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/atomic.h"
30 #include "libavutil/avstring.h"
31 #include "libavcodec/avcodec.h"
32 #include "libavutil/pixdesc.h"
37 # define kCVImageBufferColorPrimaries_ITU_R_2020 CFSTR("ITU_R_2020")
38 # define kCVImageBufferTransferFunction_ITU_R_2020 CFSTR("ITU_R_2020")
39 # define kCVImageBufferYCbCrMatrix_ITU_R_2020 CFSTR("ITU_R_2020")
42 typedef enum VT_H264Profile {
50 typedef enum VTH264Entropy{
56 static const uint8_t start_code[] = { 0, 0, 0, 1 };
58 typedef struct BufNode {
59 CMSampleBufferRef cm_buffer;
64 typedef struct VTEncContext {
66 VTCompressionSessionRef session;
67 CFStringRef ycbcr_matrix;
68 CFStringRef color_primaries;
69 CFStringRef transfer_function;
72 pthread_cond_t cv_sample_sent;
89 int64_t frames_before;
96 bool warned_color_range;
100 * NULL-safe release of *refPtr, and sets value to NULL.
102 static void vt_release_num(CFNumberRef* refPtr){
111 static void set_async_error(VTEncContext *vtctx, int err)
115 pthread_mutex_lock(&vtctx->lock);
117 vtctx->async_error = err;
119 info = vtctx->q_head;
120 vtctx->q_head = vtctx->q_tail = NULL;
123 BufNode *next = info->next;
124 CFRelease(info->cm_buffer);
129 pthread_mutex_unlock(&vtctx->lock);
132 static int vtenc_q_pop(VTEncContext *vtctx, bool wait, CMSampleBufferRef *buf)
136 pthread_mutex_lock(&vtctx->lock);
138 if (vtctx->async_error) {
139 pthread_mutex_unlock(&vtctx->lock);
140 return vtctx->async_error;
143 if (vtctx->flushing && vtctx->frame_ct_in == vtctx->frame_ct_out) {
146 pthread_mutex_unlock(&vtctx->lock);
150 while (!vtctx->q_head && !vtctx->async_error && wait) {
151 pthread_cond_wait(&vtctx->cv_sample_sent, &vtctx->lock);
154 if (!vtctx->q_head) {
155 pthread_mutex_unlock(&vtctx->lock);
160 info = vtctx->q_head;
161 vtctx->q_head = vtctx->q_head->next;
162 if (!vtctx->q_head) {
163 vtctx->q_tail = NULL;
166 pthread_mutex_unlock(&vtctx->lock);
168 *buf = info->cm_buffer;
171 vtctx->frame_ct_out++;
176 static void vtenc_q_push(VTEncContext *vtctx, CMSampleBufferRef buffer)
178 BufNode *info = av_malloc(sizeof(BufNode));
180 set_async_error(vtctx, AVERROR(ENOMEM));
185 info->cm_buffer = buffer;
188 pthread_mutex_lock(&vtctx->lock);
189 pthread_cond_signal(&vtctx->cv_sample_sent);
191 if (!vtctx->q_head) {
192 vtctx->q_head = info;
194 vtctx->q_tail->next = info;
197 vtctx->q_tail = info;
199 pthread_mutex_unlock(&vtctx->lock);
202 static CMVideoCodecType get_cm_codec_type(enum AVCodecID id)
205 case AV_CODEC_ID_H264: return kCMVideoCodecType_H264;
210 static void vtenc_free_block(void *opaque, uint8_t *data)
212 CMBlockBufferRef block = opaque;
217 * Get the parameter sets from a CMSampleBufferRef.
218 * @param dst If *dst isn't NULL, the parameters are copied into existing
219 * memory. *dst_size must be set accordingly when *dst != NULL.
220 * If *dst is NULL, it will be allocated.
221 * In all cases, *dst_size is set to the number of bytes used starting
224 static int get_params_size(
225 AVCodecContext *avctx,
226 CMVideoFormatDescriptionRef vid_fmt,
229 size_t total_size = 0;
231 int is_count_bad = 0;
234 status = CMVideoFormatDescriptionGetH264ParameterSetAtIndex(vid_fmt,
246 for (i = 0; i < ps_count || is_count_bad; i++) {
249 status = CMVideoFormatDescriptionGetH264ParameterSetAtIndex(vid_fmt,
257 * When ps_count is invalid, status != 0 ends the loop normally
258 * unless we didn't get any parameter sets.
260 if (i > 0 && is_count_bad) status = 0;
265 total_size += ps_size + sizeof(start_code);
269 av_log(avctx, AV_LOG_ERROR, "Error getting parameter set sizes: %d\n", status);
270 return AVERROR_EXTERNAL;
277 static int copy_param_sets(
278 AVCodecContext *avctx,
279 CMVideoFormatDescriptionRef vid_fmt,
284 int is_count_bad = 0;
289 status = CMVideoFormatDescriptionGetH264ParameterSetAtIndex(vid_fmt,
302 for (i = 0; i < ps_count || is_count_bad; i++) {
307 status = CMVideoFormatDescriptionGetH264ParameterSetAtIndex(vid_fmt,
314 if (i > 0 && is_count_bad) status = 0;
319 next_offset = offset + sizeof(start_code) + ps_size;
320 if (dst_size < next_offset) {
321 av_log(avctx, AV_LOG_ERROR, "Error: buffer too small for parameter sets.\n");
322 return AVERROR_BUFFER_TOO_SMALL;
325 memcpy(dst + offset, start_code, sizeof(start_code));
326 offset += sizeof(start_code);
328 memcpy(dst + offset, ps, ps_size);
329 offset = next_offset;
333 av_log(avctx, AV_LOG_ERROR, "Error getting parameter set data: %d\n", status);
334 return AVERROR_EXTERNAL;
340 static int set_extradata(AVCodecContext *avctx, CMSampleBufferRef sample_buffer)
342 CMVideoFormatDescriptionRef vid_fmt;
346 vid_fmt = CMSampleBufferGetFormatDescription(sample_buffer);
348 av_log(avctx, AV_LOG_ERROR, "No video format.\n");
349 return AVERROR_EXTERNAL;
352 status = get_params_size(avctx, vid_fmt, &total_size);
354 av_log(avctx, AV_LOG_ERROR, "Could not get parameter sets.\n");
358 avctx->extradata = av_malloc(total_size);
359 if (!avctx->extradata) {
360 return AVERROR(ENOMEM);
362 avctx->extradata_size = total_size;
364 status = copy_param_sets(avctx, vid_fmt, avctx->extradata, total_size);
367 av_log(avctx, AV_LOG_ERROR, "Could not copy param sets.\n");
374 static void vtenc_output_callback(
376 void *sourceFrameCtx,
378 VTEncodeInfoFlags flags,
379 CMSampleBufferRef sample_buffer)
381 AVCodecContext *avctx = ctx;
382 VTEncContext *vtctx = avctx->priv_data;
384 if (vtctx->async_error) {
385 if(sample_buffer) CFRelease(sample_buffer);
389 if (status || !sample_buffer) {
390 av_log(avctx, AV_LOG_ERROR, "Error encoding frame: %d\n", (int)status);
391 set_async_error(vtctx, AVERROR_EXTERNAL);
395 if (!avctx->extradata && (avctx->flags & AV_CODEC_FLAG_GLOBAL_HEADER)) {
396 int set_status = set_extradata(avctx, sample_buffer);
398 set_async_error(vtctx, set_status);
403 vtenc_q_push(vtctx, sample_buffer);
406 static int get_length_code_size(
407 AVCodecContext *avctx,
408 CMSampleBufferRef sample_buffer,
411 CMVideoFormatDescriptionRef vid_fmt;
415 vid_fmt = CMSampleBufferGetFormatDescription(sample_buffer);
417 av_log(avctx, AV_LOG_ERROR, "Error getting buffer format description.\n");
418 return AVERROR_EXTERNAL;
421 status = CMVideoFormatDescriptionGetH264ParameterSetAtIndex(vid_fmt,
428 av_log(avctx, AV_LOG_ERROR, "Error getting length code size: %d\n", status);
429 return AVERROR_EXTERNAL;
437 * Returns true on success.
