2 * AVFoundation input device
3 * Copyright (c) 2014 Thilo Borgmann <thilo.borgmann@mail.de>
5 * This file is part of FFmpeg.
7 * FFmpeg is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU Lesser General Public
9 * License as published by the Free Software Foundation; either
10 * version 2.1 of the License, or (at your option) any later version.
12 * FFmpeg is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * Lesser General Public License for more details.
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with FFmpeg; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
24 * AVFoundation input device
25 * @author Thilo Borgmann <thilo.borgmann@mail.de>
28 #import <AVFoundation/AVFoundation.h>
31 #include "libavutil/pixdesc.h"
32 #include "libavutil/opt.h"
33 #include "libavutil/avstring.h"
34 #include "libavformat/internal.h"
35 #include "libavutil/internal.h"
36 #include "libavutil/parseutils.h"
37 #include "libavutil/time.h"
38 #include "libavutil/imgutils.h"
41 static const int avf_time_base = 1000000;
43 static const AVRational avf_time_base_q = {
48 struct AVFPixelFormatSpec {
49 enum AVPixelFormat ff_id;
53 static const struct AVFPixelFormatSpec avf_pixel_formats[] = {
54 { AV_PIX_FMT_MONOBLACK, kCVPixelFormatType_1Monochrome },
55 { AV_PIX_FMT_RGB555BE, kCVPixelFormatType_16BE555 },
56 { AV_PIX_FMT_RGB555LE, kCVPixelFormatType_16LE555 },
57 { AV_PIX_FMT_RGB565BE, kCVPixelFormatType_16BE565 },
58 { AV_PIX_FMT_RGB565LE, kCVPixelFormatType_16LE565 },
59 { AV_PIX_FMT_RGB24, kCVPixelFormatType_24RGB },
60 { AV_PIX_FMT_BGR24, kCVPixelFormatType_24BGR },
61 { AV_PIX_FMT_0RGB, kCVPixelFormatType_32ARGB },
62 { AV_PIX_FMT_BGR0, kCVPixelFormatType_32BGRA },
63 { AV_PIX_FMT_0BGR, kCVPixelFormatType_32ABGR },
64 { AV_PIX_FMT_RGB0, kCVPixelFormatType_32RGBA },
65 { AV_PIX_FMT_BGR48BE, kCVPixelFormatType_48RGB },
66 { AV_PIX_FMT_UYVY422, kCVPixelFormatType_422YpCbCr8 },
67 { AV_PIX_FMT_YUVA444P, kCVPixelFormatType_4444YpCbCrA8R },
68 { AV_PIX_FMT_YUVA444P16LE, kCVPixelFormatType_4444AYpCbCr16 },
69 { AV_PIX_FMT_YUV444P, kCVPixelFormatType_444YpCbCr8 },
70 { AV_PIX_FMT_YUV422P16, kCVPixelFormatType_422YpCbCr16 },
71 { AV_PIX_FMT_YUV422P10, kCVPixelFormatType_422YpCbCr10 },
72 { AV_PIX_FMT_YUV444P10, kCVPixelFormatType_444YpCbCr10 },
73 { AV_PIX_FMT_YUV420P, kCVPixelFormatType_420YpCbCr8Planar },
74 { AV_PIX_FMT_NV12, kCVPixelFormatType_420YpCbCr8BiPlanarVideoRange },
75 { AV_PIX_FMT_YUYV422, kCVPixelFormatType_422YpCbCr8_yuvs },
76 #if !TARGET_OS_IPHONE && __MAC_OS_X_VERSION_MIN_REQUIRED >= 1080
77 { AV_PIX_FMT_GRAY8, kCVPixelFormatType_OneComponent8 },
79 { AV_PIX_FMT_NONE, 0 }
87 int audio_frames_captured;
89 int64_t first_audio_pts;
90 pthread_mutex_t frame_lock;
92 id avf_audio_delegate;
98 int capture_mouse_clicks;
100 int drop_late_frames;
104 int video_device_index;
105 int video_stream_index;
106 int audio_device_index;
107 int audio_stream_index;
109 char *video_filename;
110 char *audio_filename;
112 int num_video_devices;
115 int audio_bits_per_sample;
118 int audio_signed_integer;
120 int audio_non_interleaved;
122 int32_t *audio_buffer;
123 int audio_buffer_size;
125 enum AVPixelFormat pixel_format;
127 AVCaptureSession *capture_session;
128 AVCaptureVideoDataOutput *video_output;
129 AVCaptureAudioDataOutput *audio_output;
130 CMSampleBufferRef current_frame;
131 CMSampleBufferRef current_audio_frame;
133 AVCaptureDevice *observed_device;
134 AVCaptureDeviceTransportControlsPlaybackMode observed_mode;
138 static void lock_frames(AVFContext* ctx)
140 pthread_mutex_lock(&ctx->frame_lock);
143 static void unlock_frames(AVFContext* ctx)
145 pthread_mutex_unlock(&ctx->frame_lock);
148 /** FrameReciever class - delegate for AVCaptureSession
150 @interface AVFFrameReceiver : NSObject
152 AVFContext* _context;
155 - (id)initWithContext:(AVFContext*)context;
157 - (void) captureOutput:(AVCaptureOutput *)captureOutput
158 didOutputSampleBuffer:(CMSampleBufferRef)videoFrame
159 fromConnection:(AVCaptureConnection *)connection;
163 @implementation AVFFrameReceiver
165 - (id)initWithContext:(AVFContext*)context
167 if (self = [super init]) {
170 // start observing if a device is set for it
171 if (_context->observed_device) {
172 NSString *keyPath = NSStringFromSelector(@selector(transportControlsPlaybackMode));
173 NSKeyValueObservingOptions options = NSKeyValueObservingOptionNew;
175 [_context->observed_device addObserver: self
185 // stop observing if a device is set for it
186 NSString *keyPath = NSStringFromSelector(@selector(transportControlsPlaybackMode));
