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;
232 static int vtenc_populate_extradata(AVCodecContext *avctx,
233 CMVideoCodecType codec_type,
234 CFStringRef profile_level,
235 CFNumberRef gamma_level,
236 CFDictionaryRef enc_info,
237 CFDictionaryRef pixel_buffer_info);
240 * NULL-safe release of *refPtr, and sets value to NULL.
242 static void vt_release_num(CFNumberRef* refPtr){
251 static void set_async_error(VTEncContext *vtctx, int err)
255 pthread_mutex_lock(&vtctx->lock);
257 vtctx->async_error = err;
259 info = vtctx->q_head;
260 vtctx->q_head = vtctx->q_tail = NULL;
263 BufNode *next = info->next;
264 CFRelease(info->cm_buffer);
269 pthread_mutex_unlock(&vtctx->lock);
272 static void clear_frame_queue(VTEncContext *vtctx)
274 set_async_error(vtctx, 0);
277 static int vtenc_q_pop(VTEncContext *vtctx, bool wait, CMSampleBufferRef *buf, ExtraSEI **sei)
281 pthread_mutex_lock(&vtctx->lock);
283 if (vtctx->async_error) {
284 pthread_mutex_unlock(&vtctx->lock);
285 return vtctx->async_error;
288 if (vtctx->flushing && vtctx->frame_ct_in == vtctx->frame_ct_out) {
291 pthread_mutex_unlock(&vtctx->lock);
295 while (!vtctx->q_head && !vtctx->async_error && wait) {
296 pthread_cond_wait(&vtctx->cv_sample_sent, &vtctx->lock);
299 if (!vtctx->q_head) {
300 pthread_mutex_unlock(&vtctx->lock);
305 info = vtctx->q_head;
306 vtctx->q_head = vtctx->q_head->next;
307 if (!vtctx->q_head) {
308 vtctx->q_tail = NULL;
311 pthread_mutex_unlock(&vtctx->lock);
313 *buf = info->cm_buffer;
316 } else if (info->sei) {
317 if (info->sei->data) av_free(info->sei->data);
322 vtctx->frame_ct_out++;
327 static void vtenc_q_push(VTEncContext *vtctx, CMSampleBufferRef buffer, ExtraSEI *sei)
329 BufNode *info = av_malloc(sizeof(BufNode));
331 set_async_error(vtctx, AVERROR(ENOMEM));
336 info->cm_buffer = buffer;
340 pthread_mutex_lock(&vtctx->lock);
341 pthread_cond_signal(&vtctx->cv_sample_sent);
343 if (!vtctx->q_head) {
344 vtctx->q_head = info;
346 vtctx->q_tail->next = info;
349 vtctx->q_tail = info;
351 pthread_mutex_unlock(&vtctx->lock);
354 static int count_nalus(size_t length_code_size,
355 CMSampleBufferRef sample_buffer,
362 size_t src_size = CMSampleBufferGetTotalSampleSize(sample_buffer);
363 CMBlockBufferRef block = CMSampleBufferGetDataBuffer(sample_buffer);
365 if (length_code_size > 4)
366 return AVERROR_INVALIDDATA;
368 while (offset < src_size) {
373 status = CMBlockBufferCopyDataBytes(block,
378 for (i = 0; i < length_code_size; i++) {
380 box_len |= size_buf[i];
383 curr_src_len = box_len + length_code_size;
384 offset += curr_src_len;
393 static CMVideoCodecType get_cm_codec_type(enum AVCodecID id)
396 case AV_CODEC_ID_H264: return kCMVideoCodecType_H264;
397 case AV_CODEC_ID_HEVC: return kCMVideoCodecType_HEVC;
403 * Get the parameter sets from a CMSampleBufferRef.
404 * @param dst If *dst isn't NULL, the parameters are copied into existing
405 * memory. *dst_size must be set accordingly when *dst != NULL.
406 * If *dst is NULL, it will be allocated.
407 * In all cases, *dst_size is set to the number of bytes used starting
410 static int get_params_size(
411 AVCodecContext *avctx,
412 CMVideoFormatDescriptionRef vid_fmt,
415 VTEncContext *vtctx = avctx->priv_data;
416 size_t total_size = 0;
418 int is_count_bad = 0;
421 status = vtctx->get_param_set_func(vid_fmt,
433 for (i = 0; i < ps_count || is_count_bad; i++) {
436 status = vtctx->get_param_set_func(vid_fmt,
444 * When ps_count is invalid, status != 0 ends the loop normally
445 * unless we didn't get any parameter sets.
447 if (i > 0 && is_count_bad) status = 0;
452 total_size += ps_size + sizeof(start_code);
456 av_log(avctx, AV_LOG_ERROR, "Error getting parameter set sizes: %d\n", status);
457 return AVERROR_EXTERNAL;
464 static int copy_param_sets(
465 AVCodecContext *avctx,
466 CMVideoFormatDescriptionRef vid_fmt,
470 VTEncContext *vtctx = avctx->priv_data;
472 int is_count_bad = 0;
477 status = vtctx->get_param_set_func(vid_fmt,
490 for (i = 0; i < ps_count || is_count_bad; i++) {
495 status = vtctx->get_param_set_func(vid_fmt,
502 if (i > 0 && is_count_bad) status = 0;
507 next_offset = offset + sizeof(start_code) + ps_size;
508 if (dst_size < next_offset) {
509 av_log(avctx, AV_LOG_ERROR, "Error: buffer too small for parameter sets.\n");
510 return AVERROR_BUFFER_TOO_SMALL;
513 memcpy(dst + offset, start_code, sizeof(start_code));
514 offset += sizeof(start_code);
516 memcpy(dst + offset, ps, ps_size);
517 offset = next_offset;
521 av_log(avctx, AV_LOG_ERROR, "Error getting parameter set data: %d\n", status);
522 return AVERROR_EXTERNAL;
528 static int set_extradata(AVCodecContext *avctx, CMSampleBufferRef sample_buffer)
530 CMVideoFormatDescriptionRef vid_fmt;
534 vid_fmt = CMSampleBufferGetFormatDescription(sample_buffer);
536 av_log(avctx, AV_LOG_ERROR, "No video format.\n");
537 return AVERROR_EXTERNAL;
540 status = get_params_size(avctx, vid_fmt, &total_size);
542 av_log(avctx, AV_LOG_ERROR, "Could not get parameter sets.\n");
546 avctx->extradata = av_mallocz(total_size + AV_INPUT_BUFFER_PADDING_SIZE);
547 if (!avctx->extradata) {
548 return AVERROR(ENOMEM);
550 avctx->extradata_size = total_size;
552 status = copy_param_sets(avctx, vid_fmt, avctx->extradata, total_size);
555 av_log(avctx, AV_LOG_ERROR, "Could not copy param sets.\n");
562 static void vtenc_output_callback(
564 void *sourceFrameCtx,
566 VTEncodeInfoFlags flags,
567 CMSampleBufferRef sample_buffer)
569 AVCodecContext *avctx = ctx;
570 VTEncContext *vtctx = avctx->priv_data;
571 ExtraSEI *sei = sourceFrameCtx;
573 if (vtctx->async_error) {
574 if(sample_buffer) CFRelease(sample_buffer);
579 av_log(avctx, AV_LOG_ERROR, "Error encoding frame: %d\n", (int)status);
580 set_async_error(vtctx, AVERROR_EXTERNAL);
584 if (!sample_buffer) {
588 if (!avctx->extradata && (avctx->flags & AV_CODEC_FLAG_GLOBAL_HEADER)) {
589 int set_status = set_extradata(avctx, sample_buffer);
591 set_async_error(vtctx, set_status);
596 vtenc_q_push(vtctx, sample_buffer, sei);
599 static int get_length_code_size(
600 AVCodecContext *avctx,
601 CMSampleBufferRef sample_buffer,
604 VTEncContext *vtctx = avctx->priv_data;
605 CMVideoFormatDescriptionRef vid_fmt;
609 vid_fmt = CMSampleBufferGetFormatDescription(sample_buffer);
611 av_log(avctx, AV_LOG_ERROR, "Error getting buffer format description.\n");
612 return AVERROR_EXTERNAL;
615 status = vtctx->get_param_set_func(vid_fmt,
622 av_log(avctx, AV_LOG_ERROR, "Error getting length code size: %d\n", status);
623 return AVERROR_EXTERNAL;
631 * Returns true on success.
633 * If profile_level_val is NULL and this method returns true, don't specify the
634 * profile/level to the encoder.
636 static bool get_vt_h264_profile_level(AVCodecContext *avctx,
637 CFStringRef *profile_level_val)
639 VTEncContext *vtctx = avctx->priv_data;
640 int64_t profile = vtctx->profile;
642 if (profile == H264_PROF_AUTO && vtctx->level) {
643 //Need to pick a profile if level is not auto-selected.
