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/channel_layout.h"
32 #include "libavutil/pixdesc.h"
33 #include "libavutil/opt.h"
34 #include "libavutil/avstring.h"
35 #include "libavformat/internal.h"
36 #include "libavutil/internal.h"
37 #include "libavutil/parseutils.h"
38 #include "libavutil/time.h"
39 #include "libavutil/imgutils.h"
42 static const int avf_time_base = 1000000;
44 static const AVRational avf_time_base_q = {
49 struct AVFPixelFormatSpec {
50 enum AVPixelFormat ff_id;
54 static const struct AVFPixelFormatSpec avf_pixel_formats[] = {
55 { AV_PIX_FMT_MONOBLACK, kCVPixelFormatType_1Monochrome },
56 { AV_PIX_FMT_RGB555BE, kCVPixelFormatType_16BE555 },
57 { AV_PIX_FMT_RGB555LE, kCVPixelFormatType_16LE555 },
58 { AV_PIX_FMT_RGB565BE, kCVPixelFormatType_16BE565 },
59 { AV_PIX_FMT_RGB565LE, kCVPixelFormatType_16LE565 },
60 { AV_PIX_FMT_RGB24, kCVPixelFormatType_24RGB },
61 { AV_PIX_FMT_BGR24, kCVPixelFormatType_24BGR },
62 { AV_PIX_FMT_0RGB, kCVPixelFormatType_32ARGB },
63 { AV_PIX_FMT_BGR0, kCVPixelFormatType_32BGRA },
64 { AV_PIX_FMT_0BGR, kCVPixelFormatType_32ABGR },
65 { AV_PIX_FMT_RGB0, kCVPixelFormatType_32RGBA },
66 { AV_PIX_FMT_BGR48BE, kCVPixelFormatType_48RGB },
67 { AV_PIX_FMT_UYVY422, kCVPixelFormatType_422YpCbCr8 },
68 { AV_PIX_FMT_YUVA444P, kCVPixelFormatType_4444YpCbCrA8R },
69 { AV_PIX_FMT_YUVA444P16LE, kCVPixelFormatType_4444AYpCbCr16 },
70 { AV_PIX_FMT_YUV444P, kCVPixelFormatType_444YpCbCr8 },
71 { AV_PIX_FMT_YUV422P16, kCVPixelFormatType_422YpCbCr16 },
72 { AV_PIX_FMT_YUV422P10, kCVPixelFormatType_422YpCbCr10 },
73 { AV_PIX_FMT_YUV444P10, kCVPixelFormatType_444YpCbCr10 },
74 { AV_PIX_FMT_YUV420P, kCVPixelFormatType_420YpCbCr8Planar },
75 { AV_PIX_FMT_NV12, kCVPixelFormatType_420YpCbCr8BiPlanarVideoRange },
76 { AV_PIX_FMT_YUYV422, kCVPixelFormatType_422YpCbCr8_yuvs },
77 #if !TARGET_OS_IPHONE && __MAC_OS_X_VERSION_MIN_REQUIRED >= 1080
78 { AV_PIX_FMT_GRAY8, kCVPixelFormatType_OneComponent8 },
80 { AV_PIX_FMT_NONE, 0 }
88 int audio_frames_captured;
89 pthread_mutex_t frame_lock;
91 id avf_audio_delegate;
97 int capture_mouse_clicks;
104 int video_device_index;
105 int video_stream_index;
106 int audio_device_index;
107 int audio_stream_index;
110 char *video_filename;
111 char *audio_filename;
113 int num_video_devices;
116 int audio_bits_per_sample;
119 int audio_signed_integer;
121 int audio_non_interleaved;
123 int32_t *audio_buffer;
124 int audio_buffer_size;
126 enum AVPixelFormat pixel_format;
128 AVCaptureSession *capture_session;
129 AVCaptureVideoDataOutput *video_output;
130 AVCaptureAudioDataOutput *audio_output;
131 CMSampleBufferRef current_frame;
132 CMSampleBufferRef current_audio_frame;
134 AVCaptureDevice *observed_device;
135 #if !TARGET_OS_IPHONE && __MAC_OS_X_VERSION_MIN_REQUIRED >= 1070
136 AVCaptureDeviceTransportControlsPlaybackMode observed_mode;
141 static void lock_frames(AVFContext* ctx)
143 pthread_mutex_lock(&ctx->frame_lock);
146 static void unlock_frames(AVFContext* ctx)
148 pthread_mutex_unlock(&ctx->frame_lock);
151 /** FrameReciever class - delegate for AVCaptureSession
153 @interface AVFFrameReceiver : NSObject
155 AVFContext* _context;
158 - (id)initWithContext:(AVFContext*)context;
160 - (void) captureOutput:(AVCaptureOutput *)captureOutput
161 didOutputSampleBuffer:(CMSampleBufferRef)videoFrame
162 fromConnection:(AVCaptureConnection *)connection;
166 @implementation AVFFrameReceiver
168 - (id)initWithContext:(AVFContext*)context
170 if (self = [super init]) {
173 // start observing if a device is set for it
174 #if !TARGET_OS_IPHONE && __MAC_OS_X_VERSION_MIN_REQUIRED >= 1070
175 if (_context->observed_device) {