187 [_context->observed_device removeObserver: self forKeyPath: keyPath];
191 - (void)observeValueForKeyPath:(NSString *)keyPath
193 change:(NSDictionary *)change
194 context:(void *)context {
195 if (context == _context) {
196 AVCaptureDeviceTransportControlsPlaybackMode mode =
197 [change[NSKeyValueChangeNewKey] integerValue];
199 if (mode != _context->observed_mode) {
200 if (mode == AVCaptureDeviceTransportControlsNotPlayingMode) {
201 _context->observed_quit = 1;
203 _context->observed_mode = mode;
206 [super observeValueForKeyPath: keyPath
213 - (void) captureOutput:(AVCaptureOutput *)captureOutput
214 didOutputSampleBuffer:(CMSampleBufferRef)videoFrame
215 fromConnection:(AVCaptureConnection *)connection
217 lock_frames(_context);
219 if (_context->current_frame != nil) {
220 CFRelease(_context->current_frame);
223 _context->current_frame = (CMSampleBufferRef)CFRetain(videoFrame);
225 unlock_frames(_context);
227 ++_context->frames_captured;
232 /** AudioReciever class - delegate for AVCaptureSession
234 @interface AVFAudioReceiver : NSObject
236 AVFContext* _context;
239 - (id)initWithContext:(AVFContext*)context;
241 - (void) captureOutput:(AVCaptureOutput *)captureOutput
242 didOutputSampleBuffer:(CMSampleBufferRef)audioFrame
243 fromConnection:(AVCaptureConnection *)connection;
247 @implementation AVFAudioReceiver
249 - (id)initWithContext:(AVFContext*)context
251 if (self = [super init]) {
257 - (void) captureOutput:(AVCaptureOutput *)captureOutput
258 didOutputSampleBuffer:(CMSampleBufferRef)audioFrame
259 fromConnection:(AVCaptureConnection *)connection
261 lock_frames(_context);
263 if (_context->current_audio_frame != nil) {
264 CFRelease(_context->current_audio_frame);
267 _context->current_audio_frame = (CMSampleBufferRef)CFRetain(audioFrame);
269 unlock_frames(_context);
271 ++_context->audio_frames_captured;
276 static void destroy_context(AVFContext* ctx)
278 [ctx->capture_session stopRunning];
280 [ctx->capture_session release];
281 [ctx->video_output release];
282 [ctx->audio_output release];
283 [ctx->avf_delegate release];
284 [ctx->avf_audio_delegate release];
286 ctx->capture_session = NULL;
287 ctx->video_output = NULL;
288 ctx->audio_output = NULL;
289 ctx->avf_delegate = NULL;
290 ctx->avf_audio_delegate = NULL;
292 av_freep(&ctx->audio_buffer);
294 pthread_mutex_destroy(&ctx->frame_lock);
296 if (ctx->current_frame) {
297 CFRelease(ctx->current_frame);
301 static void parse_device_name(AVFormatContext *s)
303 AVFContext *ctx = (AVFContext*)s->priv_data;
304 char *tmp = av_strdup(s->url);
308 ctx->video_filename = av_strtok(tmp, ":", &save);
309 ctx->audio_filename = av_strtok(NULL, ":", &save);
311 ctx->audio_filename = av_strtok(tmp, ":", &save);
316 * Configure the video device.
318 * Configure the video device using a run-time approach to access properties
319 * since formats, activeFormat are available since iOS >= 7.0 or OSX >= 10.7
320 * and activeVideoMaxFrameDuration is available since i0S >= 7.0 and OSX >= 10.9.
322 * The NSUndefinedKeyException must be handled by the caller of this function.
325 static int configure_video_device(AVFormatContext *s, AVCaptureDevice *video_device)
327 AVFContext *ctx = (AVFContext*)s->priv_data;
329 double framerate = av_q2d(ctx->framerate);
330 NSObject *range = nil;
331 NSObject *format = nil;
332 NSObject *selected_range = nil;
333 NSObject *selected_format = nil;
335 // try to configure format by formats list
336 // might raise an exception if no format list is given
337 // (then fallback to default, no configuration)
339 for (format in [video_device valueForKey:@"formats"]) {
340 CMFormatDescriptionRef formatDescription;
341 CMVideoDimensions dimensions;
343 formatDescription = (CMFormatDescriptionRef) [format performSelector:@selector(formatDescription)];
344 dimensions = CMVideoFormatDescriptionGetDimensions(formatDescription);
346 if ((ctx->width == 0 && ctx->height == 0) ||
347 (dimensions.width == ctx->width && dimensions.height == ctx->height)) {
349 selected_format = format;
351 for (range in [format valueForKey:@"videoSupportedFrameRateRanges"]) {
352 double max_framerate;
354 [[range valueForKey:@"maxFrameRate"] getValue:&max_framerate];
355 if (fabs (framerate - max_framerate) < 0.01) {
356 selected_range = range;
363 if (!selected_format) {
364 av_log(s, AV_LOG_ERROR, "Selected video size (%dx%d) is not supported by the device.\n",
365 ctx->width, ctx->height);
366 goto unsupported_format;
