleBufferRef sample_buffer)
532 CMVideoFormatDescriptionRef vid_fmt;
536 vid_fmt = CMSampleBufferGetFormatDescription(sample_buffer);
538 av_log(avctx, AV_LOG_ERROR, "No video format.\n");
539 return AVERROR_EXTERNAL;
542 status = get_params_size(avctx, vid_fmt, &total_size);
544 av_log(avctx, AV_LOG_ERROR, "Could not get parameter sets.\n");
548 avctx->extradata = av_mallocz(total_size + AV_INPUT_BUFFER_PADDING_SIZE);
549 if (!avctx->extradata) {
550 return AVERROR(ENOMEM);
552 avctx->extradata_size = total_size;
554 status = copy_param_sets(avctx, vid_fmt, avctx->extradata, total_size);
557 av_log(avctx, AV_LOG_ERROR, "Could not copy param sets.\n");
564 static void vtenc_output_callback(
566 void *sourceFrameCtx,
568 VTEncodeInfoFlags flags,
569 CMSampleBufferRef sample_buffer)
571 AVCodecContext *avctx = ctx;
572 VTEncContext *vtctx = avctx->priv_data;
573 ExtraSEI *sei = sourceFrameCtx;
575 if (vtctx->async_error) {
576 if(sample_buffer) CFRelease(sample_buffer);
581 av_log(avctx, AV_LOG_ERROR, "Error encoding frame: %d\n", (int)status);
582 set_async_error(vtctx, AVERROR_EXTERNAL);
586 if (!sample_buffer) {
590 if (!avctx->extradata && (avctx->flags & AV_CODEC_FLAG_GLOBAL_HEADER)) {
591 int set_status = set_extradata(avctx, sample_buffer);
593 set_async_error(vtctx, set_status);
598 vtenc_q_push(vtctx, sample_buffer, sei);
601 static int get_length_code_size(
602 AVCodecContext *avctx,
603 CMSampleBufferRef sample_buffer,
606 VTEncContext *vtctx = avctx->priv_data;
607 CMVideoFormatDescriptionRef vid_fmt;
611 vid_fmt = CMSampleBufferGetFormatDescription(sample_buffer);
613 av_log(avctx, AV_LOG_ERROR, "Error getting buffer format description.\n");
614 return AVERROR_EXTERNAL;
617 status = vtctx->get_param_set_func(vid_fmt,
624 av_log(avctx, AV_LOG_ERROR, "Error getting length code size: %d\n", status);
625 return AVERROR_EXTERNAL;
633 * Returns true on success.
635 * If profile_level_val is NULL and this method returns true, don't specify the
636 * profile/level to the encoder.
638 static bool get_vt_h264_profile_level(AVCodecContext *avctx,
639 CFStringRef *profile_level_val)
641 VTEncContext *vtctx = avctx->priv_data;
642 int64_t profile = vtctx->profile;
644 if (profile == H264_PROF_AUTO && vtctx->level) {
645 //Need to pick a profile if level is not auto-selected.
646 profile = vtctx->has_b_frames ? H264_PROF_MAIN : H264_PROF_BASELINE;
649 *profile_level_val = NULL;
655 case H264_PROF_BASELINE:
656 switch (vtctx->level) {
657 case 0: *profile_level_val =
658 compat_keys.kVTProfileLevel_H264_Baseline_AutoLevel; break;
659 case 13: *profile_level_val = kVTProfileLevel_H264_Baseline_1_3; break;
660 case 30: *profile_level_val = kVTProfileLevel_H264_Baseline_3_0; break;
661 case 31: *profile_level_val = kVTProfileLevel_H264_Baseline_3_1; break;
662 case 32: *profile_level_val = kVTProfileLevel_H264_Baseline_3_2; break;
663 case 40: *profile_level_val =
664 compat_keys.kVTProfileLevel_H264_Baseline_4_0; break;
665 case 41: *profile_level_val = kVTProfileLevel_H264_Baseline_4_1; break;
666 case 42: *profile_level_val =
667 compat_keys.kVTProfileLevel_H264_Baseline_4_2; break;
668 case 50: *profile_level_val =
669 compat_keys.kVTProfileLevel_H264_Baseline_5_0; break;
670 case 51: *profile_level_val =
671 compat_keys.kVTProfileLevel_H264_Baseline_5_1; break;
672 case 52: *profile_level_val =
673 compat_keys.kVTProfileLevel_H264_Baseline_5_2; break;
678 switch (vtctx->level) {
679 case 0: *profile_level_val =
680 compat_keys.kVTProfileLevel_H264_Main_AutoLevel; break;
681 case 30: *profile_level_val = kVTProfileLevel_H264_Main_3_0; break;
682 case 31: *profile_level_val = kVTProfileLevel_H264_Main_3_1; break;
683 case 32: *profile_level_val = kVTProfileLevel_H264_Main_3_2; break;
684 case 40: *profile_level_val = kVTProfileLevel_H264_Main_4_0; break;
685 case 41: *profile_level_val = kVTProfileLevel_H264_Main_4_1; break;
686 case 42: *profile_level_val =
687 compat_keys.kVTProfileLevel_H264_Main_4_2; break;
688 case 50: *profile_level_val = kVTProfileLevel_H264_Main_5_0; break;
689 case 51: *profile_level_val =
690 compat_keys.kVTProfileLevel_H264_Main_5_1; break;
691 case 52: *profile_level_val =
692 compat_keys.kVTProfileLevel_H264_Main_5_2; break;
697 switch (vtctx->level) {
698 case 0: *profile_level_val =
699 compat_keys.kVTProfileLevel_H264_High_AutoLevel; break;
700 case 30: *profile_level_val =
701 compat_keys.kVTProfileLevel_H264_High_3_0; break;
702 case 31: *profile_level_val =
703 compat_keys.kVTProfileLevel_H264_High_3_1; break;
704 case 32: *profile_level_val =
705 compat_keys.kVTProfileLevel_H264_High_3_2; break;
706 case 40: *profile_level_val =
707 compat_keys.kVTProfileLevel_H264_High_4_0; break;
708 case 41: *profile_level_val =
709 compat_keys.kVTProfileLevel_H264_High_4_1; break;
710 case 42: *profile_level_val =
711 compat_keys.kVTProfileLevel_H264_High_4_2; break;
712 case 50: *profile_level_val = kVTProfileLevel_H264_High_5_0; break;
713 case 51: *profile_level_val =
714 compat_keys.kVTProfileLevel_H264_High_5_1; break;
715 case 52: *profile_level_val =
716 compat_keys.kVTProfileLevel_H264_High_5_2; break;
719 case H264_PROF_EXTENDED:
720 switch (vtctx->level) {
721 case 0: *profile_level_val =
722 compat_keys.kVTProfileLevel_H264_Extended_AutoLevel; break;
723 case 50: *profile_level_val =
724 compat_keys.kVTProfileLevel_H264_Extended_5_0; break;
729 if (!*profile_level_val) {
730 av_log(avctx, AV_LOG_ERROR, "Invalid Profile/Level.\n");
738 * Returns true on success.
740 * If profile_level_val is NULL and this method returns true, don't specify the
741 * profile/level to the encoder.
743 static bool get_vt_hevc_profile_level(AVCodecContext *avctx,
744 CFStringRef *profile_level_val)
746 VTEncContext *vtctx = avctx->priv_data;
747 int64_t profile = vtctx->profile;
749 *profile_level_val = NULL;
756 compat_keys.kVTProfileLevel_HEVC_Main_AutoLevel;
758 case HEVC_PROF_MAIN10:
760 compat_keys.kVTProfileLevel_HEVC_Main10_AutoLevel;
764 if (!*profile_level_val) {
765 av_log(avctx, AV_LOG_ERROR, "Invalid Profile/Level.\n");
772 static int get_cv_pixel_format(AVCodecContext* avctx,
773 enum AVPixelFormat fmt,
774 enum AVColorRange range,
775 int* av_pixel_format,
778 if (range_guessed) *range_guessed = range != AVCOL_RANGE_MPEG &&
779 range != AVCOL_RANGE_JPEG;
781 //MPEG range is used when no range is set
782 if (fmt == AV_PIX_FMT_NV12) {
783 *av_pixel_format = range == AVCOL_RANGE_JPEG ?