644 profile = vtctx->has_b_frames ? H264_PROF_MAIN : H264_PROF_BASELINE;
647 *profile_level_val = NULL;
653 case H264_PROF_BASELINE:
654 switch (vtctx->level) {
655 case 0: *profile_level_val =
656 compat_keys.kVTProfileLevel_H264_Baseline_AutoLevel; break;
657 case 13: *profile_level_val = kVTProfileLevel_H264_Baseline_1_3; break;
658 case 30: *profile_level_val = kVTProfileLevel_H264_Baseline_3_0; break;
659 case 31: *profile_level_val = kVTProfileLevel_H264_Baseline_3_1; break;
660 case 32: *profile_level_val = kVTProfileLevel_H264_Baseline_3_2; break;
661 case 40: *profile_level_val =
662 compat_keys.kVTProfileLevel_H264_Baseline_4_0; break;
663 case 41: *profile_level_val = kVTProfileLevel_H264_Baseline_4_1; break;
664 case 42: *profile_level_val =
665 compat_keys.kVTProfileLevel_H264_Baseline_4_2; break;
666 case 50: *profile_level_val =
667 compat_keys.kVTProfileLevel_H264_Baseline_5_0; break;
668 case 51: *profile_level_val =
669 compat_keys.kVTProfileLevel_H264_Baseline_5_1; break;
670 case 52: *profile_level_val =
671 compat_keys.kVTProfileLevel_H264_Baseline_5_2; break;
676 switch (vtctx->level) {
677 case 0: *profile_level_val =
678 compat_keys.kVTProfileLevel_H264_Main_AutoLevel; break;
679 case 30: *profile_level_val = kVTProfileLevel_H264_Main_3_0; break;
680 case 31: *profile_level_val = kVTProfileLevel_H264_Main_3_1; break;
681 case 32: *profile_level_val = kVTProfileLevel_H264_Main_3_2; break;
682 case 40: *profile_level_val = kVTProfileLevel_H264_Main_4_0; break;
683 case 41: *profile_level_val = kVTProfileLevel_H264_Main_4_1; break;
684 case 42: *profile_level_val =
685 compat_keys.kVTProfileLevel_H264_Main_4_2; break;
686 case 50: *profile_level_val = kVTProfileLevel_H264_Main_5_0; break;
687 case 51: *profile_level_val =
688 compat_keys.kVTProfileLevel_H264_Main_5_1; break;
689 case 52: *profile_level_val =
690 compat_keys.kVTProfileLevel_H264_Main_5_2; break;
695 switch (vtctx->level) {
696 case 0: *profile_level_val =
697 compat_keys.kVTProfileLevel_H264_High_AutoLevel; break;
698 case 30: *profile_level_val =
699 compat_keys.kVTProfileLevel_H264_High_3_0; break;
700 case 31: *profile_level_val =
701 compat_keys.kVTProfileLevel_H264_High_3_1; break;
702 case 32: *profile_level_val =
703 compat_keys.kVTProfileLevel_H264_High_3_2; break;
704 case 40: *profile_level_val =
705 compat_keys.kVTProfileLevel_H264_High_4_0; break;
706 case 41: *profile_level_val =
707 compat_keys.kVTProfileLevel_H264_High_4_1; break;
708 case 42: *profile_level_val =
709 compat_keys.kVTProfileLevel_H264_High_4_2; break;
710 case 50: *profile_level_val = kVTProfileLevel_H264_High_5_0; break;
711 case 51: *profile_level_val =
712 compat_keys.kVTProfileLevel_H264_High_5_1; break;
713 case 52: *profile_level_val =
714 compat_keys.kVTProfileLevel_H264_High_5_2; break;
717 case H264_PROF_EXTENDED:
718 switch (vtctx->level) {
719 case 0: *profile_level_val =
720 compat_keys.kVTProfileLevel_H264_Extended_AutoLevel; break;
721 case 50: *profile_level_val =
722 compat_keys.kVTProfileLevel_H264_Extended_5_0; break;
727 if (!*profile_level_val) {
728 av_log(avctx, AV_LOG_ERROR, "Invalid Profile/Level.\n");
736 * Returns true on success.
738 * If profile_level_val is NULL and this method returns true, don't specify the
739 * profile/level to the encoder.
741 static bool get_vt_hevc_profile_level(AVCodecContext *avctx,
742 CFStringRef *profile_level_val)
744 VTEncContext *vtctx = avctx->priv_data;
745 int64_t profile = vtctx->profile;
747 *profile_level_val = NULL;
754 compat_keys.kVTProfileLevel_HEVC_Main_AutoLevel;
756 case HEVC_PROF_MAIN10:
758 compat_keys.kVTProfileLevel_HEVC_Main10_AutoLevel;
762 if (!*profile_level_val) {
763 av_log(avctx, AV_LOG_ERROR, "Invalid Profile/Level.\n");
770 static int get_cv_pixel_format(AVCodecContext* avctx,
771 enum AVPixelFormat fmt,
772 enum AVColorRange range,
773 int* av_pixel_format,
776 if (range_guessed) *range_guessed = range != AVCOL_RANGE_MPEG &&
777 range != AVCOL_RANGE_JPEG;
779 //MPEG range is used when no range is set
780 if (fmt == AV_PIX_FMT_NV12) {
781 *av_pixel_format = range == AVCOL_RANGE_JPEG ?
782 kCVPixelFormatType_420YpCbCr8BiPlanarFullRange :
783 kCVPixelFormatType_420YpCbCr8BiPlanarVideoRange;
784 } else if (fmt == AV_PIX_FMT_YUV420P) {
785 *av_pixel_format = range == AVCOL_RANGE_JPEG ?
786 kCVPixelFormatType_420YpCbCr8PlanarFullRange :
787 kCVPixelFormatType_420YpCbCr8Planar;
788 } else if (fmt == AV_PIX_FMT_P010LE) {
789 *av_pixel_format = range == AVCOL_RANGE_JPEG ?
790 kCVPixelFormatType_420YpCbCr10BiPlanarFullRange :
791 kCVPixelFormatType_420YpCbCr10BiPlanarVideoRange;
793 return AVERROR(EINVAL);
799 static void add_color_attr(AVCodecContext *avctx, CFMutableDictionaryRef dict) {
800 VTEncContext *vtctx = avctx->priv_data;
802 if (vtctx->color_primaries) {
803 CFDictionarySetValue(dict,
804 kCVImageBufferColorPrimariesKey,
805 vtctx->color_primaries);
808 if (vtctx->transfer_function) {
809 CFDictionarySetValue(dict,
810 kCVImageBufferTransferFunctionKey,
811 vtctx->transfer_function);
814 if (vtctx->ycbcr_matrix) {
815 CFDictionarySetValue(dict,
816 kCVImageBufferYCbCrMatrixKey,
817 vtctx->ycbcr_matrix);
821 static int create_cv_pixel_buffer_info(AVCodecContext* avctx,
822 CFMutableDictionaryRef* dict)
824 CFNumberRef cv_color_format_num = NULL;
825 CFNumberRef width_num = NULL;
826 CFNumberRef height_num = NULL;
827 CFMutableDictionaryRef pixel_buffer_info = NULL;
829 int status = get_cv_pixel_format(avctx,
834 if (status) return status;
836 pixel_buffer_info = CFDictionaryCreateMutable(
839 &kCFCopyStringDictionaryKeyCallBacks,
840 &kCFTypeDictionaryValueCallBacks);
842 if (!pixel_buffer_info) goto pbinfo_nomem;
844 cv_color_format_num = CFNumberCreate(kCFAllocatorDefault,
847 if (!cv_color_format_num) goto pbinfo_nomem;
849 CFDictionarySetValue(pixel_buffer_info,
850 kCVPixelBufferPixelFormatTypeKey,
851 cv_color_format_num);
852 vt_release_num(&cv_color_format_num);
854 width_num = CFNumberCreate(kCFAllocatorDefault,
857 if (!width_num) return AVERROR(ENOMEM);
859 CFDictionarySetValue(pixel_buffer_info,
860 kCVPixelBufferWidthKey,
862 vt_release_num(&width_num);
864 height_num = CFNumberCreate(kCFAllocatorDefault,
867 if (!height_num) goto pbinfo_nomem;
869 CFDictionarySetValue(pixel_buffer_info,
870 kCVPixelBufferHeightKey,
872 vt_release_num(&height_num);
874 add_color_attr(avctx, pixel_buffer_info);
876 *dict = pixel_buffer_info;
880 vt_release_num(&cv_color_format_num);
881 vt_release_num(&width_num);
882 vt_release_num(&height_num);
883 if (pixel_buffer_info) CFRelease(pixel_buffer_info);
885 return AVERROR(ENOMEM);