369 if (!selected_range) {
370 av_log(s, AV_LOG_ERROR, "Selected framerate (%f) is not supported by the device.\n",
372 if (ctx->video_is_muxed) {
373 av_log(s, AV_LOG_ERROR, "Falling back to default.\n");
375 goto unsupported_format;
379 if ([video_device lockForConfiguration:NULL] == YES) {
380 if (selected_format) {
381 [video_device setValue:selected_format forKey:@"activeFormat"];
383 if (selected_range) {
384 NSValue *min_frame_duration = [selected_range valueForKey:@"minFrameDuration"];
385 [video_device setValue:min_frame_duration forKey:@"activeVideoMinFrameDuration"];
386 [video_device setValue:min_frame_duration forKey:@"activeVideoMaxFrameDuration"];
389 av_log(s, AV_LOG_ERROR, "Could not lock device for configuration.\n");
390 return AVERROR(EINVAL);
392 } @catch(NSException *e) {
393 av_log(ctx, AV_LOG_WARNING, "Configuration of video device failed, falling back to default.\n");
400 av_log(s, AV_LOG_ERROR, "Supported modes:\n");
401 for (format in [video_device valueForKey:@"formats"]) {
402 CMFormatDescriptionRef formatDescription;
403 CMVideoDimensions dimensions;
405 formatDescription = (CMFormatDescriptionRef) [format performSelector:@selector(formatDescription)];
406 dimensions = CMVideoFormatDescriptionGetDimensions(formatDescription);
408 for (range in [format valueForKey:@"videoSupportedFrameRateRanges"]) {
409 double min_framerate;
410 double max_framerate;
412 [[range valueForKey:@"minFrameRate"] getValue:&min_framerate];
413 [[range valueForKey:@"maxFrameRate"] getValue:&max_framerate];
414 av_log(s, AV_LOG_ERROR, " %dx%d@[%f %f]fps\n",
415 dimensions.width, dimensions.height,
416 min_framerate, max_framerate);
419 return AVERROR(EINVAL);
422 static int add_video_device(AVFormatContext *s, AVCaptureDevice *video_device)
424 AVFContext *ctx = (AVFContext*)s->priv_data;
426 NSError *error = nil;
427 AVCaptureInput* capture_input = nil;
428 struct AVFPixelFormatSpec pxl_fmt_spec;
429 NSNumber *pixel_format;
430 NSDictionary *capture_dict;
431 dispatch_queue_t queue;
433 if (ctx->video_device_index < ctx->num_video_devices) {
434 capture_input = (AVCaptureInput*) [[[AVCaptureDeviceInput alloc] initWithDevice:video_device error:&error] autorelease];
436 capture_input = (AVCaptureInput*) video_device;
439 if (!capture_input) {
440 av_log(s, AV_LOG_ERROR, "Failed to create AV capture input device: %s\n",
441 [[error localizedDescription] UTF8String]);
445 if ([ctx->capture_session canAddInput:capture_input]) {
446 [ctx->capture_session addInput:capture_input];
448 av_log(s, AV_LOG_ERROR, "can't add video input to capture session\n");
453 ctx->video_output = [[AVCaptureVideoDataOutput alloc] init];
455 if (!ctx->video_output) {
456 av_log(s, AV_LOG_ERROR, "Failed to init AV video output\n");
460 // Configure device framerate and video size
462 if ((ret = configure_video_device(s, video_device)) < 0) {
465 } @catch (NSException *exception) {
466 if (![[exception name] isEqualToString:NSUndefinedKeyException]) {
467 av_log (s, AV_LOG_ERROR, "An error occurred: %s", [exception.reason UTF8String]);
468 return AVERROR_EXTERNAL;
472 // select pixel format
473 pxl_fmt_spec.ff_id = AV_PIX_FMT_NONE;
475 for (int i = 0; avf_pixel_formats[i].ff_id != AV_PIX_FMT_NONE; i++) {
476 if (ctx->pixel_format == avf_pixel_formats[i].ff_id) {
477 pxl_fmt_spec = avf_pixel_formats[i];
482 // check if selected pixel format is supported by AVFoundation
483 if (pxl_fmt_spec.ff_id == AV_PIX_FMT_NONE) {
484 av_log(s, AV_LOG_ERROR, "Selected pixel format (%s) is not supported by AVFoundation.\n",
485 av_get_pix_fmt_name(pxl_fmt_spec.ff_id));
489 // check if the pixel format is available for this device
490 if ([[ctx->video_output availableVideoCVPixelFormatTypes] indexOfObject:[NSNumber numberWithInt:pxl_fmt_spec.avf_id]] == NSNotFound) {
491 av_log(s, AV_LOG_ERROR, "Selected pixel format (%s) is not supported by the input device.\n",
492 av_get_pix_fmt_name(pxl_fmt_spec.ff_id));
494 pxl_fmt_spec.ff_id = AV_PIX_FMT_NONE;
496 av_log(s, AV_LOG_ERROR, "Supported pixel formats:\n");
497 for (NSNumber *pxl_fmt in [ctx->video_output availableVideoCVPixelFormatTypes]) {
498 struct AVFPixelFormatSpec pxl_fmt_dummy;
499 pxl_fmt_dummy.ff_id = AV_PIX_FMT_NONE;
500 for (int i = 0; avf_pixel_formats[i].ff_id != AV_PIX_FMT_NONE; i++) {
501 if ([pxl_fmt intValue] == avf_pixel_formats[i].avf_id) {
502 pxl_fmt_dummy = avf_pixel_formats[i];
507 if (pxl_fmt_dummy.ff_id != AV_PIX_FMT_NONE) {
508 av_log(s, AV_LOG_ERROR, " %s\n", av_get_pix_fmt_name(pxl_fmt_dummy.ff_id));