784 kCVPixelFormatType_420YpCbCr8BiPlanarFullRange :
785 kCVPixelFormatType_420YpCbCr8BiPlanarVideoRange;
786 } else if (fmt == AV_PIX_FMT_YUV420P) {
787 *av_pixel_format = range == AVCOL_RANGE_JPEG ?
788 kCVPixelFormatType_420YpCbCr8PlanarFullRange :
789 kCVPixelFormatType_420YpCbCr8Planar;
790 } else if (fmt == AV_PIX_FMT_P010LE) {
791 *av_pixel_format = range == AVCOL_RANGE_JPEG ?
792 kCVPixelFormatType_420YpCbCr10BiPlanarFullRange :
793 kCVPixelFormatType_420YpCbCr10BiPlanarVideoRange;
795 return AVERROR(EINVAL);
801 static void add_color_attr(AVCodecContext *avctx, CFMutableDictionaryRef dict) {
802 VTEncContext *vtctx = avctx->priv_data;
804 if (vtctx->color_primaries) {
805 CFDictionarySetValue(dict,
806 kCVImageBufferColorPrimariesKey,
807 vtctx->color_primaries);
810 if (vtctx->transfer_function) {
811 CFDictionarySetValue(dict,
812 kCVImageBufferTransferFunctionKey,
813 vtctx->transfer_function);
816 if (vtctx->ycbcr_matrix) {
817 CFDictionarySetValue(dict,
818 kCVImageBufferYCbCrMatrixKey,
819 vtctx->ycbcr_matrix);
823 static int create_cv_pixel_buffer_info(AVCodecContext* avctx,
824 CFMutableDictionaryRef* dict)
826 CFNumberRef cv_color_format_num = NULL;
827 CFNumberRef width_num = NULL;
828 CFNumberRef height_num = NULL;
829 CFMutableDictionaryRef pixel_buffer_info = NULL;
831 int status = get_cv_pixel_format(avctx,
836 if (status) return status;
838 pixel_buffer_info = CFDictionaryCreateMutable(
841 &kCFCopyStringDictionaryKeyCallBacks,
842 &kCFTypeDictionaryValueCallBacks);
844 if (!pixel_buffer_info) goto pbinfo_nomem;
846 cv_color_format_num = CFNumberCreate(kCFAllocatorDefault,
849 if (!cv_color_format_num) goto pbinfo_nomem;
851 CFDictionarySetValue(pixel_buffer_info,
852 kCVPixelBufferPixelFormatTypeKey,
853 cv_color_format_num);
854 vt_release_num(&cv_color_format_num);
856 width_num = CFNumberCreate(kCFAllocatorDefault,
859 if (!width_num) return AVERROR(ENOMEM);
861 CFDictionarySetValue(pixel_buffer_info,
862 kCVPixelBufferWidthKey,
864 vt_release_num(&width_num);
866 height_num = CFNumberCreate(kCFAllocatorDefault,
869 if (!height_num) goto pbinfo_nomem;
871 CFDictionarySetValue(pixel_buffer_info,
872 kCVPixelBufferHeightKey,
874 vt_release_num(&height_num);
876 add_color_attr(avctx, pixel_buffer_info);
878 *dict = pixel_buffer_info;
882 vt_release_num(&cv_color_format_num);
883 vt_release_num(&width_num);
884 vt_release_num(&height_num);
885 if (pixel_buffer_info) CFRelease(pixel_buffer_info);
887 return AVERROR(ENOMEM);
890 static int get_cv_color_primaries(AVCodecContext *avctx,
891 CFStringRef *primaries)
893 enum AVColorPrimaries pri = avctx->color_primaries;
895 case AVCOL_PRI_UNSPECIFIED:
899 case AVCOL_PRI_BT470BG:
900 *primaries = kCVImageBufferColorPrimaries_EBU_3213;
903 case AVCOL_PRI_SMPTE170M:
904 *primaries = kCVImageBufferColorPrimaries_SMPTE_C;
907 case AVCOL_PRI_BT709:
908 *primaries = kCVImageBufferColorPrimaries_ITU_R_709_2;
911 case AVCOL_PRI_BT2020:
912 *primaries = compat_keys.kCVImageBufferColorPrimaries_ITU_R_2020;
916 av_log(avctx, AV_LOG_ERROR, "Color primaries %s is not supported.\n", av_color_primaries_name(pri));
924 static int get_cv_transfer_function(AVCodecContext *avctx,
925 CFStringRef *transfer_fnc,
926 CFNumberRef *gamma_level)
928 enum AVColorTransferCharacteristic trc = avctx->color_trc;
933 case AVCOL_TRC_UNSPECIFIED:
934 *transfer_fnc = NULL;
937 case AVCOL_TRC_BT709:
938 *transfer_fnc = kCVImageBufferTransferFunction_ITU_R_709_2;
941 case AVCOL_TRC_SMPTE240M:
942 *transfer_fnc = kCVImageBufferTransferFunction_SMPTE_240M_1995;
945 #if HAVE_KCVIMAGEBUFFERTRANSFERFUNCTION_SMPTE_ST_2084_PQ
946 case AVCOL_TRC_SMPTE2084:
947 *transfer_fnc = kCVImageBufferTransferFunction_SMPTE_ST_2084_PQ;
950 #if HAVE_KCVIMAGEBUFFERTRANSFERFUNCTION_LINEAR
951 case AVCOL_TRC_LINEAR:
952 *transfer_fnc = kCVImageBufferTransferFunction_Linear;
955 #if HAVE_KCVIMAGEBUFFERTRANSFERFUNCTION_ITU_R_2100_HLG
956 case AVCOL_TRC_ARIB_STD_B67:
957 *transfer_fnc = kCVImageBufferTransferFunction_ITU_R_2100_HLG;
961 case AVCOL_TRC_GAMMA22:
963 *transfer_fnc = kCVImageBufferTransferFunction_UseGamma;
964 *gamma_level = CFNumberCreate(NULL, kCFNumberFloat32Type, &gamma);
967 case AVCOL_TRC_GAMMA28:
969 *transfer_fnc = kCVImageBufferTransferFunction_UseGamma;
970 *gamma_level = CFNumberCreate(NULL, kCFNumberFloat32Type, &gamma);
973 case AVCOL_TRC_BT2020_10:
974 case AVCOL_TRC_BT2020_12:
975 *transfer_fnc = compat_keys.kCVImageBufferTransferFunction_ITU_R_2020;
979 *transfer_fnc = NULL;
980 av_log(avctx, AV_LOG_ERROR, "Transfer function %s is not supported.\n", av_color_transfer_name(trc));
987 static int get_cv_ycbcr_matrix(AVCodecContext *avctx, CFStringRef *matrix) {
988 switch(avctx->colorspace) {
989 case AVCOL_SPC_BT709:
990 *matrix = kCVImageBufferYCbCrMatrix_ITU_R_709_2;
993 case AVCOL_SPC_UNSPECIFIED:
997 case AVCOL_SPC_BT470BG:
998 case AVCOL_SPC_SMPTE170M:
999 *matrix = kCVImageBufferYCbCrMatrix_ITU_R_601_4;
1002 case AVCOL_SPC_SMPTE240M:
1003 *matrix = kCVImageBufferYCbCrMatrix_SMPTE_240M_1995;
1006 case AVCOL_SPC_BT2020_NCL:
1007 *matrix = compat_keys.kCVImageBufferYCbCrMatrix_ITU_R_2020;
1011 av_log(avctx, AV_LOG_ERROR, "Color space %s is not supported.\n", av_color_space_name(avctx->colorspace));
1018 static int vtenc_create_encoder(AVCodecContext *avctx,
1019 CMVideoCodecType codec_type,
1020 CFStringRef profile_level,
1021 CFNumberRef gamma_level,
1022 CFDictionaryRef enc_info,
1023 CFDictionaryRef pixel_buffer_info,
1024 VTCompressionSessionRef *session)
1026 VTEncContext *vtctx = avctx->priv_data;
1027 SInt32 bit_rate = avctx->bit_rate;
1028 SInt32 max_rate = avctx->rc_max_rate;
1029 CFNumberRef bit_rate_num;
1030 CFNumberRef bytes_per_second;
1031 CFNumberRef one_second;