176 NSString *keyPath = NSStringFromSelector(@selector(transportControlsPlaybackMode));
177 NSKeyValueObservingOptions options = NSKeyValueObservingOptionNew;
179 [_context->observed_device addObserver: self
190 // stop observing if a device is set for it
191 #if !TARGET_OS_IPHONE && __MAC_OS_X_VERSION_MIN_REQUIRED >= 1070
192 if (_context->observed_device) {
193 NSString *keyPath = NSStringFromSelector(@selector(transportControlsPlaybackMode));
194 [_context->observed_device removeObserver: self forKeyPath: keyPath];
200 - (void)observeValueForKeyPath:(NSString *)keyPath
202 change:(NSDictionary *)change
203 context:(void *)context {
204 if (context == _context) {
205 #if !TARGET_OS_IPHONE && __MAC_OS_X_VERSION_MIN_REQUIRED >= 1070
206 AVCaptureDeviceTransportControlsPlaybackMode mode =
207 [change[NSKeyValueChangeNewKey] integerValue];
209 if (mode != _context->observed_mode) {
210 if (mode == AVCaptureDeviceTransportControlsNotPlayingMode) {
211 _context->observed_quit = 1;
213 _context->observed_mode = mode;
217 [super observeValueForKeyPath: keyPath
224 - (void) captureOutput:(AVCaptureOutput *)captureOutput
225 didOutputSampleBuffer:(CMSampleBufferRef)videoFrame
226 fromConnection:(AVCaptureConnection *)connection
228 lock_frames(_context);
230 if (_context->current_frame != nil) {
231 CFRelease(_context->current_frame);
234 _context->current_frame = (CMSampleBufferRef)CFRetain(videoFrame);
236 unlock_frames(_context);
238 ++_context->frames_captured;
243 /** AudioReciever class - delegate for AVCaptureSession
245 @interface AVFAudioReceiver : NSObject
247 AVFContext* _context;
250 - (id)initWithContext:(AVFContext*)context;
252 - (void) captureOutput:(AVCaptureOutput *)captureOutput
253 didOutputSampleBuffer:(CMSampleBufferRef)audioFrame
254 fromConnection:(AVCaptureConnection *)connection;
258 @implementation AVFAudioReceiver
260 - (id)initWithContext:(AVFContext*)context
262 if (self = [super init]) {
268 - (void) captureOutput:(AVCaptureOutput *)captureOutput
269 didOutputSampleBuffer:(CMSampleBufferRef)audioFrame
270 fromConnection:(AVCaptureConnection *)connection
272 lock_frames(_context);
274 if (_context->current_audio_frame != nil) {
275 CFRelease(_context->current_audio_frame);
278 _context->current_audio_frame = (CMSampleBufferRef)CFRetain(audioFrame);
280 unlock_frames(_context);
282 ++_context->audio_frames_captured;
287 static void destroy_context(AVFContext* ctx)
289 [ctx->capture_session stopRunning];
291 [ctx->capture_session release];
292 [ctx->video_output release];
293 [ctx->audio_output release];
294 [ctx->avf_delegate release];
295 [ctx->avf_audio_delegate release];
297 ctx->capture_session = NULL;
298 ctx->video_output = NULL;
299 ctx->audio_output = NULL;
300 ctx->avf_delegate = NULL;
301 ctx->avf_audio_delegate = NULL;
304 av_freep(&ctx->audio_buffer);
306 pthread_mutex_destroy(&ctx->frame_lock);
308 if (ctx->current_frame) {
309 CFRelease(ctx->current_frame);
313 static int parse_device_name(AVFormatContext *s)
315 AVFContext *ctx = (AVFContext*)s->priv_data;
318 ctx->url = av_strdup(s->url);
321 return AVERROR(ENOMEM);
322 if (ctx->url[0] != ':') {
323 ctx->video_filename = av_strtok(ctx->url, ":", &save);
324 ctx->audio_filename = av_strtok(NULL, ":", &save);
326 ctx->audio_filename = av_strtok(ctx->url, ":", &save);
332 * Configure the video device.
334 * Configure the video device using a run-time approach to access properties
335 * since formats, activeFormat are available since iOS >= 7.0 or OSX >= 10.7
336 * and activeVideoMaxFrameDuration is available since i0S >= 7.0 and OSX >= 10.9.
338 * The NSUndefinedKeyException must be handled by the caller of this function.