510 // select first supported pixel format instead of user selected (or default) pixel format
511 if (pxl_fmt_spec.ff_id == AV_PIX_FMT_NONE) {
512 pxl_fmt_spec = pxl_fmt_dummy;
517 // fail if there is no appropriate pixel format or print a warning about overriding the pixel format
518 if (pxl_fmt_spec.ff_id == AV_PIX_FMT_NONE) {
521 av_log(s, AV_LOG_WARNING, "Overriding selected pixel format to use %s instead.\n",
522 av_get_pix_fmt_name(pxl_fmt_spec.ff_id));
526 // set videoSettings to an empty dict for receiving raw data of muxed devices
527 if (ctx->capture_raw_data) {
528 ctx->pixel_format = pxl_fmt_spec.ff_id;
529 ctx->video_output.videoSettings = @{ };
531 ctx->pixel_format = pxl_fmt_spec.ff_id;
532 pixel_format = [NSNumber numberWithUnsignedInt:pxl_fmt_spec.avf_id];
533 capture_dict = [NSDictionary dictionaryWithObject:pixel_format
534 forKey:(id)kCVPixelBufferPixelFormatTypeKey];
536 [ctx->video_output setVideoSettings:capture_dict];
538 [ctx->video_output setAlwaysDiscardsLateVideoFrames:ctx->drop_late_frames];
540 // check for transport control support and set observer device if supported
541 int trans_ctrl = [video_device transportControlsSupported];
542 AVCaptureDeviceTransportControlsPlaybackMode trans_mode = [video_device transportControlsPlaybackMode];
545 ctx->observed_mode = trans_mode;
546 ctx->observed_device = video_device;
549 ctx->avf_delegate = [[AVFFrameReceiver alloc] initWithContext:ctx];
551 queue = dispatch_queue_create("avf_queue", NULL);
552 [ctx->video_output setSampleBufferDelegate:ctx->avf_delegate queue:queue];
553 dispatch_release(queue);
555 if ([ctx->capture_session canAddOutput:ctx->video_output]) {
556 [ctx->capture_session addOutput:ctx->video_output];
558 av_log(s, AV_LOG_ERROR, "can't add video output to capture session\n");
565 static int add_audio_device(AVFormatContext *s, AVCaptureDevice *audio_device)
567 AVFContext *ctx = (AVFContext*)s->priv_data;
568 NSError *error = nil;
569 AVCaptureDeviceInput* audio_dev_input = [[[AVCaptureDeviceInput alloc] initWithDevice:audio_device error:&error] autorelease];
570 dispatch_queue_t queue;
572 if (!audio_dev_input) {
573 av_log(s, AV_LOG_ERROR, "Failed to create AV capture input device: %s\n",
574 [[error localizedDescription] UTF8String]);
578 if ([ctx->capture_session canAddInput:audio_dev_input]) {
579 [ctx->capture_session addInput:audio_dev_input];
581 av_log(s, AV_LOG_ERROR, "can't add audio input to capture session\n");
586 ctx->audio_output = [[AVCaptureAudioDataOutput alloc] init];
588 if (!ctx->audio_output) {
589 av_log(s, AV_LOG_ERROR, "Failed to init AV audio output\n");
593 ctx->avf_audio_delegate = [[AVFAudioReceiver alloc] initWithContext:ctx];
595 queue = dispatch_queue_create("avf_audio_queue", NULL);
596 [ctx->audio_output setSampleBufferDelegate:ctx->avf_audio_delegate queue:queue];
597 dispatch_release(queue);
599 if ([ctx->capture_session canAddOutput:ctx->audio_output]) {
600 [ctx->capture_session addOutput:ctx->audio_output];
602 av_log(s, AV_LOG_ERROR, "adding audio output to capture session failed\n");
609 static int get_video_config(AVFormatContext *s)
611 AVFContext *ctx = (AVFContext*)s->priv_data;
612 CVImageBufferRef image_buffer;
613 CMBlockBufferRef block_buffer;
614 CGSize image_buffer_size;
615 AVStream* stream = avformat_new_stream(s, NULL);
621 // Take stream info from the first frame.
622 while (ctx->frames_captured < 1) {
623 CFRunLoopRunInMode(kCFRunLoopDefaultMode, 0.1, YES);
628 ctx->video_stream_index = stream->index;
630 avpriv_set_pts_info(stream, 64, 1, avf_time_base);
632 image_buffer = CMSampleBufferGetImageBuffer(ctx->current_frame);
633 block_buffer = CMSampleBufferGetDataBuffer(ctx->current_frame);
636 image_buffer_size = CVImageBufferGetEncodedSize(image_buffer);
638 stream->codecpar->codec_id = AV_CODEC_ID_RAWVIDEO;
639 stream->codecpar->codec_type = AVMEDIA_TYPE_VIDEO;
640 stream->codecpar->width = (int)image_buffer_size.width;
641 stream->codecpar->height = (int)image_buffer_size.height;
642 stream->codecpar->format = ctx->pixel_format;
644 stream->codecpar->codec_id = AV_CODEC_ID_DVVIDEO;
645 stream->codecpar->codec_type = AVMEDIA_TYPE_VIDEO;
646 stream->codecpar->format = ctx->pixel_format;
649 CFRelease(ctx->current_frame);
650 ctx->current_frame = nil;
657 static int get_audio_config(AVFormatContext *s)
659 AVFContext *ctx = (AVFContext*)s->priv_data;
660 CMFormatDescriptionRef format_desc;
661 AVStream* stream = avformat_new_stream(s, NULL);
667 // Take stream info from the first frame.