1032 CFArrayRef data_rate_limits;
1033 int64_t bytes_per_second_value = 0;
1034 int64_t one_second_value = 0;
1037 int status = VTCompressionSessionCreate(kCFAllocatorDefault,
1043 kCFAllocatorDefault,
1044 vtenc_output_callback,
1048 if (status || !vtctx->session) {
1049 av_log(avctx, AV_LOG_ERROR, "Error: cannot create compression session: %d\n", status);
1051 #if !TARGET_OS_IPHONE
1052 if (!vtctx->allow_sw) {
1053 av_log(avctx, AV_LOG_ERROR, "Try -allow_sw 1. The hardware encoder may be busy, or not supported.\n");
1057 return AVERROR_EXTERNAL;
1060 bit_rate_num = CFNumberCreate(kCFAllocatorDefault,
1061 kCFNumberSInt32Type,
1063 if (!bit_rate_num) return AVERROR(ENOMEM);
1065 status = VTSessionSetProperty(vtctx->session,
1066 kVTCompressionPropertyKey_AverageBitRate,
1068 CFRelease(bit_rate_num);
1071 av_log(avctx, AV_LOG_ERROR, "Error setting bitrate property: %d\n", status);
1072 return AVERROR_EXTERNAL;
1075 if (vtctx->codec_id == AV_CODEC_ID_H264 && max_rate > 0) {
1076 // kVTCompressionPropertyKey_DataRateLimits is not available for HEVC
1077 bytes_per_second_value = max_rate >> 3;
1078 bytes_per_second = CFNumberCreate(kCFAllocatorDefault,
1079 kCFNumberSInt64Type,
1080 &bytes_per_second_value);
1081 if (!bytes_per_second) {
1082 return AVERROR(ENOMEM);
1084 one_second_value = 1;
1085 one_second = CFNumberCreate(kCFAllocatorDefault,
1086 kCFNumberSInt64Type,
1089 CFRelease(bytes_per_second);
1090 return AVERROR(ENOMEM);
1092 nums[0] = (void *)bytes_per_second;
1093 nums[1] = (void *)one_second;
1094 data_rate_limits = CFArrayCreate(kCFAllocatorDefault,
1095 (const void **)nums,
1097 &kCFTypeArrayCallBacks);
1099 if (!data_rate_limits) {
1100 CFRelease(bytes_per_second);
1101 CFRelease(one_second);
1102 return AVERROR(ENOMEM);
1104 status = VTSessionSetProperty(vtctx->session,
1105 kVTCompressionPropertyKey_DataRateLimits,
1108 CFRelease(bytes_per_second);
1109 CFRelease(one_second);
1110 CFRelease(data_rate_limits);
1113 av_log(avctx, AV_LOG_ERROR, "Error setting max bitrate property: %d\n", status);
1114 return AVERROR_EXTERNAL;
1118 if (profile_level) {
1119 status = VTSessionSetProperty(vtctx->session,
1120 kVTCompressionPropertyKey_ProfileLevel,
1123 av_log(avctx, AV_LOG_ERROR, "Error setting profile/level property: %d. Output will be encoded using a supported profile/level combination.\n", status);
1127 if (avctx->gop_size > 0) {
1128 CFNumberRef interval = CFNumberCreate(kCFAllocatorDefault,
1132 return AVERROR(ENOMEM);
1135 status = VTSessionSetProperty(vtctx->session,
1136 kVTCompressionPropertyKey_MaxKeyFrameInterval,
1138 CFRelease(interval);
1141 av_log(avctx, AV_LOG_ERROR, "Error setting 'max key-frame interval' property: %d\n", status);
1142 return AVERROR_EXTERNAL;
1146 if (vtctx->frames_before) {
1147 status = VTSessionSetProperty(vtctx->session,
1148 kVTCompressionPropertyKey_MoreFramesBeforeStart,
1151 if (status == kVTPropertyNotSupportedErr) {
1152 av_log(avctx, AV_LOG_WARNING, "frames_before property is not supported on this device. Ignoring.\n");
1153 } else if (status) {
1154 av_log(avctx, AV_LOG_ERROR, "Error setting frames_before property: %d\n", status);
1158 if (vtctx->frames_after) {
1159 status = VTSessionSetProperty(vtctx->session,
1160 kVTCompressionPropertyKey_MoreFramesAfterEnd,
1163 if (status == kVTPropertyNotSupportedErr) {
1164 av_log(avctx, AV_LOG_WARNING, "frames_after property is not supported on this device. Ignoring.\n");
1165 } else if (status) {
1166 av_log(avctx, AV_LOG_ERROR, "Error setting frames_after property: %d\n", status);
1170 if (avctx->sample_aspect_ratio.num != 0) {
1173 CFMutableDictionaryRef par;
1174 AVRational *avpar = &avctx->sample_aspect_ratio;
1176 av_reduce(&avpar->num, &avpar->den,
1177 avpar->num, avpar->den,
1180 num = CFNumberCreate(kCFAllocatorDefault,
1184 den = CFNumberCreate(kCFAllocatorDefault,
1190 par = CFDictionaryCreateMutable(kCFAllocatorDefault,
1192 &kCFCopyStringDictionaryKeyCallBacks,
1193 &kCFTypeDictionaryValueCallBacks);
1195 if (!par || !num || !den) {
1196 if (par) CFRelease(par);
1197 if (num) CFRelease(num);
1198 if (den) CFRelease(den);
1200 return AVERROR(ENOMEM);
1203 CFDictionarySetValue(
1205 kCMFormatDescriptionKey_PixelAspectRatioHorizontalSpacing,
1208 CFDictionarySetValue(
1210 kCMFormatDescriptionKey_PixelAspectRatioVerticalSpacing,
1213 status = VTSessionSetProperty(vtctx->session,
1214 kVTCompressionPropertyKey_PixelAspectRatio,
1224 "Error setting pixel aspect ratio to %d:%d: %d.\n",
1225 avctx->sample_aspect_ratio.num,
1226 avctx->sample_aspect_ratio.den,
1229 return AVERROR_EXTERNAL;
1234 if (vtctx->transfer_function) {
1235 status = VTSessionSetProperty(vtctx->session,
1236 kVTCompressionPropertyKey_TransferFunction,
1237 vtctx->transfer_function);
1240 av_log(avctx, AV_LOG_WARNING, "Could not set transfer function: %d\n", status);
1245 if (vtctx->ycbcr_matrix) {
1246 status = VTSessionSetProperty(vtctx->session,
1247 kVTCompressionPropertyKey_YCbCrMatrix,
1248 vtctx->ycbcr_matrix);
1251 av_log(avctx, AV_LOG_WARNING, "Could not set ycbcr matrix: %d\n", status);
1256 if (vtctx->color_primaries) {
1257 status = VTSessionSetProperty(vtctx->session,
1258 kVTCompressionPropertyKey_ColorPrimaries,
1259 vtctx->color_primaries);
1262 av_log(avctx, AV_LOG_WARNING, "Could not set color primaries: %d\n", status);
1267 status = VTSessionSetProperty(vtctx->session,
1268 kCVImageBufferGammaLevelKey,
1272 av_log(avctx, AV_LOG_WARNING, "Could not set gamma level: %d\n", status);
1276 if (!vtctx->has_b_frames) {
1277 status = VTSessionSetProperty(vtctx->session,
1278 kVTCompressionPropertyKey_AllowFrameReordering,
1282 av_log(avctx, AV_LOG_ERROR, "Error setting 'allow frame reordering' property: %d\n", status);
1283 return AVERROR_EXTERNAL;