341 static int configure_video_device(AVFormatContext *s, AVCaptureDevice *video_device)
343 AVFContext *ctx = (AVFContext*)s->priv_data;
345 double framerate = av_q2d(ctx->framerate);
346 NSObject *range = nil;
347 NSObject *format = nil;
348 NSObject *selected_range = nil;
349 NSObject *selected_format = nil;
351 // try to configure format by formats list
352 // might raise an exception if no format list is given
353 // (then fallback to default, no configuration)
355 for (format in [video_device valueForKey:@"formats"]) {
356 CMFormatDescriptionRef formatDescription;
357 CMVideoDimensions dimensions;
359 formatDescription = (CMFormatDescriptionRef) [format performSelector:@selector(formatDescription)];
360 dimensions = CMVideoFormatDescriptionGetDimensions(formatDescription);
362 if ((ctx->width == 0 && ctx->height == 0) ||
363 (dimensions.width == ctx->width && dimensions.height == ctx->height)) {
365 selected_format = format;
367 for (range in [format valueForKey:@"videoSupportedFrameRateRanges"]) {
368 double max_framerate;
370 [[range valueForKey:@"maxFrameRate"] getValue:&max_framerate];
371 if (fabs (framerate - max_framerate) < 0.01) {
372 selected_range = range;
379 if (!selected_format) {
380 av_log(s, AV_LOG_ERROR, "Selected video size (%dx%d) is not supported by the device.\n",
381 ctx->width, ctx->height);
382 goto unsupported_format;
385 if (!selected_range) {
386 av_log(s, AV_LOG_ERROR, "Selected framerate (%f) is not supported by the device.\n",
388 if (ctx->video_is_muxed) {
389 av_log(s, AV_LOG_ERROR, "Falling back to default.\n");
391 goto unsupported_format;
395 if ([video_device lockForConfiguration:NULL] == YES) {
396 if (selected_format) {
397 [video_device setValue:selected_format forKey:@"activeFormat"];
399 if (selected_range) {
400 NSValue *min_frame_duration = [selected_range valueForKey:@"minFrameDuration"];
401 [video_device setValue:min_frame_duration forKey:@"activeVideoMinFrameDuration"];
402 [video_device setValue:min_frame_duration forKey:@"activeVideoMaxFrameDuration"];
405 av_log(s, AV_LOG_ERROR, "Could not lock device for configuration.\n");
406 return AVERROR(EINVAL);
408 } @catch(NSException *e) {
409 av_log(ctx, AV_LOG_WARNING, "Configuration of video device failed, falling back to default.\n");
416 av_log(s, AV_LOG_ERROR, "Supported modes:\n");
417 for (format in [video_device valueForKey:@"formats"]) {
418 CMFormatDescriptionRef formatDescription;
419 CMVideoDimensions dimensions;
421 formatDescription = (CMFormatDescriptionRef) [format performSelector:@selector(formatDescription)];
422 dimensions = CMVideoFormatDescriptionGetDimensions(formatDescription);
424 for (range in [format valueForKey:@"videoSupportedFrameRateRanges"]) {
425 double min_framerate;
426 double max_framerate;
428 [[range valueForKey:@"minFrameRate"] getValue:&min_framerate];
429 [[range valueForKey:@"maxFrameRate"] getValue:&max_framerate];
430 av_log(s, AV_LOG_ERROR, " %dx%d@[%f %f]fps\n",
431 dimensions.width, dimensions.height,
432 min_framerate, max_framerate);
435 return AVERROR(EINVAL);
438 static int add_video_device(AVFormatContext *s, AVCaptureDevice *video_device)
440 AVFContext *ctx = (AVFContext*)s->priv_data;
442 NSError *error = nil;
443 AVCaptureInput* capture_input = nil;
444 struct AVFPixelFormatSpec pxl_fmt_spec;
445 NSNumber *pixel_format;
446 NSDictionary *capture_dict;
447 dispatch_queue_t queue;
449 if (ctx->video_device_index < ctx->num_video_devices) {
450 capture_input = (AVCaptureInput*) [[[AVCaptureDeviceInput alloc] initWithDevice:video_device error:&error] autorelease];
452 capture_input = (AVCaptureInput*) video_device;
455 if (!capture_input) {
456 av_log(s, AV_LOG_ERROR, "Failed to create AV capture input device: %s\n",
457 [[error localizedDescription] UTF8String]);
461 if ([ctx->capture_session canAddInput:capture_input]) {
462 [ctx->capture_session addInput:capture_input];
464 av_log(s, AV_LOG_ERROR, "can't add video input to capture session\n");
469 ctx->video_output = [[AVCaptureVideoDataOutput alloc] init];
471 if (!ctx->video_output) {
472 av_log(s, AV_LOG_ERROR, "Failed to init AV video output\n");
476 // Configure device framerate and video size
478 if ((ret = configure_video_device(s, video_device)) < 0) {
481 } @catch (NSException *exception) {
482 if (![[exception name] isEqualToString:NSUndefinedKeyException]) {
483 av_log (s, AV_LOG_ERROR, "An error occurred: %s", [exception.reason UTF8String]);
484 return AVERROR_EXTERNAL;
488 // select pixel format
489 pxl_fmt_spec.ff_id = AV_PIX_FMT_NONE;
491 for (int i = 0; avf_pixel_formats[i].ff_id != AV_PIX_FMT_NONE; i++) {
492 if (ctx->pixel_format == avf_pixel_formats[i].ff_id) {