668 while (ctx->audio_frames_captured < 1) {
669 CFRunLoopRunInMode(kCFRunLoopDefaultMode, 0.1, YES);
674 ctx->audio_stream_index = stream->index;
676 avpriv_set_pts_info(stream, 64, 1, avf_time_base);
678 format_desc = CMSampleBufferGetFormatDescription(ctx->current_audio_frame);
679 const AudioStreamBasicDescription *basic_desc = CMAudioFormatDescriptionGetStreamBasicDescription(format_desc);
682 av_log(s, AV_LOG_ERROR, "audio format not available\n");
686 stream->codecpar->codec_type = AVMEDIA_TYPE_AUDIO;
687 stream->codecpar->sample_rate = basic_desc->mSampleRate;
688 stream->codecpar->channels = basic_desc->mChannelsPerFrame;
689 stream->codecpar->channel_layout = av_get_default_channel_layout(stream->codecpar->channels);
691 ctx->audio_channels = basic_desc->mChannelsPerFrame;
692 ctx->audio_bits_per_sample = basic_desc->mBitsPerChannel;
693 ctx->audio_float = basic_desc->mFormatFlags & kAudioFormatFlagIsFloat;
694 ctx->audio_be = basic_desc->mFormatFlags & kAudioFormatFlagIsBigEndian;
695 ctx->audio_signed_integer = basic_desc->mFormatFlags & kAudioFormatFlagIsSignedInteger;
696 ctx->audio_packed = basic_desc->mFormatFlags & kAudioFormatFlagIsPacked;
697 ctx->audio_non_interleaved = basic_desc->mFormatFlags & kAudioFormatFlagIsNonInterleaved;
699 if (basic_desc->mFormatID == kAudioFormatLinearPCM &&
701 ctx->audio_bits_per_sample == 32 &&
703 stream->codecpar->codec_id = ctx->audio_be ? AV_CODEC_ID_PCM_F32BE : AV_CODEC_ID_PCM_F32LE;
704 } else if (basic_desc->mFormatID == kAudioFormatLinearPCM &&
705 ctx->audio_signed_integer &&
706 ctx->audio_bits_per_sample == 16 &&
708 stream->codecpar->codec_id = ctx->audio_be ? AV_CODEC_ID_PCM_S16BE : AV_CODEC_ID_PCM_S16LE;
709 } else if (basic_desc->mFormatID == kAudioFormatLinearPCM &&
710 ctx->audio_signed_integer &&
711 ctx->audio_bits_per_sample == 24 &&
713 stream->codecpar->codec_id = ctx->audio_be ? AV_CODEC_ID_PCM_S24BE : AV_CODEC_ID_PCM_S24LE;
714 } else if (basic_desc->mFormatID == kAudioFormatLinearPCM &&
715 ctx->audio_signed_integer &&
716 ctx->audio_bits_per_sample == 32 &&
718 stream->codecpar->codec_id = ctx->audio_be ? AV_CODEC_ID_PCM_S32BE : AV_CODEC_ID_PCM_S32LE;
720 av_log(s, AV_LOG_ERROR, "audio format is not supported\n");
724 if (ctx->audio_non_interleaved) {
725 CMBlockBufferRef block_buffer = CMSampleBufferGetDataBuffer(ctx->current_audio_frame);
726 ctx->audio_buffer_size = CMBlockBufferGetDataLength(block_buffer);
727 ctx->audio_buffer = av_malloc(ctx->audio_buffer_size);
728 if (!ctx->audio_buffer) {
729 av_log(s, AV_LOG_ERROR, "error allocating audio buffer\n");
734 CFRelease(ctx->current_audio_frame);
735 ctx->current_audio_frame = nil;
742 static int avf_read_header(AVFormatContext *s)
744 NSAutoreleasePool *pool = [[NSAutoreleasePool alloc] init];
745 int capture_screen = 0;
746 uint32_t num_screens = 0;
747 AVFContext *ctx = (AVFContext*)s->priv_data;
748 AVCaptureDevice *video_device = nil;
749 AVCaptureDevice *audio_device = nil;
750 // Find capture device
751 NSArray *devices = [AVCaptureDevice devicesWithMediaType:AVMediaTypeVideo];
752 NSArray *devices_muxed = [AVCaptureDevice devicesWithMediaType:AVMediaTypeMuxed];
754 ctx->num_video_devices = [devices count] + [devices_muxed count];
755 ctx->first_pts = av_gettime();
756 ctx->first_audio_pts = av_gettime();
758 pthread_mutex_init(&ctx->frame_lock, NULL);
760 #if !TARGET_OS_IPHONE && __MAC_OS_X_VERSION_MIN_REQUIRED >= 1070
761 CGGetActiveDisplayList(0, NULL, &num_screens);
764 // List devices if requested
765 if (ctx->list_devices) {
767 av_log(ctx, AV_LOG_INFO, "AVFoundation video devices:\n");
768 for (AVCaptureDevice *device in devices) {
769 const char *name = [[device localizedName] UTF8String];
770 index = [devices indexOfObject:device];
771 av_log(ctx, AV_LOG_INFO, "[%d] %s\n", index, name);
773 for (AVCaptureDevice *device in devices_muxed) {
774 const char *name = [[device localizedName] UTF8String];
775 index = [devices count] + [devices_muxed indexOfObject:device];
776 av_log(ctx, AV_LOG_INFO, "[%d] %s\n", index, name);
778 #if !TARGET_OS_IPHONE && __MAC_OS_X_VERSION_MIN_REQUIRED >= 1070
779 if (num_screens > 0) {
780 CGDirectDisplayID screens[num_screens];
781 CGGetActiveDisplayList(num_screens, screens, &num_screens);
782 for (int i = 0; i < num_screens; i++) {
783 av_log(ctx, AV_LOG_INFO, "[%d] Capture screen %d\n", ctx->num_video_devices + i, i);