1287 if (vtctx->entropy != VT_ENTROPY_NOT_SET) {
1288 CFStringRef entropy = vtctx->entropy == VT_CABAC ?
1289 compat_keys.kVTH264EntropyMode_CABAC:
1290 compat_keys.kVTH264EntropyMode_CAVLC;
1292 status = VTSessionSetProperty(vtctx->session,
1293 compat_keys.kVTCompressionPropertyKey_H264EntropyMode,
1297 av_log(avctx, AV_LOG_ERROR, "Error setting entropy property: %d\n", status);
1301 if (vtctx->realtime) {
1302 status = VTSessionSetProperty(vtctx->session,
1303 compat_keys.kVTCompressionPropertyKey_RealTime,
1307 av_log(avctx, AV_LOG_ERROR, "Error setting realtime property: %d\n", status);
1311 status = VTCompressionSessionPrepareToEncodeFrames(vtctx->session);
1313 av_log(avctx, AV_LOG_ERROR, "Error: cannot prepare encoder: %d\n", status);
1314 return AVERROR_EXTERNAL;
1320 static int vtenc_configure_encoder(AVCodecContext *avctx)
1322 CFMutableDictionaryRef enc_info;
1323 CFMutableDictionaryRef pixel_buffer_info;
1324 CMVideoCodecType codec_type;
1325 VTEncContext *vtctx = avctx->priv_data;
1326 CFStringRef profile_level;
1327 CFNumberRef gamma_level = NULL;
1330 codec_type = get_cm_codec_type(avctx->codec_id);
1332 av_log(avctx, AV_LOG_ERROR, "Error: no mapping for AVCodecID %d\n", avctx->codec_id);
1333 return AVERROR(EINVAL);
1336 vtctx->codec_id = avctx->codec_id;
1338 if (vtctx->codec_id == AV_CODEC_ID_H264) {
1339 vtctx->get_param_set_func = CMVideoFormatDescriptionGetH264ParameterSetAtIndex;
1341 vtctx->has_b_frames = avctx->max_b_frames > 0;
1342 if(vtctx->has_b_frames && vtctx->profile == H264_PROF_BASELINE){
1343 av_log(avctx, AV_LOG_WARNING, "Cannot use B-frames with baseline profile. Output will not contain B-frames.\n");
1344 vtctx->has_b_frames = false;
1347 if (vtctx->entropy == VT_CABAC && vtctx->profile == H264_PROF_BASELINE) {
1348 av_log(avctx, AV_LOG_WARNING, "CABAC entropy requires 'main' or 'high' profile, but baseline was requested. Encode will not use CABAC entropy.\n");
1349 vtctx->entropy = VT_ENTROPY_NOT_SET;
1352 if (!get_vt_h264_profile_level(avctx, &profile_level)) return AVERROR(EINVAL);
1354 vtctx->get_param_set_func = compat_keys.CMVideoFormatDescriptionGetHEVCParameterSetAtIndex;
1355 if (!vtctx->get_param_set_func) return AVERROR(EINVAL);
1356 if (!get_vt_hevc_profile_level(avctx, &profile_level)) return AVERROR(EINVAL);
1359 enc_info = CFDictionaryCreateMutable(
1360 kCFAllocatorDefault,
1362 &kCFCopyStringDictionaryKeyCallBacks,
1363 &kCFTypeDictionaryValueCallBacks
1366 if (!enc_info) return AVERROR(ENOMEM);
1368 #if !TARGET_OS_IPHONE
1369 if(vtctx->require_sw) {
1370 CFDictionarySetValue(enc_info,
1371 compat_keys.kVTVideoEncoderSpecification_EnableHardwareAcceleratedVideoEncoder,
1373 } else if (!vtctx->allow_sw) {
1374 CFDictionarySetValue(enc_info,
1375 compat_keys.kVTVideoEncoderSpecification_RequireHardwareAcceleratedVideoEncoder,
1378 CFDictionarySetValue(enc_info,
1379 compat_keys.kVTVideoEncoderSpecification_EnableHardwareAcceleratedVideoEncoder,
1384 if (avctx->pix_fmt != AV_PIX_FMT_VIDEOTOOLBOX) {
1385 status = create_cv_pixel_buffer_info(avctx, &pixel_buffer_info);
1389 pixel_buffer_info = NULL;
1392 vtctx->dts_delta = vtctx->has_b_frames ? -1 : 0;
1394 get_cv_transfer_function(avctx, &vtctx->transfer_function, &gamma_level);
1395 get_cv_ycbcr_matrix(avctx, &vtctx->ycbcr_matrix);
1396 get_cv_color_primaries(avctx, &vtctx->color_primaries);
1399 if (avctx->flags & AV_CODEC_FLAG_GLOBAL_HEADER) {
1400 status = vtenc_populate_extradata(avctx,
1410 status = vtenc_create_encoder(avctx,
1420 CFRelease(gamma_level);
1422 if (pixel_buffer_info)
1423 CFRelease(pixel_buffer_info);
1425 CFRelease(enc_info);
1430 static av_cold int vtenc_init(AVCodecContext *avctx)
1432 VTEncContext *vtctx = avctx->priv_data;
1433 CFBooleanRef has_b_frames_cfbool;
1436 pthread_once(&once_ctrl, loadVTEncSymbols);
1438 pthread_mutex_init(&vtctx->lock, NULL);
1439 pthread_cond_init(&vtctx->cv_sample_sent, NULL);
1441 vtctx->session = NULL;
1442 status = vtenc_configure_encoder(avctx);
1443 if (status) return status;
1445 status = VTSessionCopyProperty(vtctx->session,
1446 kVTCompressionPropertyKey_AllowFrameReordering,
1447 kCFAllocatorDefault,
1448 &has_b_frames_cfbool);
1450 if (!status && has_b_frames_cfbool) {
1451 //Some devices don't output B-frames for main profile, even if requested.
1452 vtctx->has_b_frames = CFBooleanGetValue(has_b_frames_cfbool);
1453 CFRelease(has_b_frames_cfbool);
1455 avctx->has_b_frames = vtctx->has_b_frames;
1460 static void vtenc_get_frame_info(CMSampleBufferRef buffer, bool *is_key_frame)
1462 CFArrayRef attachments;
1463 CFDictionaryRef attachment;
1464 CFBooleanRef not_sync;
1467 attachments = CMSampleBufferGetSampleAttachmentsArray(buffer, false);
1468 len = !attachments ? 0 : CFArrayGetCount(attachments);
1471 *is_key_frame = true;
1475 attachment = CFArrayGetValueAtIndex(attachments, 0);
1477 if (CFDictionaryGetValueIfPresent(attachment,
1478 kCMSampleAttachmentKey_NotSync,
1479 (const void **)¬_sync))
1481 *is_key_frame = !CFBooleanGetValue(not_sync);
1483 *is_key_frame = true;
1487 static int is_post_sei_nal_type(int nal_type){
1488 return nal_type != H264_NAL_SEI &&
1489 nal_type != H264_NAL_SPS &&
1490 nal_type != H264_NAL_PPS &&
1491 nal_type != H264_NAL_AUD;
1495 * Finds the sei message start/size of type find_sei_type.
1496 * If more than one of that type exists, the last one is returned.
1498 static int find_sei_end(AVCodecContext *avctx,
1504 size_t sei_payload_size = 0;
1505 int sei_payload_type = 0;
1507 uint8_t *nal_start = nal_data;
1512 nal_type = *nal_data & 0x1F;
1513 if (nal_type != H264_NAL_SEI)
1519 if (nal_data[nal_size - 1] == 0x80)
1522 while (nal_size > 0 && *nal_data > 0) {
1524 sei_payload_type += *nal_data;
1527 } while (nal_size > 0 && *nal_data == 0xFF);
1530 av_log(avctx, AV_LOG_ERROR, "Unexpected end of SEI NAL Unit parsing type.\n");
1531 return AVERROR_INVALIDDATA;
1535 sei_payload_size += *nal_data;
1538 } while (nal_size > 0 && *nal_data == 0xFF);
1540 if (nal_size < sei_payload_size) {
1541 av_log(avctx, AV_LOG_ERROR, "Unexpected end of SEI NAL Unit parsing size.\n");
1542 return AVERROR_INVALIDDATA;
1545 nal_data += sei_payload_size;
1546 nal_size -= sei_payload_size;
1549 *sei_end = nal_data;
1551 return nal_data - nal_start + 1;
1555 * Copies the data inserting emulation prevention bytes as needed.
1556 * Existing data in the destination can be taken into account by providing
1557 * dst with a dst_offset > 0.
1559 * @return The number of bytes copied on success. On failure, the negative of
1560 * the number of bytes needed to copy src is returned.