439 * If profile_level_val is NULL and this method returns true, don't specify the
440 * profile/level to the encoder.
442 static bool get_vt_profile_level(AVCodecContext *avctx,
443 CFStringRef *profile_level_val)
445 VTEncContext *vtctx = avctx->priv_data;
446 int64_t profile = vtctx->profile;
448 if (profile == H264_PROF_AUTO && vtctx->level) {
449 //Need to pick a profile if level is not auto-selected.
450 profile = vtctx->has_b_frames ? H264_PROF_MAIN : H264_PROF_BASELINE;
453 *profile_level_val = NULL;
459 case H264_PROF_BASELINE:
460 switch (vtctx->level) {
461 case 0: *profile_level_val = kVTProfileLevel_H264_Baseline_AutoLevel; break;
462 case 13: *profile_level_val = kVTProfileLevel_H264_Baseline_1_3; break;
463 case 30: *profile_level_val = kVTProfileLevel_H264_Baseline_3_0; break;
464 case 31: *profile_level_val = kVTProfileLevel_H264_Baseline_3_1; break;
465 case 32: *profile_level_val = kVTProfileLevel_H264_Baseline_3_2; break;
466 case 40: *profile_level_val = kVTProfileLevel_H264_Baseline_4_0; break;
467 case 41: *profile_level_val = kVTProfileLevel_H264_Baseline_4_1; break;
468 case 42: *profile_level_val = kVTProfileLevel_H264_Baseline_4_2; break;
469 case 50: *profile_level_val = kVTProfileLevel_H264_Baseline_5_0; break;
470 case 51: *profile_level_val = kVTProfileLevel_H264_Baseline_5_1; break;
471 case 52: *profile_level_val = kVTProfileLevel_H264_Baseline_5_2; break;
476 switch (vtctx->level) {
477 case 0: *profile_level_val = kVTProfileLevel_H264_Main_AutoLevel; break;
478 case 30: *profile_level_val = kVTProfileLevel_H264_Main_3_0; break;
479 case 31: *profile_level_val = kVTProfileLevel_H264_Main_3_1; break;
480 case 32: *profile_level_val = kVTProfileLevel_H264_Main_3_2; break;
481 case 40: *profile_level_val = kVTProfileLevel_H264_Main_4_0; break;
482 case 41: *profile_level_val = kVTProfileLevel_H264_Main_4_1; break;
483 case 42: *profile_level_val = kVTProfileLevel_H264_Main_4_2; break;
484 case 50: *profile_level_val = kVTProfileLevel_H264_Main_5_0; break;
485 case 51: *profile_level_val = kVTProfileLevel_H264_Main_5_1; break;
486 case 52: *profile_level_val = kVTProfileLevel_H264_Main_5_2; break;
491 switch (vtctx->level) {
492 case 0: *profile_level_val = kVTProfileLevel_H264_High_AutoLevel; break;
493 case 30: *profile_level_val = kVTProfileLevel_H264_High_3_0; break;
494 case 31: *profile_level_val = kVTProfileLevel_H264_High_3_1; break;
495 case 32: *profile_level_val = kVTProfileLevel_H264_High_3_2; break;
496 case 40: *profile_level_val = kVTProfileLevel_H264_High_4_0; break;
497 case 41: *profile_level_val = kVTProfileLevel_H264_High_4_1; break;
498 case 42: *profile_level_val = kVTProfileLevel_H264_High_4_2; break;
499 case 50: *profile_level_val = kVTProfileLevel_H264_High_5_0; break;
500 case 51: *profile_level_val = kVTProfileLevel_H264_High_5_1; break;
501 case 52: *profile_level_val = kVTProfileLevel_H264_High_5_2; break;
506 if (!*profile_level_val) {
507 av_log(avctx, AV_LOG_ERROR, "Invalid Profile/Level.\n");
514 static int get_cv_pixel_format(AVCodecContext* avctx,
515 enum AVPixelFormat fmt,
516 enum AVColorRange range,
517 int* av_pixel_format,
520 if (range_guessed) *range_guessed = range != AVCOL_RANGE_MPEG &&
521 range != AVCOL_RANGE_JPEG;
523 //MPEG range is used when no range is set
524 if (fmt == AV_PIX_FMT_NV12) {
525 *av_pixel_format = range == AVCOL_RANGE_JPEG ?
526 kCVPixelFormatType_420YpCbCr8BiPlanarFullRange :
527 kCVPixelFormatType_420YpCbCr8BiPlanarVideoRange;
528 } else if (fmt == AV_PIX_FMT_YUV420P) {
529 *av_pixel_format = range == AVCOL_RANGE_JPEG ?