888 static int get_cv_color_primaries(AVCodecContext *avctx,
889 CFStringRef *primaries)
891 enum AVColorPrimaries pri = avctx->color_primaries;
893 case AVCOL_PRI_UNSPECIFIED:
897 case AVCOL_PRI_BT470BG:
898 *primaries = kCVImageBufferColorPrimaries_EBU_3213;
901 case AVCOL_PRI_SMPTE170M:
902 *primaries = kCVImageBufferColorPrimaries_SMPTE_C;
905 case AVCOL_PRI_BT709:
906 *primaries = kCVImageBufferColorPrimaries_ITU_R_709_2;
909 case AVCOL_PRI_BT2020:
910 *primaries = compat_keys.kCVImageBufferColorPrimaries_ITU_R_2020;
914 av_log(avctx, AV_LOG_ERROR, "Color primaries %s is not supported.\n", av_color_primaries_name(pri));
922 static int get_cv_transfer_function(AVCodecContext *avctx,
923 CFStringRef *transfer_fnc,
924 CFNumberRef *gamma_level)
926 enum AVColorTransferCharacteristic trc = avctx->color_trc;
931 case AVCOL_TRC_UNSPECIFIED:
932 *transfer_fnc = NULL;
935 case AVCOL_TRC_BT709:
936 *transfer_fnc = kCVImageBufferTransferFunction_ITU_R_709_2;
939 case AVCOL_TRC_SMPTE240M:
940 *transfer_fnc = kCVImageBufferTransferFunction_SMPTE_240M_1995;
943 #if HAVE_KCVIMAGEBUFFERTRANSFERFUNCTION_SMPTE_ST_2084_PQ
944 case AVCOL_TRC_SMPTE2084:
945 *transfer_fnc = kCVImageBufferTransferFunction_SMPTE_ST_2084_PQ;
948 #if HAVE_KCVIMAGEBUFFERTRANSFERFUNCTION_LINEAR
949 case AVCOL_TRC_LINEAR:
950 *transfer_fnc = kCVImageBufferTransferFunction_Linear;
953 #if HAVE_KCVIMAGEBUFFERTRANSFERFUNCTION_ITU_R_2100_HLG
954 case AVCOL_TRC_ARIB_STD_B67:
955 *transfer_fnc = kCVImageBufferTransferFunction_ITU_R_2100_HLG;
959 case AVCOL_TRC_GAMMA22:
961 *transfer_fnc = kCVImageBufferTransferFunction_UseGamma;
962 *gamma_level = CFNumberCreate(NULL, kCFNumberFloat32Type, &gamma);
965 case AVCOL_TRC_GAMMA28:
967 *transfer_fnc = kCVImageBufferTransferFunction_UseGamma;
968 *gamma_level = CFNumberCreate(NULL, kCFNumberFloat32Type, &gamma);
971 case AVCOL_TRC_BT2020_10:
972 case AVCOL_TRC_BT2020_12:
973 *transfer_fnc = compat_keys.kCVImageBufferTransferFunction_ITU_R_2020;
977 *transfer_fnc = NULL;
978 av_log(avctx, AV_LOG_ERROR, "Transfer function %s is not supported.\n", av_color_transfer_name(trc));
985 static int get_cv_ycbcr_matrix(AVCodecContext *avctx, CFStringRef *matrix) {
986 switch(avctx->colorspace) {
987 case AVCOL_SPC_BT709:
988 *matrix = kCVImageBufferYCbCrMatrix_ITU_R_709_2;
991 case AVCOL_SPC_UNSPECIFIED:
995 case AVCOL_SPC_BT470BG:
996 case AVCOL_SPC_SMPTE170M:
997 *matrix = kCVImageBufferYCbCrMatrix_ITU_R_601_4;
1000 case AVCOL_SPC_SMPTE240M:
1001 *matrix = kCVImageBufferYCbCrMatrix_SMPTE_240M_1995;
1004 case AVCOL_SPC_BT2020_NCL:
1005 *matrix = compat_keys.kCVImageBufferYCbCrMatrix_ITU_R_2020;
1009 av_log(avctx, AV_LOG_ERROR, "Color space %s is not supported.\n", av_color_space_name(avctx->colorspace));
1016 static int vtenc_create_encoder(AVCodecContext *avctx,
1017 CMVideoCodecType codec_type,
1018 CFStringRef profile_level,
1019 CFNumberRef gamma_level,
1020 CFDictionaryRef enc_info,
1021 CFDictionaryRef pixel_buffer_info,
1022 VTCompressionSessionRef *session)
1024 VTEncContext *vtctx = avctx->priv_data;
1025 SInt32 bit_rate = avctx->bit_rate;
1026 SInt32 max_rate = avctx->rc_max_rate;
1027 CFNumberRef bit_rate_num;
1028 CFNumberRef bytes_per_second;
1029 CFNumberRef one_second;
1030 CFArrayRef data_rate_limits;
1031 int64_t bytes_per_second_value = 0;
1032 int64_t one_second_value = 0;
1035 int status = VTCompressionSessionCreate(kCFAllocatorDefault,
1041 kCFAllocatorDefault,
1042 vtenc_output_callback,
1046 if (status || !vtctx->session) {
1047 av_log(avctx, AV_LOG_ERROR, "Error: cannot create compression session: %d\n", status);
1049 #if !TARGET_OS_IPHONE
1050 if (!vtctx->allow_sw) {
1051 av_log(avctx, AV_LOG_ERROR, "Try -allow_sw 1. The hardware encoder may be busy, or not supported.\n");
1055 return AVERROR_EXTERNAL;
1058 bit_rate_num = CFNumberCreate(kCFAllocatorDefault,
1059 kCFNumberSInt32Type,
1061 if (!bit_rate_num) return AVERROR(ENOMEM);
1063 status = VTSessionSetProperty(vtctx->session,
1064 kVTCompressionPropertyKey_AverageBitRate,
1066 CFRelease(bit_rate_num);
1069 av_log(avctx, AV_LOG_ERROR, "Error setting bitrate property: %d\n", status);
1070 return AVERROR_EXTERNAL;
1073 if (vtctx->codec_id == AV_CODEC_ID_H264 && max_rate > 0) {
1074 // kVTCompressionPropertyKey_DataRateLimits is not available for HEVC
1075 bytes_per_second_value = max_rate >> 3;
1076 bytes_per_second = CFNumberCreate(kCFAllocatorDefault,
1077 kCFNumberSInt64Type,
1078 &bytes_per_second_value);
1079 if (!bytes_per_second) {
1080 return AVERROR(ENOMEM);
1082 one_second_value = 1;
1083 one_second = CFNumberCreate(kCFAllocatorDefault,
1084 kCFNumberSInt64Type,
1087 CFRelease(bytes_per_second);
1088 return AVERROR(ENOMEM);
1090 nums[0] = (void *)bytes_per_second;
1091 nums[1] = (void *)one_second;
1092 data_rate_limits = CFArrayCreate(kCFAllocatorDefault,
1093 (const void **)nums,
1095 &kCFTypeArrayCallBacks);
1097 if (!data_rate_limits) {
1098 CFRelease(bytes_per_second);
1099 CFRelease(one_second);
1100 return AVERROR(ENOMEM);
1102 status = VTSessionSetProperty(vtctx->session,
1103 kVTCompressionPropertyKey_DataRateLimits,
1106 CFRelease(bytes_per_second);
1107 CFRelease(one_second);
1108 CFRelease(data_rate_limits);
1111 av_log(avctx, AV_LOG_ERROR, "Error setting max bitrate property: %d\n", status);
1112 return AVERROR_EXTERNAL;
1116 if (profile_level) {
1117 status = VTSessionSetProperty(vtctx->session,
1118 kVTCompressionPropertyKey_ProfileLevel,
1121 av_log(avctx, AV_LOG_ERROR, "Error setting profile/level property: %d. Output will be encoded using a supported profile/level combination.\n", status);
1125 if (avctx->gop_size > 0) {
1126 CFNumberRef interval = CFNumberCreate(kCFAllocatorDefault,
1130 return AVERROR(ENOMEM);
1133 status = VTSessionSetProperty(vtctx->session,
1134 kVTCompressionPropertyKey_MaxKeyFrameInterval,
1136 CFRelease(interval);
1139 av_log(avctx, AV_LOG_ERROR, "Error setting 'max key-frame interval' property: %d\n", status);
1140 return AVERROR_EXTERNAL;
1144 if (vtctx->frames_before) {
1145 status = VTSessionSetProperty(vtctx->session,
1146 kVTCompressionPropertyKey_MoreFramesBeforeStart,
1149 if (status == kVTPropertyNotSupportedErr) {
1150 av_log(avctx, AV_LOG_WARNING, "frames_before property is not supported on this device. Ignoring.\n");
1151 } else if (status) {
1152 av_log(avctx, AV_LOG_ERROR, "Error setting frames_before property: %d\n", status);
1156 if (vtctx->frames_after) {
1157 status = VTSessionSetProperty(vtctx->session,
1158 kVTCompressionPropertyKey_MoreFramesAfterEnd,
1161 if (status == kVTPropertyNotSupportedErr) {
1162 av_log(avctx, AV_LOG_WARNING, "frames_after property is not supported on this device. Ignoring.\n");