1562 static int copy_emulation_prev(const uint8_t *src,
1571 uint8_t* dst_end = dst + dst_size;
1572 const uint8_t* src_end = src + src_size;
1573 int start_at = dst_offset > 2 ? dst_offset - 2 : 0;
1575 for (i = start_at; i < dst_offset && i < dst_size; i++) {
1584 for (; src < src_end; src++, dst++) {
1586 int insert_ep3_byte = *src <= 3;
1587 if (insert_ep3_byte) {
1605 wrote_bytes = dst - dst_start;
1608 return -wrote_bytes;
1613 static int write_sei(const ExtraSEI *sei,
1618 uint8_t *sei_start = dst;
1619 size_t remaining_sei_size = sei->size;
1620 size_t remaining_dst_size = dst_size;
1625 if (!remaining_dst_size)
1626 return AVERROR_BUFFER_TOO_SMALL;
1628 while (sei_type && remaining_dst_size != 0) {
1629 int sei_byte = sei_type > 255 ? 255 : sei_type;
1632 sei_type -= sei_byte;
1634 remaining_dst_size--;
1638 return AVERROR_BUFFER_TOO_SMALL;
1640 while (remaining_sei_size && remaining_dst_size != 0) {
1641 int size_byte = remaining_sei_size > 255 ? 255 : remaining_sei_size;
1644 remaining_sei_size -= size_byte;
1646 remaining_dst_size--;
1649 if (remaining_dst_size < sei->size)
1650 return AVERROR_BUFFER_TOO_SMALL;
1652 header_bytes = dst - sei_start;
1654 offset = header_bytes;
1655 bytes_written = copy_emulation_prev(sei->data,
1660 if (bytes_written < 0)
1661 return AVERROR_BUFFER_TOO_SMALL;
1663 bytes_written += header_bytes;
1664 return bytes_written;
1668 * Copies NAL units and replaces length codes with
1669 * H.264 Annex B start codes. On failure, the contents of
1670 * dst_data may have been modified.
1672 * @param length_code_size Byte length of each length code
1673 * @param sample_buffer NAL units prefixed with length codes.
1674 * @param sei Optional A53 closed captions SEI data.
1675 * @param dst_data Must be zeroed before calling this function.
1676 * Contains the copied NAL units prefixed with
1677 * start codes when the function returns
1679 * @param dst_size Length of dst_data
1680 * @return 0 on success
1681 * AVERROR_INVALIDDATA if length_code_size is invalid
1682 * AVERROR_BUFFER_TOO_SMALL if dst_data is too small
1683 * or if a length_code in src_data specifies data beyond
1684 * the end of its buffer.
1686 static int copy_replace_length_codes(
1687 AVCodecContext *avctx,
1688 size_t length_code_size,
1689 CMSampleBufferRef sample_buffer,
1694 size_t src_size = CMSampleBufferGetTotalSampleSize(sample_buffer);
1695 size_t remaining_src_size = src_size;
1696 size_t remaining_dst_size = dst_size;
1697 size_t src_offset = 0;
1700 uint8_t size_buf[4];
1702 CMBlockBufferRef block = CMSampleBufferGetDataBuffer(sample_buffer);
1704 if (length_code_size > 4) {
1705 return AVERROR_INVALIDDATA;
1708 while (remaining_src_size > 0) {
1709 size_t curr_src_len;
1710 size_t curr_dst_len;
1716 status = CMBlockBufferCopyDataBytes(block,
1721 av_log(avctx, AV_LOG_ERROR, "Cannot copy length: %d\n", status);
1722 return AVERROR_EXTERNAL;
1725 status = CMBlockBufferCopyDataBytes(block,
1726 src_offset + length_code_size,
1731 av_log(avctx, AV_LOG_ERROR, "Cannot copy type: %d\n", status);
1732 return AVERROR_EXTERNAL;
1737 for (i = 0; i < length_code_size; i++) {
1739 box_len |= size_buf[i];
1742 if (sei && !wrote_sei && is_post_sei_nal_type(nal_type)) {
1743 //No SEI NAL unit - insert.
1746 memcpy(dst_data, start_code, sizeof(start_code));
1747 dst_data += sizeof(start_code);
1748 remaining_dst_size -= sizeof(start_code);
1750 *dst_data = H264_NAL_SEI;
1752 remaining_dst_size--;
1754 wrote_bytes = write_sei(sei,
1755 H264_SEI_TYPE_USER_DATA_REGISTERED,
1757 remaining_dst_size);
1759 if (wrote_bytes < 0)
1762 remaining_dst_size -= wrote_bytes;
1763 dst_data += wrote_bytes;
1765 if (remaining_dst_size <= 0)
1766 return AVERROR_BUFFER_TOO_SMALL;
1771 remaining_dst_size--;
1776 curr_src_len = box_len + length_code_size;
1777 curr_dst_len = box_len + sizeof(start_code);
1779 if (remaining_src_size < curr_src_len) {
1780 return AVERROR_BUFFER_TOO_SMALL;
1783 if (remaining_dst_size < curr_dst_len) {
1784 return AVERROR_BUFFER_TOO_SMALL;
1787 dst_box = dst_data + sizeof(start_code);
1789 memcpy(dst_data, start_code, sizeof(start_code));
1790 status = CMBlockBufferCopyDataBytes(block,
1791 src_offset + length_code_size,
1796 av_log(avctx, AV_LOG_ERROR, "Cannot copy data: %d\n", status);
1797 return AVERROR_EXTERNAL;
1800 if (sei && !wrote_sei && nal_type == H264_NAL_SEI) {
1801 //Found SEI NAL unit - append.
1806 old_sei_length = find_sei_end(avctx, dst_box, box_len, &new_sei);
1807 if (old_sei_length < 0)
1810 wrote_bytes = write_sei(sei,
1811 H264_SEI_TYPE_USER_DATA_REGISTERED,
1813 remaining_dst_size - old_sei_length);
1814 if (wrote_bytes < 0)
1817 if (new_sei + wrote_bytes >= dst_data + remaining_dst_size)
1818 return AVERROR_BUFFER_TOO_SMALL;
1820 new_sei[wrote_bytes++] = 0x80;
1821 extra_bytes = wrote_bytes - (dst_box + box_len - new_sei);
1823 dst_data += extra_bytes;
1824 remaining_dst_size -= extra_bytes;
1829 src_offset += curr_src_len;
1830 dst_data += curr_dst_len;
1832 remaining_src_size -= curr_src_len;
1833 remaining_dst_size -= curr_dst_len;
1840 * Returns a sufficient number of bytes to contain the sei data.
1841 * It may be greater than the minimum required.