788 av_log(ctx, AV_LOG_INFO, "AVFoundation audio devices:\n");
789 devices = [AVCaptureDevice devicesWithMediaType:AVMediaTypeAudio];
790 for (AVCaptureDevice *device in devices) {
791 const char *name = [[device localizedName] UTF8String];
792 int index = [devices indexOfObject:device];
793 av_log(ctx, AV_LOG_INFO, "[%d] %s\n", index, name);
798 // parse input filename for video and audio device
799 parse_device_name(s);
801 // check for device index given in filename
802 if (ctx->video_device_index == -1 && ctx->video_filename) {
803 sscanf(ctx->video_filename, "%d", &ctx->video_device_index);
805 if (ctx->audio_device_index == -1 && ctx->audio_filename) {
806 sscanf(ctx->audio_filename, "%d", &ctx->audio_device_index);
809 if (ctx->video_device_index >= 0) {
810 if (ctx->video_device_index < ctx->num_video_devices) {
811 if (ctx->video_device_index < [devices count]) {
812 video_device = [devices objectAtIndex:ctx->video_device_index];
814 video_device = [devices_muxed objectAtIndex:(ctx->video_device_index - [devices count])];
815 ctx->video_is_muxed = 1;
817 } else if (ctx->video_device_index < ctx->num_video_devices + num_screens) {
818 #if !TARGET_OS_IPHONE && __MAC_OS_X_VERSION_MIN_REQUIRED >= 1070
819 CGDirectDisplayID screens[num_screens];
820 CGGetActiveDisplayList(num_screens, screens, &num_screens);
821 AVCaptureScreenInput* capture_screen_input = [[[AVCaptureScreenInput alloc] initWithDisplayID:screens[ctx->video_device_index - ctx->num_video_devices]] autorelease];
823 if (ctx->framerate.num > 0) {
824 capture_screen_input.minFrameDuration = CMTimeMake(ctx->framerate.den, ctx->framerate.num);
827 #if !TARGET_OS_IPHONE && __MAC_OS_X_VERSION_MIN_REQUIRED >= 1080
828 if (ctx->capture_cursor) {
829 capture_screen_input.capturesCursor = YES;
831 capture_screen_input.capturesCursor = NO;
835 if (ctx->capture_mouse_clicks) {
836 capture_screen_input.capturesMouseClicks = YES;
838 capture_screen_input.capturesMouseClicks = NO;
841 video_device = (AVCaptureDevice*) capture_screen_input;
845 av_log(ctx, AV_LOG_ERROR, "Invalid device index\n");
848 } else if (ctx->video_filename &&
849 strncmp(ctx->video_filename, "none", 4)) {
850 if (!strncmp(ctx->video_filename, "default", 7)) {
851 video_device = [AVCaptureDevice defaultDeviceWithMediaType:AVMediaTypeVideo];
853 // looking for video inputs
854 for (AVCaptureDevice *device in devices) {
855 if (!strncmp(ctx->video_filename, [[device localizedName] UTF8String], strlen(ctx->video_filename))) {
856 video_device = device;
860 // looking for muxed inputs
861 for (AVCaptureDevice *device in devices_muxed) {
862 if (!strncmp(ctx->video_filename, [[device localizedName] UTF8String], strlen(ctx->video_filename))) {
863 video_device = device;
864 ctx->video_is_muxed = 1;
869 #if !TARGET_OS_IPHONE && __MAC_OS_X_VERSION_MIN_REQUIRED >= 1070
870 // looking for screen inputs
873 if(sscanf(ctx->video_filename, "Capture screen %d", &idx) && idx < num_screens) {
874 CGDirectDisplayID screens[num_screens];
875 CGGetActiveDisplayList(num_screens, screens, &num_screens);
876 AVCaptureScreenInput* capture_screen_input = [[[AVCaptureScreenInput alloc] initWithDisplayID:screens[idx]] autorelease];
877 video_device = (AVCaptureDevice*) capture_screen_input;
878 ctx->video_device_index = ctx->num_video_devices + idx;
881 if (ctx->framerate.num > 0) {
882 capture_screen_input.minFrameDuration = CMTimeMake(ctx->framerate.den, ctx->framerate.num);
885 #if !TARGET_OS_IPHONE && __MAC_OS_X_VERSION_MIN_REQUIRED >= 1080
886 if (ctx->capture_cursor) {
887 capture_screen_input.capturesCursor = YES;
889 capture_screen_input.capturesCursor = NO;
893 if (ctx->capture_mouse_clicks) {
894 capture_screen_input.capturesMouseClicks = YES;
896 capture_screen_input.capturesMouseClicks = NO;
904 av_log(ctx, AV_LOG_ERROR, "Video device not found\n");
910 if (ctx->audio_device_index >= 0) {
911 NSArray *devices = [AVCaptureDevice devicesWithMediaType:AVMediaTypeAudio];
913 if (ctx->audio_device_index >= [devices count]) {
914 av_log(ctx, AV_LOG_ERROR, "Invalid audio device index\n");
918 audio_device = [devices objectAtIndex:ctx->audio_device_index];
919 } else if (ctx->audio_filename &&
920 strncmp(ctx->audio_filename, "none", 4)) {
921 if (!strncmp(ctx->audio_filename, "default", 7)) {
922 audio_device = [AVCaptureDevice defaultDeviceWithMediaType:AVMediaTypeAudio];