1163 } else if (status) {
1164 av_log(avctx, AV_LOG_ERROR, "Error setting frames_after property: %d\n", status);
1168 if (avctx->sample_aspect_ratio.num != 0) {
1171 CFMutableDictionaryRef par;
1172 AVRational *avpar = &avctx->sample_aspect_ratio;
1174 av_reduce(&avpar->num, &avpar->den,
1175 avpar->num, avpar->den,
1178 num = CFNumberCreate(kCFAllocatorDefault,
1182 den = CFNumberCreate(kCFAllocatorDefault,
1188 par = CFDictionaryCreateMutable(kCFAllocatorDefault,
1190 &kCFCopyStringDictionaryKeyCallBacks,
1191 &kCFTypeDictionaryValueCallBacks);
1193 if (!par || !num || !den) {
1194 if (par) CFRelease(par);
1195 if (num) CFRelease(num);
1196 if (den) CFRelease(den);
1198 return AVERROR(ENOMEM);
1201 CFDictionarySetValue(
1203 kCMFormatDescriptionKey_PixelAspectRatioHorizontalSpacing,
1206 CFDictionarySetValue(
1208 kCMFormatDescriptionKey_PixelAspectRatioVerticalSpacing,
1211 status = VTSessionSetProperty(vtctx->session,
1212 kVTCompressionPropertyKey_PixelAspectRatio,
1222 "Error setting pixel aspect ratio to %d:%d: %d.\n",
1223 avctx->sample_aspect_ratio.num,
1224 avctx->sample_aspect_ratio.den,
1227 return AVERROR_EXTERNAL;
1232 if (vtctx->transfer_function) {
1233 status = VTSessionSetProperty(vtctx->session,
1234 kVTCompressionPropertyKey_TransferFunction,
1235 vtctx->transfer_function);
1238 av_log(avctx, AV_LOG_WARNING, "Could not set transfer function: %d\n", status);
1243 if (vtctx->ycbcr_matrix) {
1244 status = VTSessionSetProperty(vtctx->session,
1245 kVTCompressionPropertyKey_YCbCrMatrix,
1246 vtctx->ycbcr_matrix);
1249 av_log(avctx, AV_LOG_WARNING, "Could not set ycbcr matrix: %d\n", status);
1254 if (vtctx->color_primaries) {
1255 status = VTSessionSetProperty(vtctx->session,
1256 kVTCompressionPropertyKey_ColorPrimaries,
1257 vtctx->color_primaries);
1260 av_log(avctx, AV_LOG_WARNING, "Could not set color primaries: %d\n", status);
1265 status = VTSessionSetProperty(vtctx->session,
1266 kCVImageBufferGammaLevelKey,
1270 av_log(avctx, AV_LOG_WARNING, "Could not set gamma level: %d\n", status);
1274 if (!vtctx->has_b_frames) {
1275 status = VTSessionSetProperty(vtctx->session,
1276 kVTCompressionPropertyKey_AllowFrameReordering,
1280 av_log(avctx, AV_LOG_ERROR, "Error setting 'allow frame reordering' property: %d\n", status);
1281 return AVERROR_EXTERNAL;
1285 if (vtctx->entropy != VT_ENTROPY_NOT_SET) {
1286 CFStringRef entropy = vtctx->entropy == VT_CABAC ?
1287 compat_keys.kVTH264EntropyMode_CABAC:
1288 compat_keys.kVTH264EntropyMode_CAVLC;
1290 status = VTSessionSetProperty(vtctx->session,
1291 compat_keys.kVTCompressionPropertyKey_H264EntropyMode,
1295 av_log(avctx, AV_LOG_ERROR, "Error setting entropy property: %d\n", status);
1299 if (vtctx->realtime) {
1300 status = VTSessionSetProperty(vtctx->session,
1301 compat_keys.kVTCompressionPropertyKey_RealTime,
1305 av_log(avctx, AV_LOG_ERROR, "Error setting realtime property: %d\n", status);
1309 status = VTCompressionSessionPrepareToEncodeFrames(vtctx->session);
1311 av_log(avctx, AV_LOG_ERROR, "Error: cannot prepare encoder: %d\n", status);
1312 return AVERROR_EXTERNAL;
1318 static int vtenc_configure_encoder(AVCodecContext *avctx)
1320 CFMutableDictionaryRef enc_info;
1321 CFMutableDictionaryRef pixel_buffer_info;
1322 CMVideoCodecType codec_type;
1323 VTEncContext *vtctx = avctx->priv_data;
1324 CFStringRef profile_level;
1325 CFNumberRef gamma_level = NULL;
1328 codec_type = get_cm_codec_type(avctx->codec_id);
1330 av_log(avctx, AV_LOG_ERROR, "Error: no mapping for AVCodecID %d\n", avctx->codec_id);
1331 return AVERROR(EINVAL);
1334 vtctx->codec_id = avctx->codec_id;
1336 if (vtctx->codec_id == AV_CODEC_ID_H264) {
1337 vtctx->get_param_set_func = CMVideoFormatDescriptionGetH264ParameterSetAtIndex;
1339 vtctx->has_b_frames = avctx->max_b_frames > 0;
1340 if(vtctx->has_b_frames && vtctx->profile == H264_PROF_BASELINE){
1341 av_log(avctx, AV_LOG_WARNING, "Cannot use B-frames with baseline profile. Output will not contain B-frames.\n");
1342 vtctx->has_b_frames = false;
1345 if (vtctx->entropy == VT_CABAC && vtctx->profile == H264_PROF_BASELINE) {
1346 av_log(avctx, AV_LOG_WARNING, "CABAC entropy requires 'main' or 'high' profile, but baseline was requested. Encode will not use CABAC entropy.\n");
1347 vtctx->entropy = VT_ENTROPY_NOT_SET;
1350 if (!get_vt_h264_profile_level(avctx, &profile_level)) return AVERROR(EINVAL);
1352 vtctx->get_param_set_func = compat_keys.CMVideoFormatDescriptionGetHEVCParameterSetAtIndex;
1353 if (!vtctx->get_param_set_func) return AVERROR(EINVAL);
1354 if (!get_vt_hevc_profile_level(avctx, &profile_level)) return AVERROR(EINVAL);
1357 enc_info = CFDictionaryCreateMutable(
1358 kCFAllocatorDefault,
1360 &kCFCopyStringDictionaryKeyCallBacks,
1361 &kCFTypeDictionaryValueCallBacks
1364 if (!enc_info) return AVERROR(ENOMEM);
1366 #if !TARGET_OS_IPHONE
1367 if(vtctx->require_sw) {
1368 CFDictionarySetValue(enc_info,
1369 compat_keys.kVTVideoEncoderSpecification_EnableHardwareAcceleratedVideoEncoder,
1371 } else if (!vtctx->allow_sw) {
1372 CFDictionarySetValue(enc_info,
1373 compat_keys.kVTVideoEncoderSpecification_RequireHardwareAcceleratedVideoEncoder,
1376 CFDictionarySetValue(enc_info,
1377 compat_keys.kVTVideoEncoderSpecification_EnableHardwareAcceleratedVideoEncoder,
1382 if (avctx->pix_fmt != AV_PIX_FMT_VIDEOTOOLBOX) {
1383 status = create_cv_pixel_buffer_info(avctx, &pixel_buffer_info);
1387 pixel_buffer_info = NULL;
1390 vtctx->dts_delta = vtctx->has_b_frames ? -1 : 0;
1392 get_cv_transfer_function(avctx, &vtctx->transfer_function, &gamma_level);
1393 get_cv_ycbcr_matrix(avctx, &vtctx->ycbcr_matrix);
1394 get_cv_color_primaries(avctx, &vtctx->color_primaries);
1397 if (avctx->flags & AV_CODEC_FLAG_GLOBAL_HEADER) {
1398 status = vtenc_populate_extradata(avctx,
1408 status = vtenc_create_encoder(avctx,
1418 CFRelease(gamma_level);
1420 if (pixel_buffer_info)
1421 CFRelease(pixel_buffer_info);
1423 CFRelease(enc_info);
1428 static av_cold int vtenc_init(AVCodecContext *avctx)
1430 VTEncContext *vtctx = avctx->priv_data;
1431 CFBooleanRef has_b_frames_cfbool;
1434 pthread_once(&once_ctrl, loadVTEncSymbols);
1436 pthread_mutex_init(&vtctx->lock, NULL);
1437 pthread_cond_init(&vtctx->cv_sample_sent, NULL);
1439 vtctx->session = NULL;
1440 status = vtenc_configure_encoder(avctx);
1441 if (status) return status;
1443 status = VTSessionCopyProperty(vtctx->session,
1444 kVTCompressionPropertyKey_AllowFrameReordering,
1445 kCFAllocatorDefault,
1446 &has_b_frames_cfbool);
1448 if (!status && has_b_frames_cfbool) {
1449 //Some devices don't output B-frames for main profile, even if requested.