1843 static int get_sei_msg_bytes(const ExtraSEI* sei, int type){
1848 copied_size = -copy_emulation_prev(sei->data,
1854 if ((sei->size % 255) == 0) //may result in an extra byte
1857 return copied_size + sei->size / 255 + 1 + type / 255 + 1;
1860 static int vtenc_cm_to_avpacket(
1861 AVCodecContext *avctx,
1862 CMSampleBufferRef sample_buffer,
1866 VTEncContext *vtctx = avctx->priv_data;
1871 size_t length_code_size;
1872 size_t header_size = 0;
1874 size_t out_buf_size;
1875 size_t sei_nalu_size = 0;
1877 int64_t time_base_num;
1881 CMVideoFormatDescriptionRef vid_fmt;
1884 vtenc_get_frame_info(sample_buffer, &is_key_frame);
1885 status = get_length_code_size(avctx, sample_buffer, &length_code_size);
1886 if (status) return status;
1888 add_header = is_key_frame && !(avctx->flags & AV_CODEC_FLAG_GLOBAL_HEADER);
1891 vid_fmt = CMSampleBufferGetFormatDescription(sample_buffer);
1893 av_log(avctx, AV_LOG_ERROR, "Cannot get format description.\n");
1894 return AVERROR_EXTERNAL;
1897 int status = get_params_size(avctx, vid_fmt, &header_size);
1898 if (status) return status;
1901 status = count_nalus(length_code_size, sample_buffer, &nalu_count);
1906 size_t msg_size = get_sei_msg_bytes(sei,
1907 H264_SEI_TYPE_USER_DATA_REGISTERED);
1909 sei_nalu_size = sizeof(start_code) + 1 + msg_size + 1;
1912 in_buf_size = CMSampleBufferGetTotalSampleSize(sample_buffer);
1913 out_buf_size = header_size +
1916 nalu_count * ((int)sizeof(start_code) - (int)length_code_size);
1918 status = ff_alloc_packet2(avctx, pkt, out_buf_size, out_buf_size);
1923 status = copy_param_sets(avctx, vid_fmt, pkt->data, out_buf_size);
1924 if(status) return status;
1927 status = copy_replace_length_codes(
1932 pkt->data + header_size,
1933 pkt->size - header_size
1937 av_log(avctx, AV_LOG_ERROR, "Error copying packet data: %d\n", status);
1942 pkt->flags |= AV_PKT_FLAG_KEY;
1945 pts = CMSampleBufferGetPresentationTimeStamp(sample_buffer);
1946 dts = CMSampleBufferGetDecodeTimeStamp (sample_buffer);
1948 if (CMTIME_IS_INVALID(dts)) {
1949 if (!vtctx->has_b_frames) {
1952 av_log(avctx, AV_LOG_ERROR, "DTS is invalid.\n");
1953 return AVERROR_EXTERNAL;
1957 dts_delta = vtctx->dts_delta >= 0 ? vtctx->dts_delta : 0;
1958 time_base_num = avctx->time_base.num;
1959 pkt->pts = pts.value / time_base_num;
1960 pkt->dts = dts.value / time_base_num - dts_delta;
1961 pkt->size = out_buf_size;
1967 * contiguous_buf_size is 0 if not contiguous, and the size of the buffer
1968 * containing all planes if so.
1970 static int get_cv_pixel_info(
1971 AVCodecContext *avctx,
1972 const AVFrame *frame,
1978 size_t *contiguous_buf_size)
1980 VTEncContext *vtctx = avctx->priv_data;
1981 int av_format = frame->format;
1982 int av_color_range = frame->color_range;
1987 status = get_cv_pixel_format(avctx, av_format, av_color_range, color, &range_guessed);
1991 "Could not get pixel format for color format '%s' range '%s'.\n",
1992 av_get_pix_fmt_name(av_format),
1993 av_color_range > AVCOL_RANGE_UNSPECIFIED &&
1994 av_color_range < AVCOL_RANGE_NB ?
1995 av_color_range_name(av_color_range) :
1998 return AVERROR(EINVAL);
2001 if (range_guessed) {
2002 if (!vtctx->warned_color_range) {
2003 vtctx->warned_color_range = true;
2006 "Color range not set for %s. Using MPEG range.\n",
2007 av_get_pix_fmt_name(av_format));
2011 switch (av_format) {
2012 case AV_PIX_FMT_NV12:
2015 widths [0] = avctx->width;
2016 heights[0] = avctx->height;
2017 strides[0] = frame ? frame->linesize[0] : avctx->width;
2019 widths [1] = (avctx->width + 1) / 2;
2020 heights[1] = (avctx->height + 1) / 2;
2021 strides[1] = frame ? frame->linesize[1] : (avctx->width + 1) & -2;
2024 case AV_PIX_FMT_YUV420P:
2027 widths [0] = avctx->width;
2028 heights[0] = avctx->height;
2029 strides[0] = frame ? frame->linesize[0] : avctx->width;
2031 widths [1] = (avctx->width + 1) / 2;
2032 heights[1] = (avctx->height + 1) / 2;
2033 strides[1] = frame ? frame->linesize[1] : (avctx->width + 1) / 2;
2035 widths [2] = (avctx->width + 1) / 2;
2036 heights[2] = (avctx->height + 1) / 2;
2037 strides[2] = frame ? frame->linesize[2] : (avctx->width + 1) / 2;
2040 case AV_PIX_FMT_P010LE:
2042 widths[0] = avctx->width;
2043 heights[0] = avctx->height;
2044 strides[0] = frame ? frame->linesize[0] : (avctx->width * 2 + 63) & -64;
2046 widths[1] = (avctx->width + 1) / 2;
2047 heights[1] = (avctx->height + 1) / 2;
2048 strides[1] = frame ? frame->linesize[1] : ((avctx->width + 1) / 2 + 63) & -64;
2055 "Could not get frame format info for color %d range %d.\n",
2059 return AVERROR(EINVAL);
2062 *contiguous_buf_size = 0;
2063 for (i = 0; i < *plane_count; i++) {
2064 if (i < *plane_count - 1 &&
2065 frame->data[i] + strides[i] * heights[i] != frame->data[i + 1]) {
2066 *contiguous_buf_size = 0;
2070 *contiguous_buf_size += strides[i] * heights[i];
2076 //Not used on OSX - frame is never copied.
2077 static int copy_avframe_to_pixel_buffer(AVCodecContext *avctx,
2078 const AVFrame *frame,
2079 CVPixelBufferRef cv_img,
2080 const size_t *plane_strides,
2081 const size_t *plane_rows)
2093 status = CVPixelBufferLockBaseAddress(cv_img, 0);
2098 "Error: Could not lock base address of CVPixelBuffer: %d.\n",
2103 if (CVPixelBufferIsPlanar(cv_img)) {
2104 plane_count = CVPixelBufferGetPlaneCount(cv_img);
2105 for (i = 0; frame->data[i]; i++) {
2106 if (i == plane_count) {
2107 CVPixelBufferUnlockBaseAddress(cv_img, 0);
2110 "Error: different number of planes in AVFrame and CVPixelBuffer.\n"
2113 return AVERROR_EXTERNAL;
2116 dst_addr = (uint8_t*)CVPixelBufferGetBaseAddressOfPlane(cv_img, i);
2117 src_addr = (uint8_t*)frame->data[i];
2118 dst_stride = CVPixelBufferGetBytesPerRowOfPlane(cv_img, i);
2119 src_stride = plane_strides[i];
2120 rows = plane_rows[i];
2122 if (dst_stride == src_stride) {
2123 memcpy(dst_addr, src_addr, src_stride * rows);
2125 copy_bytes = dst_stride < src_stride ? dst_stride : src_stride;
2127 for (j = 0; j < rows; j++) {
2128 memcpy(dst_addr + j * dst_stride, src_addr + j * src_stride, copy_bytes);
2133 if (frame->data[1]) {
2134 CVPixelBufferUnlockBaseAddress(cv_img, 0);
2137 "Error: different number of planes in AVFrame and non-planar CVPixelBuffer.\n"
2140 return AVERROR_EXTERNAL;