530 kCVPixelFormatType_420YpCbCr8PlanarFullRange :
531 kCVPixelFormatType_420YpCbCr8Planar;
533 return AVERROR(EINVAL);
539 static void add_color_attr(AVCodecContext *avctx, CFMutableDictionaryRef dict) {
540 VTEncContext *vtctx = avctx->priv_data;
542 if (vtctx->color_primaries) {
543 CFDictionarySetValue(dict,
544 kCVImageBufferColorPrimariesKey,
545 vtctx->color_primaries);
548 if (vtctx->transfer_function) {
549 CFDictionarySetValue(dict,
550 kCVImageBufferTransferFunctionKey,
551 vtctx->transfer_function);
554 if (vtctx->ycbcr_matrix) {
555 CFDictionarySetValue(dict,
556 kCVImageBufferYCbCrMatrixKey,
557 vtctx->ycbcr_matrix);
561 static int create_cv_pixel_buffer_info(AVCodecContext* avctx,
562 CFMutableDictionaryRef* dict)
564 CFNumberRef cv_color_format_num = NULL;
565 CFNumberRef width_num = NULL;
566 CFNumberRef height_num = NULL;
567 CFMutableDictionaryRef pixel_buffer_info = NULL;
569 int status = get_cv_pixel_format(avctx,
574 if (status) return status;
576 pixel_buffer_info = CFDictionaryCreateMutable(
579 &kCFCopyStringDictionaryKeyCallBacks,
580 &kCFTypeDictionaryValueCallBacks);
582 if (!pixel_buffer_info) goto pbinfo_nomem;
584 cv_color_format_num = CFNumberCreate(kCFAllocatorDefault,
587 if (!cv_color_format_num) goto pbinfo_nomem;
589 CFDictionarySetValue(pixel_buffer_info,
590 kCVPixelBufferPixelFormatTypeKey,
591 cv_color_format_num);
592 vt_release_num(&cv_color_format_num);
594 width_num = CFNumberCreate(kCFAllocatorDefault,
597 if (!width_num) return AVERROR(ENOMEM);
599 CFDictionarySetValue(pixel_buffer_info,
600 kCVPixelBufferWidthKey,
602 vt_release_num(&width_num);
604 height_num = CFNumberCreate(kCFAllocatorDefault,
607 if (!height_num) goto pbinfo_nomem;
609 CFDictionarySetValue(pixel_buffer_info,
610 kCVPixelBufferHeightKey,
612 vt_release_num(&height_num);
614 add_color_attr(avctx, pixel_buffer_info);
616 *dict = pixel_buffer_info;
620 vt_release_num(&cv_color_format_num);
621 vt_release_num(&width_num);
622 vt_release_num(&height_num);
623 if (pixel_buffer_info) CFRelease(pixel_buffer_info);
625 return AVERROR(ENOMEM);
628 static int get_cv_color_primaries(AVCodecContext *avctx,
629 CFStringRef *primaries)
631 enum AVColorPrimaries pri = avctx->color_primaries;
633 case AVCOL_PRI_UNSPECIFIED:
637 case AVCOL_PRI_BT709:
638 *primaries = kCVImageBufferColorPrimaries_ITU_R_709_2;
641 case AVCOL_PRI_BT2020:
642 *primaries = kCVImageBufferColorPrimaries_ITU_R_2020;
646 av_log(avctx, AV_LOG_ERROR, "Color primaries %s is not supported.\n", av_color_primaries_name(pri));
654 static int get_cv_transfer_function(AVCodecContext *avctx,
655 CFStringRef *transfer_fnc,
656 CFNumberRef *gamma_level)
658 enum AVColorTransferCharacteristic trc = avctx->color_trc;
663 case AVCOL_TRC_UNSPECIFIED:
664 *transfer_fnc = NULL;
667 case AVCOL_TRC_BT709:
668 *transfer_fnc = kCVImageBufferTransferFunction_ITU_R_709_2;
671 case AVCOL_TRC_SMPTE240M:
672 *transfer_fnc = kCVImageBufferTransferFunction_SMPTE_240M_1995;
675 case AVCOL_TRC_GAMMA22:
677 *transfer_fnc = kCVImageBufferTransferFunction_UseGamma;
678 *gamma_level = CFNumberCreate(NULL, kCFNumberFloat32Type, &gamma);
681 case AVCOL_TRC_GAMMA28:
683 *transfer_fnc = kCVImageBufferTransferFunction_UseGamma;
684 *gamma_level = CFNumberCreate(NULL, kCFNumberFloat32Type, &gamma);
687 case AVCOL_TRC_BT2020_10:
688 case AVCOL_TRC_BT2020_12:
689 *transfer_fnc = kCVImageBufferTransferFunction_ITU_R_2020;
693 av_log(avctx, AV_LOG_ERROR, "Transfer function %s is not supported.\n", av_color_transfer_name(trc));
700 static int get_cv_ycbcr_matrix(AVCodecContext *avctx, CFStringRef *matrix) {
701 switch(avctx->colorspace) {
702 case AVCOL_SPC_BT709:
703 *matrix = kCVImageBufferYCbCrMatrix_ITU_R_709_2;
706 case AVCOL_SPC_UNSPECIFIED:
710 case AVCOL_SPC_BT470BG:
711 case AVCOL_SPC_SMPTE170M:
712 *matrix = kCVImageBufferYCbCrMatrix_ITU_R_601_4;
715 case AVCOL_SPC_SMPTE240M:
716 *matrix = kCVImageBufferYCbCrMatrix_SMPTE_240M_1995;
719 case AVCOL_SPC_BT2020_NCL:
720 *matrix = kCVImageBufferYCbCrMatrix_ITU_R_2020;
724 av_log(avctx, AV_LOG_ERROR, "Color space %s is not supported.\n", av_color_space_name(avctx->colorspace));
732 static av_cold int vtenc_init(AVCodecContext *avctx)
734 CFMutableDictionaryRef enc_info;
735 CFMutableDictionaryRef pixel_buffer_info;
736 CMVideoCodecType codec_type;
737 VTEncContext *vtctx = avctx->priv_data;
738 CFStringRef profile_level;
739 SInt32 bit_rate = avctx->bit_rate;
740 CFNumberRef bit_rate_num;
741 CFBooleanRef has_b_frames_cfbool;
742 CFNumberRef gamma_level;
745 codec_type = get_cm_codec_type(avctx->codec_id);
747 av_log(avctx, AV_LOG_ERROR, "Error: no mapping for AVCodecID %d\n", avctx->codec_id);
748 return AVERROR(EINVAL);
751 vtctx->has_b_frames = avctx->max_b_frames > 0;
752 if(vtctx->has_b_frames && vtctx->profile == H264_PROF_BASELINE){
753 av_log(avctx, AV_LOG_WARNING, "Cannot use B-frames with baseline profile. Output will not contain B-frames.\n");
754 vtctx->has_b_frames = false;
757 if (vtctx->entropy == VT_CABAC && vtctx->profile == H264_PROF_BASELINE) {
758 av_log(avctx, AV_LOG_WARNING, "CABAC entropy requires 'main' or 'high' profile, but baseline was requested. Encode will not use CABAC entropy.\n");
759 vtctx->entropy = VT_ENTROPY_NOT_SET;