1450 vtctx->has_b_frames = CFBooleanGetValue(has_b_frames_cfbool);
1451 CFRelease(has_b_frames_cfbool);
1453 avctx->has_b_frames = vtctx->has_b_frames;
1458 static void vtenc_get_frame_info(CMSampleBufferRef buffer, bool *is_key_frame)
1460 CFArrayRef attachments;
1461 CFDictionaryRef attachment;
1462 CFBooleanRef not_sync;
1465 attachments = CMSampleBufferGetSampleAttachmentsArray(buffer, false);
1466 len = !attachments ? 0 : CFArrayGetCount(attachments);
1469 *is_key_frame = true;
1473 attachment = CFArrayGetValueAtIndex(attachments, 0);
1475 if (CFDictionaryGetValueIfPresent(attachment,
1476 kCMSampleAttachmentKey_NotSync,
1477 (const void **)¬_sync))
1479 *is_key_frame = !CFBooleanGetValue(not_sync);
1481 *is_key_frame = true;
1485 static int is_post_sei_nal_type(int nal_type){
1486 return nal_type != H264_NAL_SEI &&
1487 nal_type != H264_NAL_SPS &&
1488 nal_type != H264_NAL_PPS &&
1489 nal_type != H264_NAL_AUD;
1493 * Finds the sei message start/size of type find_sei_type.
1494 * If more than one of that type exists, the last one is returned.
1496 static int find_sei_end(AVCodecContext *avctx,
1502 size_t sei_payload_size = 0;
1503 int sei_payload_type = 0;
1505 uint8_t *nal_start = nal_data;
1510 nal_type = *nal_data & 0x1F;
1511 if (nal_type != H264_NAL_SEI)
1517 if (nal_data[nal_size - 1] == 0x80)
1520 while (nal_size > 0 && *nal_data > 0) {
1522 sei_payload_type += *nal_data;
1525 } while (nal_size > 0 && *nal_data == 0xFF);
1528 av_log(avctx, AV_LOG_ERROR, "Unexpected end of SEI NAL Unit parsing type.\n");
1529 return AVERROR_INVALIDDATA;
1533 sei_payload_size += *nal_data;
1536 } while (nal_size > 0 && *nal_data == 0xFF);
1538 if (nal_size < sei_payload_size) {
1539 av_log(avctx, AV_LOG_ERROR, "Unexpected end of SEI NAL Unit parsing size.\n");
1540 return AVERROR_INVALIDDATA;
1543 nal_data += sei_payload_size;
1544 nal_size -= sei_payload_size;
1547 *sei_end = nal_data;
1549 return nal_data - nal_start + 1;
1553 * Copies the data inserting emulation prevention bytes as needed.
1554 * Existing data in the destination can be taken into account by providing
1555 * dst with a dst_offset > 0.
1557 * @return The number of bytes copied on success. On failure, the negative of
1558 * the number of bytes needed to copy src is returned.
1560 static int copy_emulation_prev(const uint8_t *src,
1569 uint8_t* dst_end = dst + dst_size;
1570 const uint8_t* src_end = src + src_size;
1571 int start_at = dst_offset > 2 ? dst_offset - 2 : 0;
1573 for (i = start_at; i < dst_offset && i < dst_size; i++) {
1582 for (; src < src_end; src++, dst++) {
1584 int insert_ep3_byte = *src <= 3;
1585 if (insert_ep3_byte) {
1603 wrote_bytes = dst - dst_start;
1606 return -wrote_bytes;
1611 static int write_sei(const ExtraSEI *sei,
1616 uint8_t *sei_start = dst;
1617 size_t remaining_sei_size = sei->size;
1618 size_t remaining_dst_size = dst_size;
1623 if (!remaining_dst_size)
1624 return AVERROR_BUFFER_TOO_SMALL;
1626 while (sei_type && remaining_dst_size != 0) {
1627 int sei_byte = sei_type > 255 ? 255 : sei_type;
1630 sei_type -= sei_byte;
1632 remaining_dst_size--;
1636 return AVERROR_BUFFER_TOO_SMALL;
1638 while (remaining_sei_size && remaining_dst_size != 0) {
1639 int size_byte = remaining_sei_size > 255 ? 255 : remaining_sei_size;
1642 remaining_sei_size -= size_byte;
1644 remaining_dst_size--;
1647 if (remaining_dst_size < sei->size)
1648 return AVERROR_BUFFER_TOO_SMALL;
1650 header_bytes = dst - sei_start;
1652 offset = header_bytes;
1653 bytes_written = copy_emulation_prev(sei->data,
1658 if (bytes_written < 0)
1659 return AVERROR_BUFFER_TOO_SMALL;
1661 bytes_written += header_bytes;
1662 return bytes_written;
1666 * Copies NAL units and replaces length codes with
1667 * H.264 Annex B start codes. On failure, the contents of
1668 * dst_data may have been modified.
1670 * @param length_code_size Byte length of each length code
1671 * @param sample_buffer NAL units prefixed with length codes.
1672 * @param sei Optional A53 closed captions SEI data.
1673 * @param dst_data Must be zeroed before calling this function.
1674 * Contains the copied NAL units prefixed with
1675 * start codes when the function returns
1677 * @param dst_size Length of dst_data
1678 * @return 0 on success
1679 * AVERROR_INVALIDDATA if length_code_size is invalid
1680 * AVERROR_BUFFER_TOO_SMALL if dst_data is too small
1681 * or if a length_code in src_data specifies data beyond
1682 * the end of its buffer.
1684 static int copy_replace_length_codes(
1685 AVCodecContext *avctx,
1686 size_t length_code_size,
1687 CMSampleBufferRef sample_buffer,
1692 size_t src_size = CMSampleBufferGetTotalSampleSize(sample_buffer);
1693 size_t remaining_src_size = src_size;
1694 size_t remaining_dst_size = dst_size;
1695 size_t src_offset = 0;
1698 uint8_t size_buf[4];
1700 CMBlockBufferRef block = CMSampleBufferGetDataBuffer(sample_buffer);
1702 if (length_code_size > 4) {
1703 return AVERROR_INVALIDDATA;
1706 while (remaining_src_size > 0) {
1707 size_t curr_src_len;
1708 size_t curr_dst_len;
1714 status = CMBlockBufferCopyDataBytes(block,
1719 av_log(avctx, AV_LOG_ERROR, "Cannot copy length: %d\n", status);
1720 return AVERROR_EXTERNAL;
1723 status = CMBlockBufferCopyDataBytes(block,
1724 src_offset + length_code_size,
1729 av_log(avctx, AV_LOG_ERROR, "Cannot copy type: %d\n", status);
1730 return AVERROR_EXTERNAL;
1735 for (i = 0; i < length_code_size; i++) {
1737 box_len |= size_buf[i];
1740 if (sei && !wrote_sei && is_post_sei_nal_type(nal_type)) {
1741 //No SEI NAL unit - insert.
1744 memcpy(dst_data, start_code, sizeof(start_code));
1745 dst_data += sizeof(start_code);
1746 remaining_dst_size -= sizeof(start_code);
1748 *dst_data = H264_NAL_SEI;
1750 remaining_dst_size--;
1752 wrote_bytes = write_sei(sei,
1753 H264_SEI_TYPE_USER_DATA_REGISTERED,
1755 remaining_dst_size);
1757 if (wrote_bytes < 0)
1760 remaining_dst_size -= wrote_bytes;
1761 dst_data += wrote_bytes;
1763 if (remaining_dst_size <= 0)
1764 return AVERROR_BUFFER_TOO_SMALL;
1769 remaining_dst_size--;
1774 curr_src_len = box_len + length_code_size;
1775 curr_dst_len = box_len + sizeof(start_code);
1777 if (remaining_src_size < curr_src_len) {
1778 return AVERROR_BUFFER_TOO_SMALL;
1781 if (remaining_dst_size < curr_dst_len) {
1782 return AVERROR_BUFFER_TOO_SMALL;
1785 dst_box = dst_data + sizeof(start_code);
1787 memcpy(dst_data, start_code, sizeof(start_code));
1788 status = CMBlockBufferCopyDataBytes(block,
1789 src_offset + length_code_size,
1794 av_log(avctx, AV_LOG_ERROR, "Cannot copy data: %d\n", status);
1795 return AVERROR_EXTERNAL;
1798 if (sei && !wrote_sei && nal_type == H264_NAL_SEI) {
1799 //Found SEI NAL unit - append.