2143 dst_addr = (uint8_t*)CVPixelBufferGetBaseAddress(cv_img);
2144 src_addr = (uint8_t*)frame->data[0];
2145 dst_stride = CVPixelBufferGetBytesPerRow(cv_img);
2146 src_stride = plane_strides[0];
2147 rows = plane_rows[0];
2149 if (dst_stride == src_stride) {
2150 memcpy(dst_addr, src_addr, src_stride * rows);
2152 copy_bytes = dst_stride < src_stride ? dst_stride : src_stride;
2154 for (j = 0; j < rows; j++) {
2155 memcpy(dst_addr + j * dst_stride, src_addr + j * src_stride, copy_bytes);
2160 status = CVPixelBufferUnlockBaseAddress(cv_img, 0);
2162 av_log(avctx, AV_LOG_ERROR, "Error: Could not unlock CVPixelBuffer base address: %d.\n", status);
2163 return AVERROR_EXTERNAL;
2169 static int create_cv_pixel_buffer(AVCodecContext *avctx,
2170 const AVFrame *frame,
2171 CVPixelBufferRef *cv_img)
2175 size_t widths [AV_NUM_DATA_POINTERS];
2176 size_t heights[AV_NUM_DATA_POINTERS];
2177 size_t strides[AV_NUM_DATA_POINTERS];
2179 size_t contiguous_buf_size;
2180 CVPixelBufferPoolRef pix_buf_pool;
2181 VTEncContext* vtctx = avctx->priv_data;
2183 if (avctx->pix_fmt == AV_PIX_FMT_VIDEOTOOLBOX) {
2184 av_assert0(frame->format == AV_PIX_FMT_VIDEOTOOLBOX);
2186 *cv_img = (CVPixelBufferRef)frame->data[3];
2187 av_assert0(*cv_img);
2193 memset(widths, 0, sizeof(widths));
2194 memset(heights, 0, sizeof(heights));
2195 memset(strides, 0, sizeof(strides));
2197 status = get_cv_pixel_info(
2205 &contiguous_buf_size
2212 "Error: Cannot convert format %d color_range %d: %d\n",
2218 return AVERROR_EXTERNAL;
2221 pix_buf_pool = VTCompressionSessionGetPixelBufferPool(vtctx->session);
2222 if (!pix_buf_pool) {
2223 /* On iOS, the VT session is invalidated when the APP switches from
2224 * foreground to background and vice versa. Fetch the actual error code
2225 * of the VT session to detect that case and restart the VT session
2229 vtstatus = VTCompressionSessionPrepareToEncodeFrames(vtctx->session);
2230 if (vtstatus == kVTInvalidSessionErr) {
2231 CFRelease(vtctx->session);
2232 vtctx->session = NULL;
2233 status = vtenc_configure_encoder(avctx);
2235 pix_buf_pool = VTCompressionSessionGetPixelBufferPool(vtctx->session);
2237 if (!pix_buf_pool) {
2238 av_log(avctx, AV_LOG_ERROR, "Could not get pixel buffer pool.\n");
2239 return AVERROR_EXTERNAL;
2242 av_log(avctx, AV_LOG_WARNING, "VT session restarted because of a "
2243 "kVTInvalidSessionErr error.\n");
2246 status = CVPixelBufferPoolCreatePixelBuffer(NULL,
2252 av_log(avctx, AV_LOG_ERROR, "Could not create pixel buffer from pool: %d.\n", status);
2253 return AVERROR_EXTERNAL;
2256 status = copy_avframe_to_pixel_buffer(avctx, frame, *cv_img, strides, heights);
2266 static int create_encoder_dict_h264(const AVFrame *frame,
2267 CFDictionaryRef* dict_out)
2269 CFDictionaryRef dict = NULL;
2270 if (frame->pict_type == AV_PICTURE_TYPE_I) {
2271 const void *keys[] = { kVTEncodeFrameOptionKey_ForceKeyFrame };
2272 const void *vals[] = { kCFBooleanTrue };
2274 dict = CFDictionaryCreate(NULL, keys, vals, 1, NULL, NULL);
2275 if(!dict) return AVERROR(ENOMEM);
2282 static int vtenc_send_frame(AVCodecContext *avctx,
2283 VTEncContext *vtctx,
2284 const AVFrame *frame)
2287 CFDictionaryRef frame_dict;
2288 CVPixelBufferRef cv_img = NULL;
2289 AVFrameSideData *side_data = NULL;
2290 ExtraSEI *sei = NULL;
2291 int status = create_cv_pixel_buffer(avctx, frame, &cv_img);
2293 if (status) return status;
2295 status = create_encoder_dict_h264(frame, &frame_dict);
2301 side_data = av_frame_get_side_data(frame, AV_FRAME_DATA_A53_CC);
2302 if (vtctx->a53_cc && side_data && side_data->size) {
2303 sei = av_mallocz(sizeof(*sei));
2305 av_log(avctx, AV_LOG_ERROR, "Not enough memory for closed captions, skipping\n");
2307 int ret = ff_alloc_a53_sei(frame, 0, &sei->data, &sei->size);
2309 av_log(avctx, AV_LOG_ERROR, "Not enough memory for closed captions, skipping\n");
2316 time = CMTimeMake(frame->pts * avctx->time_base.num, avctx->time_base.den);
2317 status = VTCompressionSessionEncodeFrame(
2327 if (frame_dict) CFRelease(frame_dict);
2331 av_log(avctx, AV_LOG_ERROR, "Error: cannot encode frame: %d\n", status);
2332 return AVERROR_EXTERNAL;
2338 static av_cold int vtenc_frame(
2339 AVCodecContext *avctx,
2341 const AVFrame *frame,
2344 VTEncContext *vtctx = avctx->priv_data;
2347 CMSampleBufferRef buf = NULL;
2348 ExtraSEI *sei = NULL;
2351 status = vtenc_send_frame(avctx, vtctx, frame);
2354 status = AVERROR_EXTERNAL;
2358 if (vtctx->frame_ct_in == 0) {
2359 vtctx->first_pts = frame->pts;
2360 } else if(vtctx->frame_ct_in == 1 && vtctx->has_b_frames) {
2361 vtctx->dts_delta = frame->pts - vtctx->first_pts;
2364 vtctx->frame_ct_in++;
2365 } else if(!vtctx->flushing) {
2366 vtctx->flushing = true;
2368 status = VTCompressionSessionCompleteFrames(vtctx->session,
2372 av_log(avctx, AV_LOG_ERROR, "Error flushing frames: %d\n", status);
2373 status = AVERROR_EXTERNAL;
2379 get_frame = vtctx->dts_delta >= 0 || !frame;
2385 status = vtenc_q_pop(vtctx, !frame, &buf, &sei);
2386 if (status) goto end_nopkt;
2387 if (!buf) goto end_nopkt;
2389 status = vtenc_cm_to_avpacket(avctx, buf, pkt, sei);
2391 if (sei->data) av_free(sei->data);
2395 if (status) goto end_nopkt;
2401 av_packet_unref(pkt);
2405 static int vtenc_populate_extradata(AVCodecContext *avctx,
2406 CMVideoCodecType codec_type,
2407 CFStringRef profile_level,
2408 CFNumberRef gamma_level,
2409 CFDictionaryRef enc_info,
2410 CFDictionaryRef pixel_buffer_info)
2412 VTEncContext *vtctx = avctx->priv_data;
2414 CVPixelBufferPoolRef pool = NULL;
2415 CVPixelBufferRef pix_buf = NULL;
2417 CMSampleBufferRef buf = NULL;
2419 status = vtenc_create_encoder(avctx,
2429 pool = VTCompressionSessionGetPixelBufferPool(vtctx->session);
2431 av_log(avctx, AV_LOG_ERROR, "Error getting pixel buffer pool.\n");
2435 status = CVPixelBufferPoolCreatePixelBuffer(NULL,
2439 if(status != kCVReturnSuccess){
2440 av_log(avctx, AV_LOG_ERROR, "Error creating frame from pool: %d\n", status);
2444 time = CMTimeMake(0, avctx->time_base.den);
2445 status = VTCompressionSessionEncodeFrame(vtctx->session,
2456 "Error sending frame for extradata: %d\n",
2462 //Populates extradata - output frames are flushed and param sets are available.