493 pxl_fmt_spec = avf_pixel_formats[i];
498 // check if selected pixel format is supported by AVFoundation
499 if (pxl_fmt_spec.ff_id == AV_PIX_FMT_NONE) {
500 av_log(s, AV_LOG_ERROR, "Selected pixel format (%s) is not supported by AVFoundation.\n",
501 av_get_pix_fmt_name(pxl_fmt_spec.ff_id));
505 // check if the pixel format is available for this device
506 if ([[ctx->video_output availableVideoCVPixelFormatTypes] indexOfObject:[NSNumber numberWithInt:pxl_fmt_spec.avf_id]] == NSNotFound) {
507 av_log(s, AV_LOG_ERROR, "Selected pixel format (%s) is not supported by the input device.\n",
508 av_get_pix_fmt_name(pxl_fmt_spec.ff_id));
510 pxl_fmt_spec.ff_id = AV_PIX_FMT_NONE;
512 av_log(s, AV_LOG_ERROR, "Supported pixel formats:\n");
513 for (NSNumber *pxl_fmt in [ctx->video_output availableVideoCVPixelFormatTypes]) {
514 struct AVFPixelFormatSpec pxl_fmt_dummy;
515 pxl_fmt_dummy.ff_id = AV_PIX_FMT_NONE;
516 for (int i = 0; avf_pixel_formats[i].ff_id != AV_PIX_FMT_NONE; i++) {
517 if ([pxl_fmt intValue] == avf_pixel_formats[i].avf_id) {
518 pxl_fmt_dummy = avf_pixel_formats[i];
523 if (pxl_fmt_dummy.ff_id != AV_PIX_FMT_NONE) {
524 av_log(s, AV_LOG_ERROR, " %s\n", av_get_pix_fmt_name(pxl_fmt_dummy.ff_id));
526 // select first supported pixel format instead of user selected (or default) pixel format
527 if (pxl_fmt_spec.ff_id == AV_PIX_FMT_NONE) {
528 pxl_fmt_spec = pxl_fmt_dummy;
533 // fail if there is no appropriate pixel format or print a warning about overriding the pixel format
534 if (pxl_fmt_spec.ff_id == AV_PIX_FMT_NONE) {
537 av_log(s, AV_LOG_WARNING, "Overriding selected pixel format to use %s instead.\n",
538 av_get_pix_fmt_name(pxl_fmt_spec.ff_id));
542 // set videoSettings to an empty dict for receiving raw data of muxed devices
543 if (ctx->capture_raw_data) {
544 ctx->pixel_format = pxl_fmt_spec.ff_id;
545 ctx->video_output.videoSettings = @{ };
547 ctx->pixel_format = pxl_fmt_spec.ff_id;
548 pixel_format = [NSNumber numberWithUnsignedInt:pxl_fmt_spec.avf_id];
549 capture_dict = [NSDictionary dictionaryWithObject:pixel_format
550 forKey:(id)kCVPixelBufferPixelFormatTypeKey];
552 [ctx->video_output setVideoSettings:capture_dict];
554 [ctx->video_output setAlwaysDiscardsLateVideoFrames:ctx->drop_late_frames];
556 #if !TARGET_OS_IPHONE && __MAC_OS_X_VERSION_MIN_REQUIRED >= 1070
557 // check for transport control support and set observer device if supported
558 if (!ctx->video_is_screen) {
559 int trans_ctrl = [video_device transportControlsSupported];
560 AVCaptureDeviceTransportControlsPlaybackMode trans_mode = [video_device transportControlsPlaybackMode];
563 ctx->observed_mode = trans_mode;
564 ctx->observed_device = video_device;
569 ctx->avf_delegate = [[AVFFrameReceiver alloc] initWithContext:ctx];
571 queue = dispatch_queue_create("avf_queue", NULL);
572 [ctx->video_output setSampleBufferDelegate:ctx->avf_delegate queue:queue];
573 dispatch_release(queue);
575 if ([ctx->capture_session canAddOutput:ctx->video_output]) {
576 [ctx->capture_session addOutput:ctx->video_output];
578 av_log(s, AV_LOG_ERROR, "can't add video output to capture session\n");
585 static int add_audio_device(AVFormatContext *s, AVCaptureDevice *audio_device)
587 AVFContext *ctx = (AVFContext*)s->priv_data;
588 NSError *error = nil;
589 AVCaptureDeviceInput* audio_dev_input = [[[AVCaptureDeviceInput alloc] initWithDevice:audio_device error:&error] autorelease];
590 dispatch_queue_t queue;
592 if (!audio_dev_input) {
593 av_log(s, AV_LOG_ERROR, "Failed to create AV capture input device: %s\n",
594 [[error localizedDescription] UTF8String]);
598 if ([ctx->capture_session canAddInput:audio_dev_input]) {
599 [ctx->capture_session addInput:audio_dev_input];
601 av_log(s, AV_LOG_ERROR, "can't add audio input to capture session\n");
606 ctx->audio_output = [[AVCaptureAudioDataOutput alloc] init];
608 if (!ctx->audio_output) {
609 av_log(s, AV_LOG_ERROR, "Failed to init AV audio output\n");
613 ctx->avf_audio_delegate = [[AVFAudioReceiver alloc] initWithContext:ctx];
615 queue = dispatch_queue_create("avf_audio_queue", NULL);
616 [ctx->audio_output setSampleBufferDelegate:ctx->avf_audio_delegate queue:queue];
617 dispatch_release(queue);
619 if ([ctx->capture_session canAddOutput:ctx->audio_output]) {
620 [ctx->capture_session addOutput:ctx->audio_output];
622 av_log(s, AV_LOG_ERROR, "adding audio output to capture session failed\n");
629 static int get_video_config(AVFormatContext *s)
631 AVFContext *ctx = (AVFContext*)s->priv_data;
632 CVImageBufferRef image_buffer;
633 CMBlockBufferRef block_buffer;
634 CGSize image_buffer_size;
635 AVStream* stream = avformat_new_stream(s, NULL);
641 // Take stream info from the first frame.