762 if (!get_vt_profile_level(avctx, &profile_level)) return AVERROR(EINVAL);
764 vtctx->session = NULL;
766 enc_info = CFDictionaryCreateMutable(
769 &kCFCopyStringDictionaryKeyCallBacks,
770 &kCFTypeDictionaryValueCallBacks
773 if (!enc_info) return AVERROR(ENOMEM);
775 #if !TARGET_OS_IPHONE
776 if (!vtctx->allow_sw) {
777 CFDictionarySetValue(enc_info, kVTVideoEncoderSpecification_RequireHardwareAcceleratedVideoEncoder, kCFBooleanTrue);
779 CFDictionarySetValue(enc_info, kVTVideoEncoderSpecification_EnableHardwareAcceleratedVideoEncoder, kCFBooleanTrue);
783 if (avctx->pix_fmt != AV_PIX_FMT_VIDEOTOOLBOX) {
784 status = create_cv_pixel_buffer_info(avctx, &pixel_buffer_info);
790 pixel_buffer_info = NULL;
793 status = VTCompressionSessionCreate(
801 vtenc_output_callback,
806 if (pixel_buffer_info) CFRelease(pixel_buffer_info);
809 if (status || !vtctx->session) {
810 av_log(avctx, AV_LOG_ERROR, "Error: cannot create compression session: %d\n", status);
812 #if !TARGET_OS_IPHONE
813 if (!vtctx->allow_sw) {
814 av_log(avctx, AV_LOG_ERROR, "Try -allow_sw 1. The hardware encoder may be busy, or not supported.\n");
818 return AVERROR_EXTERNAL;
821 bit_rate_num = CFNumberCreate(kCFAllocatorDefault,
824 if (!bit_rate_num) return AVERROR(ENOMEM);
826 status = VTSessionSetProperty(vtctx->session,
827 kVTCompressionPropertyKey_AverageBitRate,
829 CFRelease(bit_rate_num);
832 av_log(avctx, AV_LOG_ERROR, "Error setting bitrate property: %d\n", status);
833 return AVERROR_EXTERNAL;
837 status = VTSessionSetProperty(vtctx->session,
838 kVTCompressionPropertyKey_ProfileLevel,
841 av_log(avctx, AV_LOG_ERROR, "Error setting profile/level property: %d\n", status);
842 return AVERROR_EXTERNAL;
846 if (avctx->gop_size > 0) {
847 CFNumberRef interval = CFNumberCreate(kCFAllocatorDefault,
851 return AVERROR(ENOMEM);
854 status = VTSessionSetProperty(vtctx->session,
855 kVTCompressionPropertyKey_MaxKeyFrameInterval,
860 av_log(avctx, AV_LOG_ERROR, "Error setting 'max key-frame interval' property: %d\n", status);
861 return AVERROR_EXTERNAL;
865 if (vtctx->frames_before) {
866 status = VTSessionSetProperty(vtctx->session,
867 kVTCompressionPropertyKey_MoreFramesBeforeStart,
870 if (status == kVTPropertyNotSupportedErr) {
871 av_log(avctx, AV_LOG_WARNING, "frames_before property is not supported on this device. Ignoring.\n");
873 av_log(avctx, AV_LOG_ERROR, "Error setting frames_before property: %d\n", status);
877 if (vtctx->frames_after) {
878 status = VTSessionSetProperty(vtctx->session,
879 kVTCompressionPropertyKey_MoreFramesAfterEnd,
882 if (status == kVTPropertyNotSupportedErr) {
883 av_log(avctx, AV_LOG_WARNING, "frames_after property is not supported on this device. Ignoring.\n");
885 av_log(avctx, AV_LOG_ERROR, "Error setting frames_after property: %d\n", status);
889 if (avctx->sample_aspect_ratio.num != 0) {
892 CFMutableDictionaryRef par;
893 AVRational *avpar = &avctx->sample_aspect_ratio;
895 av_reduce(&avpar->num, &avpar->den,
896 avpar->num, avpar->den,
899 num = CFNumberCreate(kCFAllocatorDefault,
903 den = CFNumberCreate(kCFAllocatorDefault,
909 par = CFDictionaryCreateMutable(kCFAllocatorDefault,
911 &kCFCopyStringDictionaryKeyCallBacks,
912 &kCFTypeDictionaryValueCallBacks);
914 if (!par || !num || !den) {
915 if (par) CFRelease(par);
916 if (num) CFRelease(num);
917 if (den) CFRelease(den);
919 return AVERROR(ENOMEM);
922 CFDictionarySetValue(
924 kCMFormatDescriptionKey_PixelAspectRatioHorizontalSpacing,
927 CFDictionarySetValue(
929 kCMFormatDescriptionKey_PixelAspectRatioVerticalSpacing,
932 status = VTSessionSetProperty(vtctx->session,
933 kVTCompressionPropertyKey_PixelAspectRatio,
943 "Error setting pixel aspect ratio to %d:%d: %d.\n",
944 avctx->sample_aspect_ratio.num,
945 avctx->sample_aspect_ratio.den,
948 return AVERROR_EXTERNAL;
952 status = get_cv_transfer_function(avctx, &vtctx->transfer_function, &gamma_level);
953 if (!status && vtctx->transfer_function) {
954 status = VTSessionSetProperty(vtctx->session,
955 kVTCompressionPropertyKey_TransferFunction,
956 vtctx->transfer_function);
959 av_log(avctx, AV_LOG_WARNING, "Could not set transfer function: %d\n", status);
963 status = get_cv_ycbcr_matrix(avctx, &vtctx->ycbcr_matrix);
964 if (!status && vtctx->ycbcr_matrix) {
965 status = VTSessionSetProperty(vtctx->session,
966 kVTCompressionPropertyKey_YCbCrMatrix,
967 vtctx->ycbcr_matrix);
970 av_log(avctx, AV_LOG_WARNING, "Could not set ycbcr matrix: %d\n", status);
974 status = get_cv_color_primaries(avctx, &vtctx->color_primaries);
975 if (!status && vtctx->color_primaries) {
976 status = VTSessionSetProperty(vtctx->session,
977 kVTCompressionPropertyKey_ColorPrimaries,
978 vtctx->color_primaries);
981 av_log(avctx, AV_LOG_WARNING, "Could not set color primaries: %d\n", status);
985 if (!status && gamma_level) {
986 status = VTSessionSetProperty(vtctx->session,
987 kCVImageBufferGammaLevelKey,
991 av_log(avctx, AV_LOG_WARNING, "Could not set gamma level: %d\n", status);
995 if (!vtctx->has_b_frames) {
996 status = VTSessionSetProperty(vtctx->session,
997 kVTCompressionPropertyKey_AllowFrameReordering,
1001 av_log(avctx, AV_LOG_ERROR, "Error setting 'allow frame reordering' property: %d\n", status);
1002 return AVERROR_EXTERNAL;