2463 status = VTCompressionSessionCompleteFrames(vtctx->session,
2469 status = vtenc_q_pop(vtctx, 0, &buf, NULL);
2471 av_log(avctx, AV_LOG_ERROR, "popping: %d\n", status);
2481 CFRelease(vtctx->session);
2483 vtctx->session = NULL;
2484 vtctx->frame_ct_out = 0;
2486 av_assert0(status != 0 || (avctx->extradata && avctx->extradata_size > 0));
2491 static av_cold int vtenc_close(AVCodecContext *avctx)
2493 VTEncContext *vtctx = avctx->priv_data;
2495 if(!vtctx->session) {
2496 pthread_cond_destroy(&vtctx->cv_sample_sent);
2497 pthread_mutex_destroy(&vtctx->lock);
2501 VTCompressionSessionCompleteFrames(vtctx->session,
2503 clear_frame_queue(vtctx);
2504 pthread_cond_destroy(&vtctx->cv_sample_sent);
2505 pthread_mutex_destroy(&vtctx->lock);
2506 CFRelease(vtctx->session);
2507 vtctx->session = NULL;
2509 if (vtctx->color_primaries) {
2510 CFRelease(vtctx->color_primaries);
2511 vtctx->color_primaries = NULL;
2514 if (vtctx->transfer_function) {
2515 CFRelease(vtctx->transfer_function);
2516 vtctx->transfer_function = NULL;
2519 if (vtctx->ycbcr_matrix) {
2520 CFRelease(vtctx->ycbcr_matrix);
2521 vtctx->ycbcr_matrix = NULL;
2527 static const enum AVPixelFormat avc_pix_fmts[] = {
2528 AV_PIX_FMT_VIDEOTOOLBOX,
2534 static const enum AVPixelFormat hevc_pix_fmts[] = {
2535 AV_PIX_FMT_VIDEOTOOLBOX,
2542 #define VE AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
2543 #define COMMON_OPTIONS \
2544 { "allow_sw", "Allow software encoding", OFFSET(allow_sw), AV_OPT_TYPE_BOOL, \
2545 { .i64 = 0 }, 0, 1, VE }, \
2546 { "require_sw", "Require software encoding", OFFSET(require_sw), AV_OPT_TYPE_BOOL, \
2547 { .i64 = 0 }, 0, 1, VE }, \
2548 { "realtime", "Hint that encoding should happen in real-time if not faster (e.g. capturing from camera).", \
2549 OFFSET(realtime), AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, VE }, \
2550 { "frames_before", "Other frames will come before the frames in this session. This helps smooth concatenation issues.", \
2551 OFFSET(frames_before), AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, VE }, \
2552 { "frames_after", "Other frames will come after the frames in this session. This helps smooth concatenation issues.", \
2553 OFFSET(frames_after), AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, VE },
2555 #define OFFSET(x) offsetof(VTEncContext, x)
2556 static const AVOption h264_options[] = {
2557 { "profile", "Profile", OFFSET(profile), AV_OPT_TYPE_INT, { .i64 = H264_PROF_AUTO }, H264_PROF_AUTO, H264_PROF_COUNT, VE, "profile" },
2558 { "baseline", "Baseline Profile", 0, AV_OPT_TYPE_CONST, { .i64 = H264_PROF_BASELINE }, INT_MIN, INT_MAX, VE, "profile" },
2559 { "main", "Main Profile", 0, AV_OPT_TYPE_CONST, { .i64 = H264_PROF_MAIN }, INT_MIN, INT_MAX, VE, "profile" },
2560 { "high", "High Profile", 0, AV_OPT_TYPE_CONST, { .i64 = H264_PROF_HIGH }, INT_MIN, INT_MAX, VE, "profile" },
2561 { "extended", "Extend Profile", 0, AV_OPT_TYPE_CONST, { .i64 = H264_PROF_EXTENDED }, INT_MIN, INT_MAX, VE, "profile" },
2563 { "level", "Level", OFFSET(level), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 52, VE, "level" },
2564 { "1.3", "Level 1.3, only available with Baseline Profile", 0, AV_OPT_TYPE_CONST, { .i64 = 13 }, INT_MIN, INT_MAX, VE, "level" },
2565 { "3.0", "Level 3.0", 0, AV_OPT_TYPE_CONST, { .i64 = 30 }, INT_MIN, INT_MAX, VE, "level" },
2566 { "3.1", "Level 3.1", 0, AV_OPT_TYPE_CONST, { .i64 = 31 }, INT_MIN, INT_MAX, VE, "level" },
2567 { "3.2", "Level 3.2", 0, AV_OPT_TYPE_CONST, { .i64 = 32 }, INT_MIN, INT_MAX, VE, "level" },
2568 { "4.0", "Level 4.0", 0, AV_OPT_TYPE_CONST, { .i64 = 40 }, INT_MIN, INT_MAX, VE, "level" },
2569 { "4.1", "Level 4.1", 0, AV_OPT_TYPE_CONST, { .i64 = 41 }, INT_MIN, INT_MAX, VE, "level" },
2570 { "4.2", "Level 4.2", 0, AV_OPT_TYPE_CONST, { .i64 = 42 }, INT_MIN, INT_MAX, VE, "level" },
2571 { "5.0", "Level 5.0", 0, AV_OPT_TYPE_CONST, { .i64 = 50 }, INT_MIN, INT_MAX, VE, "level" },
2572 { "5.1", "Level 5.1", 0, AV_OPT_TYPE_CONST, { .i64 = 51 }, INT_MIN, INT_MAX, VE, "level" },
2573 { "5.2", "Level 5.2", 0, AV_OPT_TYPE_CONST, { .i64 = 52 }, INT_MIN, INT_MAX, VE, "level" },
2575 { "coder", "Entropy coding", OFFSET(entropy), AV_OPT_TYPE_INT, { .i64 = VT_ENTROPY_NOT_SET }, VT_ENTROPY_NOT_SET, VT_CABAC, VE, "coder" },
2576 { "cavlc", "CAVLC entropy coding", 0, AV_OPT_TYPE_CONST, { .i64 = VT_CAVLC }, INT_MIN, INT_MAX, VE, "coder" },
2577 { "vlc", "CAVLC entropy coding", 0, AV_OPT_TYPE_CONST, { .i64 = VT_CAVLC }, INT_MIN, INT_MAX, VE, "coder" },
2578 { "cabac", "CABAC entropy coding", 0, AV_OPT_TYPE_CONST, { .i64 = VT_CABAC }, INT_MIN, INT_MAX, VE, "coder" },
2579 { "ac", "CABAC entropy coding", 0, AV_OPT_TYPE_CONST, { .i64 = VT_CABAC }, INT_MIN, INT_MAX, VE, "coder" },
2581 { "a53cc", "Use A53 Closed Captions (if available)", OFFSET(a53_cc), AV_OPT_TYPE_BOOL, {.i64 = 1}, 0, 1, VE },
2587 static const AVClass h264_videotoolbox_class = {
2588 .class_name = "h264_videotoolbox",
2589 .item_name = av_default_item_name,
2590 .option = h264_options,
2591 .version = LIBAVUTIL_VERSION_INT,
2594 AVCodec ff_h264_videotoolbox_encoder = {
2595 .name = "h264_videotoolbox",
2596 .long_name = NULL_IF_CONFIG_SMALL("VideoToolbox H.264 Encoder"),
2597 .type = AVMEDIA_TYPE_VIDEO,
2598 .id = AV_CODEC_ID_H264,
2599 .priv_data_size = sizeof(VTEncContext),
2600 .pix_fmts = avc_pix_fmts,
2602 .encode2 = vtenc_frame,
2603 .close = vtenc_close,
2604 .capabilities = AV_CODEC_CAP_DELAY,
2605 .priv_class = &h264_videotoolbox_class,
2606 .caps_internal = FF_CODEC_CAP_INIT_THREADSAFE |
2607 FF_CODEC_CAP_INIT_CLEANUP,
2610 static const AVOption hevc_options[] = {
2611 { "profile", "Profile", OFFSET(profile), AV_OPT_TYPE_INT, { .i64 = HEVC_PROF_AUTO }, HEVC_PROF_AUTO, HEVC_PROF_COUNT, VE, "profile" },
2612 { "main", "Main Profile", 0, AV_OPT_TYPE_CONST, { .i64 = HEVC_PROF_MAIN }, INT_MIN, INT_MAX, VE, "profile" },
2613 { "main10", "Main10 Profile", 0, AV_OPT_TYPE_CONST, { .i64 = HEVC_PROF_MAIN10 }, INT_MIN, INT_MAX, VE, "profile" },
2619 static const AVClass hevc_videotoolbox_class = {
2620 .class_name = "hevc_videotoolbox",
2621 .item_name = av_default_item_name,
2622 .option = hevc_options,
2623 .version = LIBAVUTIL_VERSION_INT,
2626 AVCodec ff_hevc_videotoolbox_encoder = {
2627 .name = "hevc_videotoolbox",
2628 .long_name = NULL_IF_CONFIG_SMALL("VideoToolbox H.265 Encoder"),
2629 .type = AVMEDIA_TYPE_VIDEO,
2630 .id = AV_CODEC_ID_HEVC,
2631 .priv_data_size = sizeof(VTEncContext),
2632 .pix_fmts = hevc_pix_fmts,
2634 .encode2 = vtenc_frame,
2635 .close = vtenc_close,
2636 .capabilities = AV_CODEC_CAP_DELAY | AV_CODEC_CAP_HARDWARE,
2637 .priv_class = &hevc_videotoolbox_class,
2638 .caps_internal = FF_CODEC_CAP_INIT_THREADSAFE |
2639 FF_CODEC_CAP_INIT_CLEANUP,
2640 .wrapper_name = "videotoolbox",