642 while (ctx->frames_captured < 1) {
643 CFRunLoopRunInMode(kCFRunLoopDefaultMode, 0.1, YES);
648 ctx->video_stream_index = stream->index;
650 avpriv_set_pts_info(stream, 64, 1, avf_time_base);
652 image_buffer = CMSampleBufferGetImageBuffer(ctx->current_frame);
653 block_buffer = CMSampleBufferGetDataBuffer(ctx->current_frame);
656 image_buffer_size = CVImageBufferGetEncodedSize(image_buffer);
658 stream->codecpar->codec_id = AV_CODEC_ID_RAWVIDEO;
659 stream->codecpar->codec_type = AVMEDIA_TYPE_VIDEO;
660 stream->codecpar->width = (int)image_buffer_size.width;
661 stream->codecpar->height = (int)image_buffer_size.height;
662 stream->codecpar->format = ctx->pixel_format;
664 stream->codecpar->codec_id = AV_CODEC_ID_DVVIDEO;
665 stream->codecpar->codec_type = AVMEDIA_TYPE_VIDEO;
666 stream->codecpar->format = ctx->pixel_format;
669 CFRelease(ctx->current_frame);
670 ctx->current_frame = nil;
677 static int get_audio_config(AVFormatContext *s)
679 AVFContext *ctx = (AVFContext*)s->priv_data;
680 CMFormatDescriptionRef format_desc;
681 AVStream* stream = avformat_new_stream(s, NULL);
687 // Take stream info from the first frame.
688 while (ctx->audio_frames_captured < 1) {
689 CFRunLoopRunInMode(kCFRunLoopDefaultMode, 0.1, YES);
694 ctx->audio_stream_index = stream->index;
696 avpriv_set_pts_info(stream, 64, 1, avf_time_base);
698 format_desc = CMSampleBufferGetFormatDescription(ctx->current_audio_frame);
699 const AudioStreamBasicDescription *basic_desc = CMAudioFormatDescriptionGetStreamBasicDescription(format_desc);
703 av_log(s, AV_LOG_ERROR, "audio format not available\n");
707 stream->codecpar->codec_type = AVMEDIA_TYPE_AUDIO;
708 stream->codecpar->sample_rate = basic_desc->mSampleRate;
709 av_channel_layout_default(&stream->codecpar->ch_layout, basic_desc->mChannelsPerFrame);
711 ctx->audio_channels = basic_desc->mChannelsPerFrame;
712 ctx->audio_bits_per_sample = basic_desc->mBitsPerChannel;
713 ctx->audio_float = basic_desc->mFormatFlags & kAudioFormatFlagIsFloat;
714 ctx->audio_be = basic_desc->mFormatFlags & kAudioFormatFlagIsBigEndian;
715 ctx->audio_signed_integer = basic_desc->mFormatFlags & kAudioFormatFlagIsSignedInteger;
716 ctx->audio_packed = basic_desc->mFormatFlags & kAudioFormatFlagIsPacked;
717 ctx->audio_non_interleaved = basic_desc->mFormatFlags & kAudioFormatFlagIsNonInterleaved;
719 if (basic_desc->mFormatID == kAudioFormatLinearPCM &&
721 ctx->audio_bits_per_sample == 32 &&
723 stream->codecpar->codec_id = ctx->audio_be ? AV_CODEC_ID_PCM_F32BE : AV_CODEC_ID_PCM_F32LE;
724 } else if (basic_desc->mFormatID == kAudioFormatLinearPCM &&
725 ctx->audio_signed_integer &&
726 ctx->audio_bits_per_sample == 16 &&
728 stream->codecpar->codec_id = ctx->audio_be ? AV_CODEC_ID_PCM_S16BE : AV_CODEC_ID_PCM_S16LE;
729 } else if (basic_desc->mFormatID == kAudioFormatLinearPCM &&
730 ctx->audio_signed_integer &&
731 ctx->audio_bits_per_sample == 24 &&
733 stream->codecpar->codec_id = ctx->audio_be ? AV_CODEC_ID_PCM_S24BE : AV_CODEC_ID_PCM_S24LE;
734 } else if (basic_desc->mFormatID == kAudioFormatLinearPCM &&
735 ctx->audio_signed_integer &&
736 ctx->audio_bits_per_sample == 32 &&
738 stream->codecpar->codec_id = ctx->audio_be ? AV_CODEC_ID_PCM_S32BE : AV_CODEC_ID_PCM_S32LE;
741 av_log(s, AV_LOG_ERROR, "audio format is not supported\n");
745 if (ctx->audio_non_interleaved) {
746 CMBlockBufferRef block_buffer = CMSampleBufferGetDataBuffer(ctx->current_audio_frame);
747 ctx->audio_buffer_size = CMBlockBufferGetDataLength(block_buffer);
748 ctx->audio_buffer = av_malloc(ctx->audio_buffer_size);
749 if (!ctx->audio_buffer) {
751 av_log(s, AV_LOG_ERROR, "error allocating audio buffer\n");
756 CFRelease(ctx->current_audio_frame);
757 ctx->current_audio_frame = nil;
764 static int avf_read_header(AVFormatContext *s)
767 NSAutoreleasePool *pool = [[NSAutoreleasePool alloc] init];
768 uint32_t num_screens = 0;
769 AVFContext *ctx = (AVFContext*)s->priv_data;
770 AVCaptureDevice *video_device = nil;
771 AVCaptureDevice *audio_device = nil;
772 // Find capture device
773 NSArray *devices = [AVCaptureDevice devicesWithMediaType:AVMediaTypeVideo];