1804 old_sei_length = find_sei_end(avctx, dst_box, box_len, &new_sei);
1805 if (old_sei_length < 0)
1808 wrote_bytes = write_sei(sei,
1809 H264_SEI_TYPE_USER_DATA_REGISTERED,
1811 remaining_dst_size - old_sei_length);
1812 if (wrote_bytes < 0)
1815 if (new_sei + wrote_bytes >= dst_data + remaining_dst_size)
1816 return AVERROR_BUFFER_TOO_SMALL;
1818 new_sei[wrote_bytes++] = 0x80;
1819 extra_bytes = wrote_bytes - (dst_box + box_len - new_sei);
1821 dst_data += extra_bytes;
1822 remaining_dst_size -= extra_bytes;
1827 src_offset += curr_src_len;
1828 dst_data += curr_dst_len;
1830 remaining_src_size -= curr_src_len;
1831 remaining_dst_size -= curr_dst_len;
1838 * Returns a sufficient number of bytes to contain the sei data.
1839 * It may be greater than the minimum required.
1841 static int get_sei_msg_bytes(const ExtraSEI* sei, int type){
1846 copied_size = -copy_emulation_prev(sei->data,
1852 if ((sei->size % 255) == 0) //may result in an extra byte
1855 return copied_size + sei->size / 255 + 1 + type / 255 + 1;
1858 static int vtenc_cm_to_avpacket(
1859 AVCodecContext *avctx,
1860 CMSampleBufferRef sample_buffer,
1864 VTEncContext *vtctx = avctx->priv_data;
1869 size_t length_code_size;
1870 size_t header_size = 0;
1872 size_t out_buf_size;
1873 size_t sei_nalu_size = 0;
1875 int64_t time_base_num;
1879 CMVideoFormatDescriptionRef vid_fmt;
1882 vtenc_get_frame_info(sample_buffer, &is_key_frame);
1883 status = get_length_code_size(avctx, sample_buffer, &length_code_size);
1884 if (status) return status;
1886 add_header = is_key_frame && !(avctx->flags & AV_CODEC_FLAG_GLOBAL_HEADER);
1889 vid_fmt = CMSampleBufferGetFormatDescription(sample_buffer);
1891 av_log(avctx, AV_LOG_ERROR, "Cannot get format description.\n");
1892 return AVERROR_EXTERNAL;
1895 int status = get_params_size(avctx, vid_fmt, &header_size);
1896 if (status) return status;
1899 status = count_nalus(length_code_size, sample_buffer, &nalu_count);
1904 size_t msg_size = get_sei_msg_bytes(sei,
1905 H264_SEI_TYPE_USER_DATA_REGISTERED);
1907 sei_nalu_size = sizeof(start_code) + 1 + msg_size + 1;
1910 in_buf_size = CMSampleBufferGetTotalSampleSize(sample_buffer);
1911 out_buf_size = header_size +
1914 nalu_count * ((int)sizeof(start_code) - (int)length_code_size);
1916 status = ff_alloc_packet2(avctx, pkt, out_buf_size, out_buf_size);
1921 status = copy_param_sets(avctx, vid_fmt, pkt->data, out_buf_size);
1922 if(status) return status;
1925 status = copy_replace_length_codes(
1930 pkt->data + header_size,
1931 pkt->size - header_size
1935 av_log(avctx, AV_LOG_ERROR, "Error copying packet data: %d\n", status);
1940 pkt->flags |= AV_PKT_FLAG_KEY;
1943 pts = CMSampleBufferGetPresentationTimeStamp(sample_buffer);
1944 dts = CMSampleBufferGetDecodeTimeStamp (sample_buffer);
1946 if (CMTIME_IS_INVALID(dts)) {
1947 if (!vtctx->has_b_frames) {
1950 av_log(avctx, AV_LOG_ERROR, "DTS is invalid.\n");
1951 return AVERROR_EXTERNAL;
1955 dts_delta = vtctx->dts_delta >= 0 ? vtctx->dts_delta : 0;
1956 time_base_num = avctx->time_base.num;
1957 pkt->pts = pts.value / time_base_num;
1958 pkt->dts = dts.value / time_base_num - dts_delta;
1959 pkt->size = out_buf_size;
1965 * contiguous_buf_size is 0 if not contiguous, and the size of the buffer
1966 * containing all planes if so.
1968 static int get_cv_pixel_info(
1969 AVCodecContext *avctx,
1970 const AVFrame *frame,
1976 size_t *contiguous_buf_size)
1978 VTEncContext *vtctx = avctx->priv_data;
1979 int av_format = frame->format;
1980 int av_color_range = frame->color_range;
1985 status = get_cv_pixel_format(avctx, av_format, av_color_range, color, &range_guessed);
1989 "Could not get pixel format for color format '%s' range '%s'.\n",
1990 av_get_pix_fmt_name(av_format),
1991 av_color_range > AVCOL_RANGE_UNSPECIFIED &&
1992 av_color_range < AVCOL_RANGE_NB ?
1993 av_color_range_name(av_color_range) :
1996 return AVERROR(EINVAL);
1999 if (range_guessed) {
2000 if (!vtctx->warned_color_range) {
2001 vtctx->warned_color_range = true;
2004 "Color range not set for %s. Using MPEG range.\n",
2005 av_get_pix_fmt_name(av_format));
2009 switch (av_format) {
2010 case AV_PIX_FMT_NV12:
2013 widths [0] = avctx->width;
2014 heights[0] = avctx->height;
2015 strides[0] = frame ? frame->linesize[0] : avctx->width;
2017 widths [1] = (avctx->width + 1) / 2;
2018 heights[1] = (avctx->height + 1) / 2;
2019 strides[1] = frame ? frame->linesize[1] : (avctx->width + 1) & -2;
2022 case AV_PIX_FMT_YUV420P:
2025 widths [0] = avctx->width;
2026 heights[0] = avctx->height;
2027 strides[0] = frame ? frame->linesize[0] : avctx->width;
2029 widths [1] = (avctx->width + 1) / 2;
2030 heights[1] = (avctx->height + 1) / 2;
2031 strides[1] = frame ? frame->linesize[1] : (avctx->width + 1) / 2;
2033 widths [2] = (avctx->width + 1) / 2;
2034 heights[2] = (avctx->height + 1) / 2;
2035 strides[2] = frame ? frame->linesize[2] : (avctx->width + 1) / 2;
2038 case AV_PIX_FMT_P010LE:
2040 widths[0] = avctx->width;
2041 heights[0] = avctx->height;
2042 strides[0] = frame ? frame->linesize[0] : (avctx->width * 2 + 63) & -64;
2044 widths[1] = (avctx->width + 1) / 2;
2045 heights[1] = (avctx->height + 1) / 2;
2046 strides[1] = frame ? frame->linesize[1] : ((avctx->width + 1) / 2 + 63) & -64;
2053 "Could not get frame format info for color %d range %d.\n",
2057 return AVERROR(EINVAL);
2060 *contiguous_buf_size = 0;
2061 for (i = 0; i < *plane_count; i++) {
2062 if (i < *plane_count - 1 &&
2063 frame->data[i] + strides[i] * heights[i] != frame->data[i + 1]) {
2064 *contiguous_buf_size = 0;
2068 *contiguous_buf_size += strides[i] * heights[i];
2074 //Not used on OSX - frame is never copied.