1006 if (vtctx->entropy != VT_ENTROPY_NOT_SET) {
1007 CFStringRef entropy = vtctx->entropy == VT_CABAC ?
1008 kVTH264EntropyMode_CABAC:
1009 kVTH264EntropyMode_CAVLC;
1011 status = VTSessionSetProperty(vtctx->session,
1012 kVTCompressionPropertyKey_H264EntropyMode,
1016 av_log(avctx, AV_LOG_ERROR, "Error setting entropy property: %d\n", status);
1017 return AVERROR_EXTERNAL;
1021 if (vtctx->realtime) {
1022 status = VTSessionSetProperty(vtctx->session,
1023 kVTCompressionPropertyKey_RealTime,
1027 av_log(avctx, AV_LOG_ERROR, "Error setting realtime property: %d\n", status);
1031 status = VTCompressionSessionPrepareToEncodeFrames(vtctx->session);
1033 av_log(avctx, AV_LOG_ERROR, "Error: cannot prepare encoder: %d\n", status);
1034 return AVERROR_EXTERNAL;
1037 pthread_mutex_init(&vtctx->lock, NULL);
1038 pthread_cond_init(&vtctx->cv_sample_sent, NULL);
1039 vtctx->dts_delta = vtctx->has_b_frames ? -1 : 0;
1041 status = VTSessionCopyProperty(vtctx->session,
1042 kVTCompressionPropertyKey_AllowFrameReordering,
1043 kCFAllocatorDefault,
1044 &has_b_frames_cfbool);
1047 //Some devices don't output B-frames for main profile, even if requested.
1048 vtctx->has_b_frames = CFBooleanGetValue(has_b_frames_cfbool);
1049 CFRelease(has_b_frames_cfbool);
1051 avctx->has_b_frames = vtctx->has_b_frames;
1056 static void vtenc_get_frame_info(CMSampleBufferRef buffer, bool *is_key_frame)
1058 CFArrayRef attachments;
1059 CFDictionaryRef attachment;
1060 CFBooleanRef not_sync;
1063 attachments = CMSampleBufferGetSampleAttachmentsArray(buffer, false);
1064 len = !attachments ? 0 : CFArrayGetCount(attachments);
1067 *is_key_frame = true;
1071 attachment = CFArrayGetValueAtIndex(attachments, 0);
1073 if (CFDictionaryGetValueIfPresent(attachment,
1074 kCMSampleAttachmentKey_NotSync,
1075 (const void **)¬_sync))
1077 *is_key_frame = !CFBooleanGetValue(not_sync);
1079 *is_key_frame = true;
1084 * Replaces length codes with H.264 Annex B start codes.
1085 * length_code_size must equal sizeof(start_code).
1086 * On failure, the contents of data may have been modified.
1088 * @param length_code_size Byte length of each length code
1089 * @param data Call with NAL units prefixed with length codes.
1090 * On success, the length codes are replace with
1092 * @param size Length of data, excluding any padding.
1093 * @return 0 on success
1094 * AVERROR_BUFFER_TOO_SMALL if length code size is smaller
1095 * than a start code or if a length_code in data specifies
1096 * data beyond the end of its buffer.
1098 static int replace_length_codes(size_t length_code_size,
1102 size_t remaining_size = size;
1104 if (length_code_size != sizeof(start_code)) {
1105 av_log(NULL, AV_LOG_ERROR, "Start code size and length code size not equal.\n");
1106 return AVERROR_BUFFER_TOO_SMALL;
1109 while (remaining_size > 0) {
1113 for (i = 0; i < length_code_size; i++) {
1118 if (remaining_size < box_len + sizeof(start_code)) {
1119 av_log(NULL, AV_LOG_ERROR, "Length is out of range.\n");
1120 AVERROR_BUFFER_TOO_SMALL;
1123 memcpy(data, start_code, sizeof(start_code));
1124 data += box_len + sizeof(start_code);
1125 remaining_size -= box_len + sizeof(start_code);
1132 * Copies NAL units and replaces length codes with
1133 * H.264 Annex B start codes. On failure, the contents of
1134 * dst_data may have been modified.
1136 * @param length_code_size Byte length of each length code
1137 * @param src_data NAL units prefixed with length codes.
1138 * @param src_size Length of buffer, excluding any padding.
1139 * @param dst_data Must be zeroed before calling this function.
1140 * Contains the copied NAL units prefixed with
1141 * start codes when the function returns
1143 * @param dst_size Length of dst_data
1144 * @return 0 on success
1145 * AVERROR_INVALIDDATA if length_code_size is invalid
1146 * AVERROR_BUFFER_TOO_SMALL if dst_data is too small
1147 * or if a length_code in src_data specifies data beyond
1148 * the end of its buffer.
1150 static int copy_replace_length_codes(
1151 size_t length_code_size,
1152 const uint8_t *src_data,
1157 size_t remaining_src_size = src_size;
1158 size_t remaining_dst_size = dst_size;
1160 if (length_code_size > 4) {
1161 return AVERROR_INVALIDDATA;
1164 while (remaining_src_size > 0) {
1165 size_t curr_src_len;
1166 size_t curr_dst_len;
1171 const uint8_t *src_box;
1173 for (i = 0; i < length_code_size; i++) {
1175 box_len |= src_data[i];
1178 curr_src_len = box_len + length_code_size;
1179 curr_dst_len = box_len + sizeof(start_code);
1181 if (remaining_src_size < curr_src_len) {
1182 return AVERROR_BUFFER_TOO_SMALL;
1185 if (remaining_dst_size < curr_dst_len) {
1186 return AVERROR_BUFFER_TOO_SMALL;
1189 dst_box = dst_data + sizeof(start_code);
1190 src_box = src_data + length_code_size;
1192 memcpy(dst_data, start_code, sizeof(start_code));
1193 memcpy(dst_box, src_box, box_len);
1195 src_data += curr_src_len;
1196 dst_data += curr_dst_len;
1198 remaining_src_size -= curr_src_len;
1199 remaining_dst_size -= curr_dst_len;
1205 static int vtenc_cm_to_avpacket(