774 NSArray *devices_muxed = [AVCaptureDevice devicesWithMediaType:AVMediaTypeMuxed];
776 ctx->num_video_devices = [devices count] + [devices_muxed count];
778 pthread_mutex_init(&ctx->frame_lock, NULL);
780 #if !TARGET_OS_IPHONE && __MAC_OS_X_VERSION_MIN_REQUIRED >= 1070
781 CGGetActiveDisplayList(0, NULL, &num_screens);
784 // List devices if requested
785 if (ctx->list_devices) {
787 av_log(ctx, AV_LOG_INFO, "AVFoundation video devices:\n");
788 for (AVCaptureDevice *device in devices) {
789 const char *name = [[device localizedName] UTF8String];
790 index = [devices indexOfObject:device];
791 av_log(ctx, AV_LOG_INFO, "[%d] %s\n", index, name);
793 for (AVCaptureDevice *device in devices_muxed) {
794 const char *name = [[device localizedName] UTF8String];
795 index = [devices count] + [devices_muxed indexOfObject:device];
796 av_log(ctx, AV_LOG_INFO, "[%d] %s\n", index, name);
798 #if !TARGET_OS_IPHONE && __MAC_OS_X_VERSION_MIN_REQUIRED >= 1070
799 if (num_screens > 0) {
800 CGDirectDisplayID screens[num_screens];
801 CGGetActiveDisplayList(num_screens, screens, &num_screens);
802 for (int i = 0; i < num_screens; i++) {
803 av_log(ctx, AV_LOG_INFO, "[%d] Capture screen %d\n", ctx->num_video_devices + i, i);
808 av_log(ctx, AV_LOG_INFO, "AVFoundation audio devices:\n");
809 devices = [AVCaptureDevice devicesWithMediaType:AVMediaTypeAudio];
810 for (AVCaptureDevice *device in devices) {
811 const char *name = [[device localizedName] UTF8String];
812 int index = [devices indexOfObject:device];
813 av_log(ctx, AV_LOG_INFO, "[%d] %s\n", index, name);
818 // parse input filename for video and audio device
819 ret = parse_device_name(s);
823 // check for device index given in filename
824 if (ctx->video_device_index == -1 && ctx->video_filename) {
825 sscanf(ctx->video_filename, "%d", &ctx->video_device_index);
827 if (ctx->audio_device_index == -1 && ctx->audio_filename) {
828 sscanf(ctx->audio_filename, "%d", &ctx->audio_device_index);
831 if (ctx->video_device_index >= 0) {
832 if (ctx->video_device_index < ctx->num_video_devices) {
833 if (ctx->video_device_index < [devices count]) {
834 video_device = [devices objectAtIndex:ctx->video_device_index];
836 video_device = [devices_muxed objectAtIndex:(ctx->video_device_index - [devices count])];
837 ctx->video_is_muxed = 1;
839 } else if (ctx->video_device_index < ctx->num_video_devices + num_screens) {
840 #if !TARGET_OS_IPHONE && __MAC_OS_X_VERSION_MIN_REQUIRED >= 1070
841 CGDirectDisplayID screens[num_screens];
842 CGGetActiveDisplayList(num_screens, screens, &num_screens);
843 AVCaptureScreenInput* capture_screen_input = [[[AVCaptureScreenInput alloc] initWithDisplayID:screens[ctx->video_device_index - ctx->num_video_devices]] autorelease];
845 if (ctx->framerate.num > 0) {
846 capture_screen_input.minFrameDuration = CMTimeMake(ctx->framerate.den, ctx->framerate.num);
849 #if !TARGET_OS_IPHONE && __MAC_OS_X_VERSION_MIN_REQUIRED >= 1080
850 if (ctx->capture_cursor) {
851 capture_screen_input.capturesCursor = YES;
853 capture_screen_input.capturesCursor = NO;
857 if (ctx->capture_mouse_clicks) {
858 capture_screen_input.capturesMouseClicks = YES;
860 capture_screen_input.capturesMouseClicks = NO;
863 video_device = (AVCaptureDevice*) capture_screen_input;
864 ctx->video_is_screen = 1;
867 av_log(ctx, AV_LOG_ERROR, "Invalid device index\n");
870 } else if (ctx->video_filename &&
871 strncmp(ctx->video_filename, "none", 4)) {
872 if (!strncmp(ctx->video_filename, "default", 7)) {
873 video_device = [AVCaptureDevice defaultDeviceWithMediaType:AVMediaTypeVideo];
875 // looking for video inputs
876 for (AVCaptureDevice *device in devices) {
877 if (!strncmp(ctx->video_filename, [[device localizedName] UTF8String], strlen(ctx->video_filename))) {
878 video_device = device;
882 // looking for muxed inputs
883 for (AVCaptureDevice *device in devices_muxed) {
884 if (!strncmp(ctx->video_filename, [[device localizedName] UTF8String], strlen(ctx->video_filename))) {
885 video_device = device;
886 ctx->video_is_muxed = 1;
891 #if !TARGET_OS_IPHONE && __MAC_OS_X_VERSION_MIN_REQUIRED >= 1070