924 NSArray *devices = [AVCaptureDevice devicesWithMediaType:AVMediaTypeAudio];
926 for (AVCaptureDevice *device in devices) {
927 if (!strncmp(ctx->audio_filename, [[device localizedName] UTF8String], strlen(ctx->audio_filename))) {
928 audio_device = device;
935 av_log(ctx, AV_LOG_ERROR, "Audio device not found\n");
940 // Video nor Audio capture device not found, looking for AVMediaTypeVideo/Audio
941 if (!video_device && !audio_device) {
942 av_log(s, AV_LOG_ERROR, "No AV capture device found\n");
947 if (ctx->video_device_index < ctx->num_video_devices) {
948 av_log(s, AV_LOG_DEBUG, "'%s' opened\n", [[video_device localizedName] UTF8String]);
950 av_log(s, AV_LOG_DEBUG, "'%s' opened\n", [[video_device description] UTF8String]);
954 av_log(s, AV_LOG_DEBUG, "audio device '%s' opened\n", [[audio_device localizedName] UTF8String]);
957 // Initialize capture session
958 ctx->capture_session = [[AVCaptureSession alloc] init];
960 if (video_device && add_video_device(s, video_device)) {
963 if (audio_device && add_audio_device(s, audio_device)) {
966 [ctx->capture_session startRunning];
968 /* Unlock device configuration only after the session is started so it
969 * does not reset the capture formats */
970 if (!capture_screen) {
971 [video_device unlockForConfiguration];
974 if (video_device && get_video_config(s)) {
979 if (audio_device && get_audio_config(s)) {
988 destroy_context(ctx);
992 static int copy_cvpixelbuffer(AVFormatContext *s,
993 CVPixelBufferRef image_buffer,
996 AVFContext *ctx = s->priv_data;
998 const uint8_t *src_data[4];
999 int width = CVPixelBufferGetWidth(image_buffer);
1000 int height = CVPixelBufferGetHeight(image_buffer);
1003 memset(src_linesize, 0, sizeof(src_linesize));
1004 memset(src_data, 0, sizeof(src_data));
1006 status = CVPixelBufferLockBaseAddress(image_buffer, 0);
1007 if (status != kCVReturnSuccess) {
1008 av_log(s, AV_LOG_ERROR, "Could not lock base address: %d (%dx%d)\n", status, width, height);
1009 return AVERROR_EXTERNAL;
1012 if (CVPixelBufferIsPlanar(image_buffer)) {
1013 size_t plane_count = CVPixelBufferGetPlaneCount(image_buffer);
1015 for(i = 0; i < plane_count; i++){
1016 src_linesize[i] = CVPixelBufferGetBytesPerRowOfPlane(image_buffer, i);
1017 src_data[i] = CVPixelBufferGetBaseAddressOfPlane(image_buffer, i);
1020 src_linesize[0] = CVPixelBufferGetBytesPerRow(image_buffer);
1021 src_data[0] = CVPixelBufferGetBaseAddress(image_buffer);
1024 status = av_image_copy_to_buffer(pkt->data, pkt->size,
1025 src_data, src_linesize,
1026 ctx->pixel_format, width, height, 1);
1030 CVPixelBufferUnlockBaseAddress(image_buffer, 0);
1035 static int avf_read_packet(AVFormatContext *s, AVPacket *pkt)
1037 AVFContext* ctx = (AVFContext*)s->priv_data;
1040 CVImageBufferRef image_buffer;
1041 CMBlockBufferRef block_buffer;
1044 if (ctx->current_frame != nil) {
1048 image_buffer = CMSampleBufferGetImageBuffer(ctx->current_frame);
1049 block_buffer = CMSampleBufferGetDataBuffer(ctx->current_frame);
1051 if (image_buffer != nil) {
1052 length = (int)CVPixelBufferGetDataSize(image_buffer);
1053 } else if (block_buffer != nil) {
1054 length = (int)CMBlockBufferGetDataLength(block_buffer);
1056 return AVERROR(EINVAL);
1059 if (av_new_packet(pkt, length) < 0) {
1060 return AVERROR(EIO);
1064 CMSampleTimingInfo timing_info;
1066 if (CMSampleBufferGetOutputSampleTimingInfoArray(ctx->current_frame, 1, &timing_info, &count) == noErr) {
1067 AVRational timebase_q = av_make_q(1, timing_info.presentationTimeStamp.timescale);
1068 pkt->pts = pkt->dts = av_rescale_q(timing_info.presentationTimeStamp.value, timebase_q, avf_time_base_q);
1071 pkt->stream_index = ctx->video_stream_index;
1072 pkt->flags |= AV_PKT_FLAG_KEY;
1075 status = copy_cvpixelbuffer(s, image_buffer, pkt);
1078 OSStatus ret = CMBlockBufferCopyDataBytes(block_buffer, 0, pkt->size, pkt->data);
1079 if (ret != kCMBlockBufferNoErr) {
1080 status = AVERROR(EIO);
1083 CFRelease(ctx->current_frame);
1084 ctx->current_frame = nil;
1088 } else if (ctx->current_audio_frame != nil) {
1089 CMBlockBufferRef block_buffer = CMSampleBufferGetDataBuffer(ctx->current_audio_frame);
1090 int block_buffer_size = CMBlockBufferGetDataLength(block_buffer);
1092 if (!block_buffer || !block_buffer_size) {
1093 return AVERROR(EIO);
1096 if (ctx->audio_non_interleaved && block_buffer_size > ctx->audio_buffer_size) {