1206 AVCodecContext *avctx,
1207 CMSampleBufferRef sample_buffer,
1210 VTEncContext *vtctx = avctx->priv_data;
1216 size_t length_code_size;
1217 size_t header_size = 0;
1220 int64_t time_base_num;
1224 CMBlockBufferRef block;
1225 CMVideoFormatDescriptionRef vid_fmt;
1228 vtenc_get_frame_info(sample_buffer, &is_key_frame);
1229 status = get_length_code_size(avctx, sample_buffer, &length_code_size);
1230 if (status) return status;
1232 add_header = is_key_frame && !(avctx->flags & AV_CODEC_FLAG_GLOBAL_HEADER);
1235 vid_fmt = CMSampleBufferGetFormatDescription(sample_buffer);
1237 av_log(avctx, AV_LOG_ERROR, "Cannot get format description.\n");
1240 int status = get_params_size(avctx, vid_fmt, &header_size);
1241 if (status) return status;
1244 block = CMSampleBufferGetDataBuffer(sample_buffer);
1246 av_log(avctx, AV_LOG_ERROR, "Could not get block buffer from sample buffer.\n");
1247 return AVERROR_EXTERNAL;
1251 status = CMBlockBufferGetDataPointer(block, 0, &in_buf_size, NULL, &buf_data);
1253 av_log(avctx, AV_LOG_ERROR, "Error: cannot get data pointer: %d\n", status);
1254 return AVERROR_EXTERNAL;
1257 size_t out_buf_size = header_size + in_buf_size;
1258 bool can_reuse_cmbuffer = !add_header &&
1260 length_code_size == sizeof(start_code);
1262 av_init_packet(pkt);
1264 if (can_reuse_cmbuffer) {
1265 AVBufferRef* buf_ref = av_buffer_create(
1273 if (!buf_ref) return AVERROR(ENOMEM);
1278 pkt->data = buf_data;
1279 pkt->size = in_buf_size;
1281 status = replace_length_codes(length_code_size, pkt->data, pkt->size);
1283 av_log(avctx, AV_LOG_ERROR, "Error replacing length codes: %d\n", status);
1288 status = av_new_packet(pkt, out_buf_size);
1289 if(status) return status;
1292 if (pkt->size < out_buf_size) {
1293 av_log(avctx, AV_LOG_ERROR, "Error: packet's buffer is too small.\n");
1294 return AVERROR_BUFFER_TOO_SMALL;
1298 status = copy_param_sets(avctx, vid_fmt, pkt->data, out_buf_size);
1299 if(status) return status;
1302 status = copy_replace_length_codes(
1306 pkt->data + header_size,
1307 pkt->size - header_size
1311 av_log(avctx, AV_LOG_ERROR, "Error copying packet data: %d", status);
1317 pkt->flags |= AV_PKT_FLAG_KEY;
1320 pts = CMSampleBufferGetPresentationTimeStamp(sample_buffer);
1321 dts = CMSampleBufferGetDecodeTimeStamp (sample_buffer);
1323 if (CMTIME_IS_INVALID(dts)) {
1324 if (!vtctx->has_b_frames) {
1327 av_log(avctx, AV_LOG_ERROR, "DTS is invalid.\n");
1328 return AVERROR_EXTERNAL;
1332 dts_delta = vtctx->dts_delta >= 0 ? vtctx->dts_delta : 0;
1333 time_base_num = avctx->time_base.num;
1334 pkt->pts = pts.value / time_base_num;
1335 pkt->dts = dts.value / time_base_num - dts_delta;
1341 * contiguous_buf_size is 0 if not contiguous, and the size of the buffer
1342 * containing all planes if so.
1344 static int get_cv_pixel_info(
1345 AVCodecContext *avctx,
1346 const AVFrame *frame,
1352 size_t *contiguous_buf_size)
1354 VTEncContext *vtctx = avctx->priv_data;
1355 int av_format = frame->format;
1356 int av_color_range = av_frame_get_color_range(frame);
1361 status = get_cv_pixel_format(avctx, av_format, av_color_range, color, &range_guessed);
1365 "Could not get pixel format for color format '%s' range '%s'.\n",
1366 av_get_pix_fmt_name(av_format),
1367 av_color_range > AVCOL_RANGE_UNSPECIFIED &&
1368 av_color_range < AVCOL_RANGE_NB ?
1369 av_color_range_name(av_color_range) :
1372 return AVERROR(EINVAL);
1375 if (range_guessed) {
1376 if (!vtctx->warned_color_range) {
1377 vtctx->warned_color_range = true;
1380 "Color range not set for %s. Using MPEG range.\n",
1381 av_get_pix_fmt_name(av_format));
1384 av_log(avctx, AV_LOG_WARNING, "");
1387 switch (av_format) {
1388 case AV_PIX_FMT_NV12:
1391 widths [0] = avctx->width;
1392 heights[0] = avctx->height;
1393 strides[0] = frame ? frame->linesize[0] : avctx->width;
1395 widths [1] = (avctx->width + 1) / 2;
1396 heights[1] = (avctx->height + 1) / 2;
1397 strides[1] = frame ? frame->linesize[1] : (avctx->width + 1) & -2;
1400 case AV_PIX_FMT_YUV420P:
1403 widths [0] = avctx->width;
1404 heights[0] = avctx->height;
1405 strides[0] = frame ? frame->linesize[0] : avctx->width;
1407 widths [1] = (avctx->width + 1) / 2;
1408 heights[1] = (avctx->height + 1) / 2;
1409 strides[1] = frame ? frame->linesize[1] : (avctx->width + 1) / 2;
1411 widths [2] = (avctx->width + 1) / 2;
1412 heights[2] = (avctx->height + 1) / 2;
1413 strides[2] = frame ? frame->linesize[2] : (avctx->width + 1) / 2;
1420 "Could not get frame format info for color %d range %d.\n",
1424 return AVERROR(EINVAL);
1427 *contiguous_buf_size = 0;
1428 for (i = 0; i < *plane_count; i++) {
1429 if (i < *plane_count - 1 &&
1430 frame->data[i] + strides[i] * heights[i] != frame->data[i + 1]) {
1431 *contiguous_buf_size = 0;
1435 *contiguous_buf_size += strides[i] * heights[i];
1441 #if !TARGET_OS_IPHONE
1442 //Not used on iOS - frame is always copied.
1443 static void free_avframe(
1448 const void *plane_addresses[])
1450 AVFrame *frame = release_ctx;
1451 av_frame_free(&frame);
1454 //Not used on OSX - frame is never copied.