892 // looking for screen inputs
895 if(sscanf(ctx->video_filename, "Capture screen %d", &idx) && idx < num_screens) {
896 CGDirectDisplayID screens[num_screens];
897 CGGetActiveDisplayList(num_screens, screens, &num_screens);
898 AVCaptureScreenInput* capture_screen_input = [[[AVCaptureScreenInput alloc] initWithDisplayID:screens[idx]] autorelease];
899 video_device = (AVCaptureDevice*) capture_screen_input;
900 ctx->video_device_index = ctx->num_video_devices + idx;
901 ctx->video_is_screen = 1;
903 if (ctx->framerate.num > 0) {
904 capture_screen_input.minFrameDuration = CMTimeMake(ctx->framerate.den, ctx->framerate.num);
907 #if !TARGET_OS_IPHONE && __MAC_OS_X_VERSION_MIN_REQUIRED >= 1080
908 if (ctx->capture_cursor) {
909 capture_screen_input.capturesCursor = YES;
911 capture_screen_input.capturesCursor = NO;
915 if (ctx->capture_mouse_clicks) {
916 capture_screen_input.capturesMouseClicks = YES;
918 capture_screen_input.capturesMouseClicks = NO;
926 av_log(ctx, AV_LOG_ERROR, "Video device not found\n");
932 if (ctx->audio_device_index >= 0) {
933 NSArray *devices = [AVCaptureDevice devicesWithMediaType:AVMediaTypeAudio];
935 if (ctx->audio_device_index >= [devices count]) {
936 av_log(ctx, AV_LOG_ERROR, "Invalid audio device index\n");
940 audio_device = [devices objectAtIndex:ctx->audio_device_index];
941 } else if (ctx->audio_filename &&
942 strncmp(ctx->audio_filename, "none", 4)) {
943 if (!strncmp(ctx->audio_filename, "default", 7)) {
944 audio_device = [AVCaptureDevice defaultDeviceWithMediaType:AVMediaTypeAudio];
946 NSArray *devices = [AVCaptureDevice devicesWithMediaType:AVMediaTypeAudio];
948 for (AVCaptureDevice *device in devices) {
949 if (!strncmp(ctx->audio_filename, [[device localizedName] UTF8String], strlen(ctx->audio_filename))) {
950 audio_device = device;
957 av_log(ctx, AV_LOG_ERROR, "Audio device not found\n");
962 // Video nor Audio capture device not found, looking for AVMediaTypeVideo/Audio
963 if (!video_device && !audio_device) {
964 av_log(s, AV_LOG_ERROR, "No AV capture device found\n");
969 if (ctx->video_device_index < ctx->num_video_devices) {
970 av_log(s, AV_LOG_DEBUG, "'%s' opened\n", [[video_device localizedName] UTF8String]);
972 av_log(s, AV_LOG_DEBUG, "'%s' opened\n", [[video_device description] UTF8String]);
976 av_log(s, AV_LOG_DEBUG, "audio device '%s' opened\n", [[audio_device localizedName] UTF8String]);
979 // Initialize capture session
980 ctx->capture_session = [[AVCaptureSession alloc] init];
982 if (video_device && add_video_device(s, video_device)) {
985 if (audio_device && add_audio_device(s, audio_device)) {
988 [ctx->capture_session startRunning];
990 /* Unlock device configuration only after the session is started so it
991 * does not reset the capture formats */
992 if (!ctx->video_is_screen) {
993 [video_device unlockForConfiguration];
996 if (video_device && get_video_config(s)) {
1001 if (audio_device && get_audio_config(s)) {
1010 destroy_context(ctx);
1013 return AVERROR(EIO);
1016 static int copy_cvpixelbuffer(AVFormatContext *s,
1017 CVPixelBufferRef image_buffer,
1020 AVFContext *ctx = s->priv_data;
1021 int src_linesize[4];
1022 const uint8_t *src_data[4];
1023 int width = CVPixelBufferGetWidth(image_buffer);
1024 int height = CVPixelBufferGetHeight(image_buffer);
1027 memset(src_linesize, 0, sizeof(src_linesize));
1028 memset(src_data, 0, sizeof(src_data));
1030 status = CVPixelBufferLockBaseAddress(image_buffer, 0);
1031 if (status != kCVReturnSuccess) {
1032 av_log(s, AV_LOG_ERROR, "Could not lock base address: %d (%dx%d)\n", status, width, height);
1033 return AVERROR_EXTERNAL;
1036 if (CVPixelBufferIsPlanar(image_buffer)) {
1037 size_t plane_count = CVPixelBufferGetPlaneCount(image_buffer);
1039 for(i = 0; i < plane_count; i++){
1040 src_linesize[i] = CVPixelBufferGetBytesPerRowOfPlane(image_buffer, i);
1041 src_data[i] = CVPixelBufferGetBaseAddressOfPlane(image_buffer, i);
1044 src_linesize[0] = CVPixelBufferGetBytesPerRow(image_buffer);
1045 src_data[0] = CVPixelBufferGetBaseAddress(image_buffer);