1097 return AVERROR_BUFFER_TOO_SMALL;
1100 if (av_new_packet(pkt, block_buffer_size) < 0) {
1101 return AVERROR(EIO);
1105 CMSampleTimingInfo timing_info;
1107 if (CMSampleBufferGetOutputSampleTimingInfoArray(ctx->current_audio_frame, 1, &timing_info, &count) == noErr) {
1108 AVRational timebase_q = av_make_q(1, timing_info.presentationTimeStamp.timescale);
1109 pkt->pts = pkt->dts = av_rescale_q(timing_info.presentationTimeStamp.value, timebase_q, avf_time_base_q);
1112 pkt->stream_index = ctx->audio_stream_index;
1113 pkt->flags |= AV_PKT_FLAG_KEY;
1115 if (ctx->audio_non_interleaved) {
1116 int sample, c, shift, num_samples;
1118 OSStatus ret = CMBlockBufferCopyDataBytes(block_buffer, 0, pkt->size, ctx->audio_buffer);
1119 if (ret != kCMBlockBufferNoErr) {
1120 return AVERROR(EIO);
1123 num_samples = pkt->size / (ctx->audio_channels * (ctx->audio_bits_per_sample >> 3));
1125 // transform decoded frame into output format
1126 #define INTERLEAVE_OUTPUT(bps) \
1128 int##bps##_t **src; \
1129 int##bps##_t *dest; \
1130 src = av_malloc(ctx->audio_channels * sizeof(int##bps##_t*)); \
1131 if (!src) return AVERROR(EIO); \
1132 for (c = 0; c < ctx->audio_channels; c++) { \
1133 src[c] = ((int##bps##_t*)ctx->audio_buffer) + c * num_samples; \
1135 dest = (int##bps##_t*)pkt->data; \
1136 shift = bps - ctx->audio_bits_per_sample; \
1137 for (sample = 0; sample < num_samples; sample++) \
1138 for (c = 0; c < ctx->audio_channels; c++) \
1139 *dest++ = src[c][sample] << shift; \
1143 if (ctx->audio_bits_per_sample <= 16) {
1144 INTERLEAVE_OUTPUT(16)
1146 INTERLEAVE_OUTPUT(32)
1149 OSStatus ret = CMBlockBufferCopyDataBytes(block_buffer, 0, pkt->size, pkt->data);
1150 if (ret != kCMBlockBufferNoErr) {
1151 return AVERROR(EIO);
1155 CFRelease(ctx->current_audio_frame);
1156 ctx->current_audio_frame = nil;
1160 if (ctx->observed_quit) {
1163 return AVERROR(EAGAIN);
1168 } while (!pkt->data);
1173 static int avf_close(AVFormatContext *s)
1175 AVFContext* ctx = (AVFContext*)s->priv_data;
1176 destroy_context(ctx);
1180 static const AVOption options[] = {
1181 { "list_devices", "list available devices", offsetof(AVFContext, list_devices), AV_OPT_TYPE_BOOL, {.i64=0}, 0, 1, AV_OPT_FLAG_DECODING_PARAM },
1182 { "video_device_index", "select video device by index for devices with same name (starts at 0)", offsetof(AVFContext, video_device_index), AV_OPT_TYPE_INT, {.i64 = -1}, -1, INT_MAX, AV_OPT_FLAG_DECODING_PARAM },
1183 { "audio_device_index", "select audio device by index for devices with same name (starts at 0)", offsetof(AVFContext, audio_device_index), AV_OPT_TYPE_INT, {.i64 = -1}, -1, INT_MAX, AV_OPT_FLAG_DECODING_PARAM },
1184 { "pixel_format", "set pixel format", offsetof(AVFContext, pixel_format), AV_OPT_TYPE_PIXEL_FMT, {.i64 = AV_PIX_FMT_YUV420P}, 0, INT_MAX, AV_OPT_FLAG_DECODING_PARAM},
1185 { "framerate", "set frame rate", offsetof(AVFContext, framerate), AV_OPT_TYPE_VIDEO_RATE, {.str = "ntsc"}, 0, INT_MAX, AV_OPT_FLAG_DECODING_PARAM },
1186 { "video_size", "set video size", offsetof(AVFContext, width), AV_OPT_TYPE_IMAGE_SIZE, {.str = NULL}, 0, 0, AV_OPT_FLAG_DECODING_PARAM },
1187 { "capture_cursor", "capture the screen cursor", offsetof(AVFContext, capture_cursor), AV_OPT_TYPE_BOOL, {.i64=0}, 0, 1, AV_OPT_FLAG_DECODING_PARAM },
1188 { "capture_mouse_clicks", "capture the screen mouse clicks", offsetof(AVFContext, capture_mouse_clicks), AV_OPT_TYPE_BOOL, {.i64=0}, 0, 1, AV_OPT_FLAG_DECODING_PARAM },
1189 { "capture_raw_data", "capture the raw data from device connection", offsetof(AVFContext, capture_raw_data), AV_OPT_TYPE_BOOL, {.i64=0}, 0, 1, AV_OPT_FLAG_DECODING_PARAM },
1190 { "drop_late_frames", "drop frames that are available later than expected", offsetof(AVFContext, drop_late_frames), AV_OPT_TYPE_BOOL, {.i64=1}, 0, 1, AV_OPT_FLAG_DECODING_PARAM },
1195 static const AVClass avf_class = {
1196 .class_name = "AVFoundation indev",
1197 .item_name = av_default_item_name,
1199 .version = LIBAVUTIL_VERSION_INT,
1200 .category = AV_CLASS_CATEGORY_DEVICE_VIDEO_INPUT,
1203 AVInputFormat ff_avfoundation_demuxer = {
1204 .name = "avfoundation",
1205 .long_name = NULL_IF_CONFIG_SMALL("AVFoundation input device"),
1206 .priv_data_size = sizeof(AVFContext),
1207 .read_header = avf_read_header,
1208 .read_packet = avf_read_packet,
1209 .read_close = avf_close,
1210 .flags = AVFMT_NOFILE,
1211 .priv_class = &avf_class,