1455 static int copy_avframe_to_pixel_buffer(AVCodecContext *avctx,
1456 const AVFrame *frame,
1457 CVPixelBufferRef cv_img,
1458 const size_t *plane_strides,
1459 const size_t *plane_rows)
1471 status = CVPixelBufferLockBaseAddress(cv_img, 0);
1476 "Error: Could not lock base address of CVPixelBuffer: %d.\n",
1481 if (CVPixelBufferIsPlanar(cv_img)) {
1482 plane_count = CVPixelBufferGetPlaneCount(cv_img);
1483 for (i = 0; frame->data[i]; i++) {
1484 if (i == plane_count) {
1485 CVPixelBufferUnlockBaseAddress(cv_img, 0);
1488 "Error: different number of planes in AVFrame and CVPixelBuffer.\n"
1491 return AVERROR_EXTERNAL;
1494 dst_addr = (uint8_t*)CVPixelBufferGetBaseAddressOfPlane(cv_img, i);
1495 src_addr = (uint8_t*)frame->data[i];
1496 dst_stride = CVPixelBufferGetBytesPerRowOfPlane(cv_img, i);
1497 src_stride = plane_strides[i];
1498 rows = plane_rows[i];
1500 if (dst_stride == src_stride) {
1501 memcpy(dst_addr, src_addr, src_stride * rows);
1503 copy_bytes = dst_stride < src_stride ? dst_stride : src_stride;
1505 for (j = 0; j < rows; j++) {
1506 memcpy(dst_addr + j * dst_stride, src_addr + j * src_stride, copy_bytes);
1511 if (frame->data[1]) {
1512 CVPixelBufferUnlockBaseAddress(cv_img, 0);
1515 "Error: different number of planes in AVFrame and non-planar CVPixelBuffer.\n"
1518 return AVERROR_EXTERNAL;
1521 dst_addr = (uint8_t*)CVPixelBufferGetBaseAddress(cv_img);
1522 src_addr = (uint8_t*)frame->data[0];
1523 dst_stride = CVPixelBufferGetBytesPerRow(cv_img);
1524 src_stride = plane_strides[0];
1525 rows = plane_rows[0];
1527 if (dst_stride == src_stride) {
1528 memcpy(dst_addr, src_addr, src_stride * rows);
1530 copy_bytes = dst_stride < src_stride ? dst_stride : src_stride;
1532 for (j = 0; j < rows; j++) {
1533 memcpy(dst_addr + j * dst_stride, src_addr + j * src_stride, copy_bytes);
1538 status = CVPixelBufferUnlockBaseAddress(cv_img, 0);
1540 av_log(avctx, AV_LOG_ERROR, "Error: Could not unlock CVPixelBuffer base address: %d.\n", status);
1541 return AVERROR_EXTERNAL;
1546 #endif //!TARGET_OS_IPHONE
1548 static int create_cv_pixel_buffer(AVCodecContext *avctx,
1549 const AVFrame *frame,
1550 CVPixelBufferRef *cv_img)
1554 size_t widths [AV_NUM_DATA_POINTERS];
1555 size_t heights[AV_NUM_DATA_POINTERS];
1556 size_t strides[AV_NUM_DATA_POINTERS];
1558 size_t contiguous_buf_size;
1559 #if TARGET_OS_IPHONE
1560 CVPixelBufferPoolRef pix_buf_pool;
1561 VTEncContext* vtctx = avctx->priv_data;
1563 CFMutableDictionaryRef pix_buf_attachments = CFDictionaryCreateMutable(
1564 kCFAllocatorDefault,
1566 &kCFCopyStringDictionaryKeyCallBacks,
1567 &kCFTypeDictionaryValueCallBacks);
1569 if (!pix_buf_attachments) return AVERROR(ENOMEM);
1572 if (avctx->pix_fmt == AV_PIX_FMT_VIDEOTOOLBOX) {
1573 av_assert0(frame->format == AV_PIX_FMT_VIDEOTOOLBOX);
1575 *cv_img = (CVPixelBufferRef)frame->data[3];
1576 av_assert0(*cv_img);
1582 memset(widths, 0, sizeof(widths));
1583 memset(heights, 0, sizeof(heights));
1584 memset(strides, 0, sizeof(strides));
1586 status = get_cv_pixel_info(
1594 &contiguous_buf_size
1601 "Error: Cannot convert format %d color_range %d: %d\n",
1603 av_frame_get_color_range(frame),
1607 return AVERROR_EXTERNAL;
1610 #if TARGET_OS_IPHONE
1611 pix_buf_pool = VTCompressionSessionGetPixelBufferPool(vtctx->session);
1612 if (!pix_buf_pool) {
1613 av_log(avctx, AV_LOG_ERROR, "Could not get pixel buffer pool.\n");
1614 return AVERROR_EXTERNAL;
1617 status = CVPixelBufferPoolCreatePixelBuffer(NULL,
1623 av_log(avctx, AV_LOG_ERROR, "Could not create pixel buffer from pool: %d.\n", status);
1624 return AVERROR_EXTERNAL;
1627 status = copy_avframe_to_pixel_buffer(avctx, frame, *cv_img, strides, heights);
1634 AVFrame *enc_frame = av_frame_alloc();
1635 if (!enc_frame) return AVERROR(ENOMEM);
1637 status = av_frame_ref(enc_frame, frame);
1639 av_frame_free(&enc_frame);
1643 status = CVPixelBufferCreateWithPlanarBytes(
1644 kCFAllocatorDefault,
1649 contiguous_buf_size,
1651 (void **)enc_frame->data,
1661 add_color_attr(avctx, pix_buf_attachments);
1662 CVBufferSetAttachments(*cv_img, pix_buf_attachments, kCVAttachmentMode_ShouldPropagate);
1663 CFRelease(pix_buf_attachments);
1666 av_log(avctx, AV_LOG_ERROR, "Error: Could not create CVPixelBuffer: %d\n", status);
1667 return AVERROR_EXTERNAL;
1674 static int create_encoder_dict_h264(const AVFrame *frame,
1675 CFDictionaryRef* dict_out)
1677 CFDictionaryRef dict = NULL;
1678 if (frame->pict_type == AV_PICTURE_TYPE_I) {
1679 const void *keys[] = { kVTEncodeFrameOptionKey_ForceKeyFrame };
1680 const void *vals[] = { kCFBooleanTrue };
1682 dict = CFDictionaryCreate(NULL, keys, vals, 1, NULL, NULL);
1683 if(!dict) return AVERROR(ENOMEM);
1690 static int vtenc_send_frame(AVCodecContext *avctx,
1691 VTEncContext *vtctx,
1692 const AVFrame *frame)
1695 CFDictionaryRef frame_dict;
1696 CVPixelBufferRef cv_img = NULL;
1697 int status = create_cv_pixel_buffer(avctx, frame, &cv_img);
1699 if (status) return status;
1701 status = create_encoder_dict_h264(frame, &frame_dict);
1707 time = CMTimeMake(frame->pts * avctx->time_base.num, avctx->time_base.den);
1708 status = VTCompressionSessionEncodeFrame(
1718 if (frame_dict) CFRelease(frame_dict);
1722 av_log(avctx, AV_LOG_ERROR, "Error: cannot encode frame: %d\n", status);
1723 return AVERROR_EXTERNAL;
1729 static av_cold int vtenc_frame(
1730 AVCodecContext *avctx,
1732 const AVFrame *frame,