2075 static int copy_avframe_to_pixel_buffer(AVCodecContext *avctx,
2076 const AVFrame *frame,
2077 CVPixelBufferRef cv_img,
2078 const size_t *plane_strides,
2079 const size_t *plane_rows)
2091 status = CVPixelBufferLockBaseAddress(cv_img, 0);
2096 "Error: Could not lock base address of CVPixelBuffer: %d.\n",
2101 if (CVPixelBufferIsPlanar(cv_img)) {
2102 plane_count = CVPixelBufferGetPlaneCount(cv_img);
2103 for (i = 0; frame->data[i]; i++) {
2104 if (i == plane_count) {
2105 CVPixelBufferUnlockBaseAddress(cv_img, 0);
2108 "Error: different number of planes in AVFrame and CVPixelBuffer.\n"
2111 return AVERROR_EXTERNAL;
2114 dst_addr = (uint8_t*)CVPixelBufferGetBaseAddressOfPlane(cv_img, i);
2115 src_addr = (uint8_t*)frame->data[i];
2116 dst_stride = CVPixelBufferGetBytesPerRowOfPlane(cv_img, i);
2117 src_stride = plane_strides[i];
2118 rows = plane_rows[i];
2120 if (dst_stride == src_stride) {
2121 memcpy(dst_addr, src_addr, src_stride * rows);
2123 copy_bytes = dst_stride < src_stride ? dst_stride : src_stride;
2125 for (j = 0; j < rows; j++) {
2126 memcpy(dst_addr + j * dst_stride, src_addr + j * src_stride, copy_bytes);
2131 if (frame->data[1]) {
2132 CVPixelBufferUnlockBaseAddress(cv_img, 0);
2135 "Error: different number of planes in AVFrame and non-planar CVPixelBuffer.\n"
2138 return AVERROR_EXTERNAL;
2141 dst_addr = (uint8_t*)CVPixelBufferGetBaseAddress(cv_img);
2142 src_addr = (uint8_t*)frame->data[0];
2143 dst_stride = CVPixelBufferGetBytesPerRow(cv_img);
2144 src_stride = plane_strides[0];
2145 rows = plane_rows[0];
2147 if (dst_stride == src_stride) {
2148 memcpy(dst_addr, src_addr, src_stride * rows);
2150 copy_bytes = dst_stride < src_stride ? dst_stride : src_stride;
2152 for (j = 0; j < rows; j++) {
2153 memcpy(dst_addr + j * dst_stride, src_addr + j * src_stride, copy_bytes);
2158 status = CVPixelBufferUnlockBaseAddress(cv_img, 0);
2160 av_log(avctx, AV_LOG_ERROR, "Error: Could not unlock CVPixelBuffer base address: %d.\n", status);
2161 return AVERROR_EXTERNAL;
2167 static int create_cv_pixel_buffer(AVCodecContext *avctx,
2168 const AVFrame *frame,
2169 CVPixelBufferRef *cv_img)
2173 size_t widths [AV_NUM_DATA_POINTERS];
2174 size_t heights[AV_NUM_DATA_POINTERS];
2175 size_t strides[AV_NUM_DATA_POINTERS];
2177 size_t contiguous_buf_size;
2178 CVPixelBufferPoolRef pix_buf_pool;
2179 VTEncContext* vtctx = avctx->priv_data;
2181 if (avctx->pix_fmt == AV_PIX_FMT_VIDEOTOOLBOX) {
2182 av_assert0(frame->format == AV_PIX_FMT_VIDEOTOOLBOX);
2184 *cv_img = (CVPixelBufferRef)frame->data[3];
2185 av_assert0(*cv_img);
2191 memset(widths, 0, sizeof(widths));
2192 memset(heights, 0, sizeof(heights));
2193 memset(strides, 0, sizeof(strides));
2195 status = get_cv_pixel_info(
2203 &contiguous_buf_size
2210 "Error: Cannot convert format %d color_range %d: %d\n",
2216 return AVERROR_EXTERNAL;
2219 pix_buf_pool = VTCompressionSessionGetPixelBufferPool(vtctx->session);
2220 if (!pix_buf_pool) {
2221 /* On iOS, the VT session is invalidated when the APP switches from
2222 * foreground to background and vice versa. Fetch the actual error code
2223 * of the VT session to detect that case and restart the VT session
2227 vtstatus = VTCompressionSessionPrepareToEncodeFrames(vtctx->session);
2228 if (vtstatus == kVTInvalidSessionErr) {
2229 CFRelease(vtctx->session);
2230 vtctx->session = NULL;
2231 status = vtenc_configure_encoder(avctx);
2233 pix_buf_pool = VTCompressionSessionGetPixelBufferPool(vtctx->session);
2235 if (!pix_buf_pool) {
2236 av_log(avctx, AV_LOG_ERROR, "Could not get pixel buffer pool.\n");
2237 return AVERROR_EXTERNAL;
2240 av_log(avctx, AV_LOG_WARNING, "VT session restarted because of a "
2241 "kVTInvalidSessionErr error.\n");
2244 status = CVPixelBufferPoolCreatePixelBuffer(NULL,
2250 av_log(avctx, AV_LOG_ERROR, "Could not create pixel buffer from pool: %d.\n", status);
2251 return AVERROR_EXTERNAL;
2254 status = copy_avframe_to_pixel_buffer(avctx, frame, *cv_img, strides, heights);
2264 static int create_encoder_dict_h264(const AVFrame *frame,
2265 CFDictionaryRef* dict_out)
2267 CFDictionaryRef dict = NULL;
2268 if (frame->pict_type == AV_PICTURE_TYPE_I) {
2269 const void *keys[] = { kVTEncodeFrameOptionKey_ForceKeyFrame };
2270 const void *vals[] = { kCFBooleanTrue };
2272 dict = CFDictionaryCreate(NULL, keys, vals, 1, NULL, NULL);
2273 if(!dict) return AVERROR(ENOMEM);
2280 static int vtenc_send_frame(AVCodecContext *avctx,
2281 VTEncContext *vtctx,
2282 const AVFrame *frame)
2285 CFDictionaryRef frame_dict;
2286 CVPixelBufferRef cv_img = NULL;
2287 AVFrameSideData *side_data = NULL;
2288 ExtraSEI *sei = NULL;
2289 int status = create_cv_pixel_buffer(avctx, frame, &cv_img);
2291 if (status) return status;
2293 status = create_encoder_dict_h264(frame, &frame_dict);
2299 side_data = av_frame_get_side_data(frame, AV_FRAME_DATA_A53_CC);
2300 if (vtctx->a53_cc && side_data && side_data->size) {
2301 sei = av_mallocz(sizeof(*sei));
2303 av_log(avctx, AV_LOG_ERROR, "Not enough memory for closed captions, skipping\n");
2305 int ret = ff_alloc_a53_sei(frame, 0, &sei->data, &sei->size);
2307 av_log(avctx, AV_LOG_ERROR, "Not enough memory for closed captions, skipping\n");
2314 time = CMTimeMake(frame->pts * avctx->time_base.num, avctx->time_base.den);
2315 status = VTCompressionSessionEncodeFrame(
2325 if (frame_dict) CFRelease(frame_dict);
2329 av_log(avctx, AV_LOG_ERROR, "Error: cannot encode frame: %d\n", status);
2330 return AVERROR_EXTERNAL;
2336 static av_cold int vtenc_frame(
2337 AVCodecContext *avctx,
2339 const AVFrame *frame,
2342 VTEncContext *vtctx = avctx->priv_data;
2345 CMSampleBufferRef buf = NULL;
2346 ExtraSEI *sei = NULL;
2349 status = vtenc_send_frame(avctx, vtctx, frame);
2352 status = AVERROR_EXTERNAL;
2356 if (vtctx->frame_ct_in == 0) {
2357 vtctx->first_pts = frame->pts;
2358 } else if(vtctx->frame_ct_in == 1 && vtctx->has_b_frames) {
2359 vtctx->dts_delta = frame->pts - vtctx->first_pts;
2362 vtctx->frame_ct_in++;
2363 } else if(!vtctx->flushing) {
2364 vtctx->flushing = true;
2366 status = VTCompressionSessionCompleteFrames(vtctx->session,
2370 av_log(avctx, AV_LOG_ERROR, "Error flushing frames: %d\n", status);
2371 status = AVERROR_EXTERNAL;
2377 get_frame = vtctx->dts_delta >= 0 || !frame;
2383 status = vtenc_q_pop(vtctx, !frame, &buf, &sei);
2384 if (status) goto end_nopkt;
2385 if (!buf) goto end_nopkt;
2387 status = vtenc_cm_to_avpacket(avctx, buf, pkt, sei);
2389 if (sei->data) av_free(sei->data);
2393 if (status) goto end_nopkt;
2399 av_packet_unref(pkt);
2403 static int vtenc_populate_extradata(AVCodecContext *avctx,
2404 CMVideoCodecType codec_type,
2405 CFStringRef profile_level,
2406 CFNumberRef gamma_level,
2407 CFDictionaryRef enc_info,
2408 CFDictionaryRef pixel_buffer_info)
2410 VTEncContext *vtctx = avctx->priv_data;
2412 CVPixelBufferPoolRef pool = NULL;
2413 CVPixelBufferRef pix_buf = NULL;
2415 CMSampleBufferRef buf = NULL;
2417 status = vtenc_create_encoder(avctx,
2427 pool = VTCompressionSessionGetPixelBufferPool(vtctx->session);
2429 av_log(avctx, AV_LOG_ERROR, "Error getting pixel buffer pool.\n");
2433 status = CVPixelBufferPoolCreatePixelBuffer(NULL,
2437 if(status != kCVReturnSuccess){
2438 av_log(avctx, AV_LOG_ERROR, "Error creating frame from pool: %d\n", status);
2442 time = CMTimeMake(0, avctx->time_base.den);
2443 status = VTCompressionSessionEncodeFrame(vtctx->session,
2454 "Error sending frame for extradata: %d\n",
2460 //Populates extradata - output frames are flushed and param sets are available.