1735 VTEncContext *vtctx = avctx->priv_data;
1738 CMSampleBufferRef buf = NULL;
1741 status = vtenc_send_frame(avctx, vtctx, frame);
1744 status = AVERROR_EXTERNAL;
1748 if (vtctx->frame_ct_in == 0) {
1749 vtctx->first_pts = frame->pts;
1750 } else if(vtctx->frame_ct_in == 1 && vtctx->has_b_frames) {
1751 vtctx->dts_delta = frame->pts - vtctx->first_pts;
1754 vtctx->frame_ct_in++;
1755 } else if(!vtctx->flushing) {
1756 vtctx->flushing = true;
1758 status = VTCompressionSessionCompleteFrames(vtctx->session,
1762 av_log(avctx, AV_LOG_ERROR, "Error flushing frames: %d\n", status);
1763 status = AVERROR_EXTERNAL;
1769 get_frame = vtctx->dts_delta >= 0 || !frame;
1775 status = vtenc_q_pop(vtctx, !frame, &buf);
1776 if (status) goto end_nopkt;
1777 if (!buf) goto end_nopkt;
1779 status = vtenc_cm_to_avpacket(avctx, buf, pkt);
1781 if (status) goto end_nopkt;
1787 av_packet_unref(pkt);
1791 static av_cold int vtenc_close(AVCodecContext *avctx)
1793 VTEncContext *vtctx = avctx->priv_data;
1795 if(!vtctx->session) return 0;
1797 pthread_cond_destroy(&vtctx->cv_sample_sent);
1798 pthread_mutex_destroy(&vtctx->lock);
1799 CFRelease(vtctx->session);
1800 vtctx->session = NULL;
1802 if (vtctx->color_primaries) {
1803 CFRelease(vtctx->color_primaries);
1804 vtctx->color_primaries = NULL;
1807 if (vtctx->transfer_function) {
1808 CFRelease(vtctx->transfer_function);
1809 vtctx->transfer_function = NULL;
1812 if (vtctx->ycbcr_matrix) {
1813 CFRelease(vtctx->ycbcr_matrix);
1814 vtctx->ycbcr_matrix = NULL;
1820 static const enum AVPixelFormat pix_fmts[] = {
1821 AV_PIX_FMT_VIDEOTOOLBOX,
1827 #define OFFSET(x) offsetof(VTEncContext, x)
1828 #define VE AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
1829 static const AVOption options[] = {
1830 { "profile", "Profile", OFFSET(profile), AV_OPT_TYPE_INT, { .i64 = H264_PROF_AUTO }, H264_PROF_AUTO, H264_PROF_COUNT, VE, "profile" },
1831 { "baseline", "Baseline Profile", 0, AV_OPT_TYPE_CONST, { .i64 = H264_PROF_BASELINE }, INT_MIN, INT_MAX, VE, "profile" },
1832 { "main", "Main Profile", 0, AV_OPT_TYPE_CONST, { .i64 = H264_PROF_MAIN }, INT_MIN, INT_MAX, VE, "profile" },
1833 { "high", "High Profile", 0, AV_OPT_TYPE_CONST, { .i64 = H264_PROF_HIGH }, INT_MIN, INT_MAX, VE, "profile" },
1835 { "level", "Level", OFFSET(level), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 52, VE, "level" },
1836 { "1.3", "Level 1.3, only available with Baseline Profile", 0, AV_OPT_TYPE_CONST, { .i64 = 13 }, INT_MIN, INT_MAX, VE, "level" },
1837 { "3.0", "Level 3.0", 0, AV_OPT_TYPE_CONST, { .i64 = 30 }, INT_MIN, INT_MAX, VE, "level" },
1838 { "3.1", "Level 3.1", 0, AV_OPT_TYPE_CONST, { .i64 = 31 }, INT_MIN, INT_MAX, VE, "level" },
1839 { "3.2", "Level 3.2", 0, AV_OPT_TYPE_CONST, { .i64 = 32 }, INT_MIN, INT_MAX, VE, "level" },
1840 { "4.0", "Level 4.0", 0, AV_OPT_TYPE_CONST, { .i64 = 40 }, INT_MIN, INT_MAX, VE, "level" },
1841 { "4.1", "Level 4.1", 0, AV_OPT_TYPE_CONST, { .i64 = 41 }, INT_MIN, INT_MAX, VE, "level" },
1842 { "4.2", "Level 4.2", 0, AV_OPT_TYPE_CONST, { .i64 = 42 }, INT_MIN, INT_MAX, VE, "level" },
1843 { "5.0", "Level 5.0", 0, AV_OPT_TYPE_CONST, { .i64 = 50 }, INT_MIN, INT_MAX, VE, "level" },
1844 { "5.1", "Level 5.1", 0, AV_OPT_TYPE_CONST, { .i64 = 51 }, INT_MIN, INT_MAX, VE, "level" },
1845 { "5.2", "Level 5.2", 0, AV_OPT_TYPE_CONST, { .i64 = 52 }, INT_MIN, INT_MAX, VE, "level" },
1847 { "allow_sw", "Allow software encoding", OFFSET(allow_sw), AV_OPT_TYPE_BOOL,
1848 { .i64 = 0 }, 0, 1, VE },
1850 { "coder", "Entropy coding", OFFSET(entropy), AV_OPT_TYPE_INT, { .i64 = VT_ENTROPY_NOT_SET }, VT_ENTROPY_NOT_SET, VT_CABAC, VE, "coder" },
1851 { "cavlc", "CAVLC entropy coding", 0, AV_OPT_TYPE_CONST, { .i64 = VT_CAVLC }, INT_MIN, INT_MAX, VE, "coder" },
1852 { "vlc", "CAVLC entropy coding", 0, AV_OPT_TYPE_CONST, { .i64 = VT_CAVLC }, INT_MIN, INT_MAX, VE, "coder" },
1853 { "cabac", "CABAC entropy coding", 0, AV_OPT_TYPE_CONST, { .i64 = VT_CABAC }, INT_MIN, INT_MAX, VE, "coder" },
1854 { "ac", "CABAC entropy coding", 0, AV_OPT_TYPE_CONST, { .i64 = VT_CABAC }, INT_MIN, INT_MAX, VE, "coder" },
1856 { "realtime", "Hint that encoding should happen in real-time if not faster (e.g. capturing from camera).",
1857 OFFSET(realtime), AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, VE },
1859 { "frames_before", "Other frames will come before the frames in this session. This helps smooth concatenation issues.",
1860 OFFSET(frames_before), AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, VE },
1861 { "frames_after", "Other frames will come after the frames in this session. This helps smooth concatenation issues.",
1862 OFFSET(frames_after), AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, VE },
1867 static const AVClass h264_videotoolbox_class = {
1868 .class_name = "h264_videotoolbox",
1869 .item_name = av_default_item_name,
1871 .version = LIBAVUTIL_VERSION_INT,
1874 AVCodec ff_h264_videotoolbox_encoder = {
1875 .name = "h264_videotoolbox",
1876 .long_name = NULL_IF_CONFIG_SMALL("VideoToolbox H.264 Encoder"),
1877 .type = AVMEDIA_TYPE_VIDEO,
1878 .id = AV_CODEC_ID_H264,
1879 .priv_data_size = sizeof(VTEncContext),
1880 .pix_fmts = pix_fmts,
1882 .encode2 = vtenc_frame,
1883 .close = vtenc_close,
1884 .capabilities = AV_CODEC_CAP_DELAY,
1885 .priv_class = &h264_videotoolbox_class,
1886 .caps_internal = FF_CODEC_CAP_INIT_THREADSAFE |
1887 FF_CODEC_CAP_INIT_CLEANUP,