2461 status = VTCompressionSessionCompleteFrames(vtctx->session,
2467 status = vtenc_q_pop(vtctx, 0, &buf, NULL);
2469 av_log(avctx, AV_LOG_ERROR, "popping: %d\n", status);
2479 CFRelease(vtctx->session);
2481 vtctx->session = NULL;
2482 vtctx->frame_ct_out = 0;
2484 av_assert0(status != 0 || (avctx->extradata && avctx->extradata_size > 0));
2489 static av_cold int vtenc_close(AVCodecContext *avctx)
2491 VTEncContext *vtctx = avctx->priv_data;
2493 pthread_cond_destroy(&vtctx->cv_sample_sent);
2494 pthread_mutex_destroy(&vtctx->lock);
2496 if(!vtctx->session) return 0;
2498 VTCompressionSessionCompleteFrames(vtctx->session,
2500 clear_frame_queue(vtctx);
2501 CFRelease(vtctx->session);
2502 vtctx->session = NULL;
2504 if (vtctx->color_primaries) {
2505 CFRelease(vtctx->color_primaries);
2506 vtctx->color_primaries = NULL;
2509 if (vtctx->transfer_function) {
2510 CFRelease(vtctx->transfer_function);
2511 vtctx->transfer_function = NULL;
2514 if (vtctx->ycbcr_matrix) {
2515 CFRelease(vtctx->ycbcr_matrix);
2516 vtctx->ycbcr_matrix = NULL;
2522 static const enum AVPixelFormat avc_pix_fmts[] = {
2523 AV_PIX_FMT_VIDEOTOOLBOX,
2529 static const enum AVPixelFormat hevc_pix_fmts[] = {
2530 AV_PIX_FMT_VIDEOTOOLBOX,
2537 #define VE AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
2538 #define COMMON_OPTIONS \
2539 { "allow_sw", "Allow software encoding", OFFSET(allow_sw), AV_OPT_TYPE_BOOL, \
2540 { .i64 = 0 }, 0, 1, VE }, \
2541 { "require_sw", "Require software encoding", OFFSET(require_sw), AV_OPT_TYPE_BOOL, \
2542 { .i64 = 0 }, 0, 1, VE }, \
2543 { "realtime", "Hint that encoding should happen in real-time if not faster (e.g. capturing from camera).", \
2544 OFFSET(realtime), AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, VE }, \
2545 { "frames_before", "Other frames will come before the frames in this session. This helps smooth concatenation issues.", \
2546 OFFSET(frames_before), AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, VE }, \
2547 { "frames_after", "Other frames will come after the frames in this session. This helps smooth concatenation issues.", \
2548 OFFSET(frames_after), AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, VE },
2550 #define OFFSET(x) offsetof(VTEncContext, x)
2551 static const AVOption h264_options[] = {
2552 { "profile", "Profile", OFFSET(profile), AV_OPT_TYPE_INT, { .i64 = H264_PROF_AUTO }, H264_PROF_AUTO, H264_PROF_COUNT, VE, "profile" },
2553 { "baseline", "Baseline Profile", 0, AV_OPT_TYPE_CONST, { .i64 = H264_PROF_BASELINE }, INT_MIN, INT_MAX, VE, "profile" },
2554 { "main", "Main Profile", 0, AV_OPT_TYPE_CONST, { .i64 = H264_PROF_MAIN }, INT_MIN, INT_MAX, VE, "profile" },
2555 { "high", "High Profile", 0, AV_OPT_TYPE_CONST, { .i64 = H264_PROF_HIGH }, INT_MIN, INT_MAX, VE, "profile" },
2556 { "extended", "Extend Profile", 0, AV_OPT_TYPE_CONST, { .i64 = H264_PROF_EXTENDED }, INT_MIN, INT_MAX, VE, "profile" },
2558 { "level", "Level", OFFSET(level), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 52, VE, "level" },
2559 { "1.3", "Level 1.3, only available with Baseline Profile", 0, AV_OPT_TYPE_CONST, { .i64 = 13 }, INT_MIN, INT_MAX, VE, "level" },
2560 { "3.0", "Level 3.0", 0, AV_OPT_TYPE_CONST, { .i64 = 30 }, INT_MIN, INT_MAX, VE, "level" },
2561 { "3.1", "Level 3.1", 0, AV_OPT_TYPE_CONST, { .i64 = 31 }, INT_MIN, INT_MAX, VE, "level" },
2562 { "3.2", "Level 3.2", 0, AV_OPT_TYPE_CONST, { .i64 = 32 }, INT_MIN, INT_MAX, VE, "level" },
2563 { "4.0", "Level 4.0", 0, AV_OPT_TYPE_CONST, { .i64 = 40 }, INT_MIN, INT_MAX, VE, "level" },
2564 { "4.1", "Level 4.1", 0, AV_OPT_TYPE_CONST, { .i64 = 41 }, INT_MIN, INT_MAX, VE, "level" },
2565 { "4.2", "Level 4.2", 0, AV_OPT_TYPE_CONST, { .i64 = 42 }, INT_MIN, INT_MAX, VE, "level" },
2566 { "5.0", "Level 5.0", 0, AV_OPT_TYPE_CONST, { .i64 = 50 }, INT_MIN, INT_MAX, VE, "level" },
2567 { "5.1", "Level 5.1", 0, AV_OPT_TYPE_CONST, { .i64 = 51 }, INT_MIN, INT_MAX, VE, "level" },
2568 { "5.2", "Level 5.2", 0, AV_OPT_TYPE_CONST, { .i64 = 52 }, INT_MIN, INT_MAX, VE, "level" },
2570 { "coder", "Entropy coding", OFFSET(entropy), AV_OPT_TYPE_INT, { .i64 = VT_ENTROPY_NOT_SET }, VT_ENTROPY_NOT_SET, VT_CABAC, VE, "coder" },
2571 { "cavlc", "CAVLC entropy coding", 0, AV_OPT_TYPE_CONST, { .i64 = VT_CAVLC }, INT_MIN, INT_MAX, VE, "coder" },
2572 { "vlc", "CAVLC entropy coding", 0, AV_OPT_TYPE_CONST, { .i64 = VT_CAVLC }, INT_MIN, INT_MAX, VE, "coder" },
2573 { "cabac", "CABAC entropy coding", 0, AV_OPT_TYPE_CONST, { .i64 = VT_CABAC }, INT_MIN, INT_MAX, VE, "coder" },
2574 { "ac", "CABAC entropy coding", 0, AV_OPT_TYPE_CONST, { .i64 = VT_CABAC }, INT_MIN, INT_MAX, VE, "coder" },
2576 { "a53cc", "Use A53 Closed Captions (if available)", OFFSET(a53_cc), AV_OPT_TYPE_BOOL, {.i64 = 1}, 0, 1, VE },
2582 static const AVClass h264_videotoolbox_class = {
2583 .class_name = "h264_videotoolbox",
2584 .item_name = av_default_item_name,
2585 .option = h264_options,
2586 .version = LIBAVUTIL_VERSION_INT,
2589 AVCodec ff_h264_videotoolbox_encoder = {
2590 .name = "h264_videotoolbox",
2591 .long_name = NULL_IF_CONFIG_SMALL("VideoToolbox H.264 Encoder"),
2592 .type = AVMEDIA_TYPE_VIDEO,
2593 .id = AV_CODEC_ID_H264,
2594 .priv_data_size = sizeof(VTEncContext),
2595 .pix_fmts = avc_pix_fmts,
2597 .encode2 = vtenc_frame,
2598 .close = vtenc_close,
2599 .capabilities = AV_CODEC_CAP_DELAY,
2600 .priv_class = &h264_videotoolbox_class,
2601 .caps_internal = FF_CODEC_CAP_INIT_THREADSAFE |
2602 FF_CODEC_CAP_INIT_CLEANUP,
2605 static const AVOption hevc_options[] = {
2606 { "profile", "Profile", OFFSET(profile), AV_OPT_TYPE_INT, { .i64 = HEVC_PROF_AUTO }, HEVC_PROF_AUTO, HEVC_PROF_COUNT, VE, "profile" },
2607 { "main", "Main Profile", 0, AV_OPT_TYPE_CONST, { .i64 = HEVC_PROF_MAIN }, INT_MIN, INT_MAX, VE, "profile" },
2608 { "main10", "Main10 Profile", 0, AV_OPT_TYPE_CONST, { .i64 = HEVC_PROF_MAIN10 }, INT_MIN, INT_MAX, VE, "profile" },
2614 static const AVClass hevc_videotoolbox_class = {
2615 .class_name = "hevc_videotoolbox",
2616 .item_name = av_default_item_name,
2617 .option = hevc_options,
2618 .version = LIBAVUTIL_VERSION_INT,
2621 AVCodec ff_hevc_videotoolbox_encoder = {
2622 .name = "hevc_videotoolbox",
2623 .long_name = NULL_IF_CONFIG_SMALL("VideoToolbox H.265 Encoder"),
2624 .type = AVMEDIA_TYPE_VIDEO,
2625 .id = AV_CODEC_ID_HEVC,
2626 .priv_data_size = sizeof(VTEncContext),
2627 .pix_fmts = hevc_pix_fmts,
2629 .encode2 = vtenc_frame,
2630 .close = vtenc_close,
2631 .capabilities = AV_CODEC_CAP_DELAY | AV_CODEC_CAP_HARDWARE,
2632 .priv_class = &hevc_videotoolbox_class,
2633 .caps_internal = FF_CODEC_CAP_INIT_THREADSAFE |
2634 FF_CODEC_CAP_INIT_CLEANUP,
2635 .wrapper_name = "videotoolbox",