1048 status = av_image_copy_to_buffer(pkt->data, pkt->size,
1049 src_data, src_linesize,
1050 ctx->pixel_format, width, height, 1);
1054 CVPixelBufferUnlockBaseAddress(image_buffer, 0);
1059 static int avf_read_packet(AVFormatContext *s, AVPacket *pkt)
1061 AVFContext* ctx = (AVFContext*)s->priv_data;
1064 CVImageBufferRef image_buffer;
1065 CMBlockBufferRef block_buffer;
1068 if (ctx->current_frame != nil) {
1072 image_buffer = CMSampleBufferGetImageBuffer(ctx->current_frame);
1073 block_buffer = CMSampleBufferGetDataBuffer(ctx->current_frame);
1075 if (image_buffer != nil) {
1076 length = (int)CVPixelBufferGetDataSize(image_buffer);
1077 } else if (block_buffer != nil) {
1078 length = (int)CMBlockBufferGetDataLength(block_buffer);
1081 return AVERROR(EINVAL);
1084 if (av_new_packet(pkt, length) < 0) {
1086 return AVERROR(EIO);
1090 CMSampleTimingInfo timing_info;
1092 if (CMSampleBufferGetOutputSampleTimingInfoArray(ctx->current_frame, 1, &timing_info, &count) == noErr) {
1093 AVRational timebase_q = av_make_q(1, timing_info.presentationTimeStamp.timescale);
1094 pkt->pts = pkt->dts = av_rescale_q(timing_info.presentationTimeStamp.value, timebase_q, avf_time_base_q);
1097 pkt->stream_index = ctx->video_stream_index;
1098 pkt->flags |= AV_PKT_FLAG_KEY;
1101 status = copy_cvpixelbuffer(s, image_buffer, pkt);
1104 OSStatus ret = CMBlockBufferCopyDataBytes(block_buffer, 0, pkt->size, pkt->data);
1105 if (ret != kCMBlockBufferNoErr) {
1106 status = AVERROR(EIO);
1109 CFRelease(ctx->current_frame);
1110 ctx->current_frame = nil;
1116 } else if (ctx->current_audio_frame != nil) {
1117 CMBlockBufferRef block_buffer = CMSampleBufferGetDataBuffer(ctx->current_audio_frame);
1118 int block_buffer_size = CMBlockBufferGetDataLength(block_buffer);
1120 if (!block_buffer || !block_buffer_size) {
1122 return AVERROR(EIO);
1125 if (ctx->audio_non_interleaved && block_buffer_size > ctx->audio_buffer_size) {
1127 return AVERROR_BUFFER_TOO_SMALL;
1130 if (av_new_packet(pkt, block_buffer_size) < 0) {
1132 return AVERROR(EIO);
1136 CMSampleTimingInfo timing_info;
1138 if (CMSampleBufferGetOutputSampleTimingInfoArray(ctx->current_audio_frame, 1, &timing_info, &count) == noErr) {
1139 AVRational timebase_q = av_make_q(1, timing_info.presentationTimeStamp.timescale);
1140 pkt->pts = pkt->dts = av_rescale_q(timing_info.presentationTimeStamp.value, timebase_q, avf_time_base_q);
1143 pkt->stream_index = ctx->audio_stream_index;
1144 pkt->flags |= AV_PKT_FLAG_KEY;
1146 if (ctx->audio_non_interleaved) {
1147 int sample, c, shift, num_samples;
1149 OSStatus ret = CMBlockBufferCopyDataBytes(block_buffer, 0, pkt->size, ctx->audio_buffer);
1150 if (ret != kCMBlockBufferNoErr) {
1152 return AVERROR(EIO);
1155 num_samples = pkt->size / (ctx->audio_channels * (ctx->audio_bits_per_sample >> 3));
1157 // transform decoded frame into output format
1158 #define INTERLEAVE_OUTPUT(bps) \
1160 int##bps##_t **src; \
1161 int##bps##_t *dest; \
1162 src = av_malloc(ctx->audio_channels * sizeof(int##bps##_t*)); \
1164 unlock_frames(ctx); \
1165 return AVERROR(EIO); \
1168 for (c = 0; c < ctx->audio_channels; c++) { \
1169 src[c] = ((int##bps##_t*)ctx->audio_buffer) + c * num_samples; \
1171 dest = (int##bps##_t*)pkt->data; \
1172 shift = bps - ctx->audio_bits_per_sample; \
1173 for (sample = 0; sample < num_samples; sample++) \
1174 for (c = 0; c < ctx->audio_channels; c++) \
1175 *dest++ = src[c][sample] << shift; \
1179 if (ctx->audio_bits_per_sample <= 16) {
1180 INTERLEAVE_OUTPUT(16)
1182 INTERLEAVE_OUTPUT(32)
1185 OSStatus ret = CMBlockBufferCopyDataBytes(block_buffer, 0, pkt->size, pkt->data);
1186 if (ret != kCMBlockBufferNoErr) {
1188 return AVERROR(EIO);
1192 CFRelease(ctx->current_audio_frame);
1193 ctx->current_audio_frame = nil;
1197 if (ctx->observed_quit) {
1200 return AVERROR(EAGAIN);
1205 } while (!pkt->data);
1210 static int avf_close(AVFormatContext *s)
1212 AVFContext* ctx = (AVFContext*)s->priv_data;
1213 destroy_context(ctx);
1217 static const AVOption options[] = {
1218 { "list_devices", "list available devices", offsetof(AVFContext, list_devices), AV_OPT_TYPE_BOOL, {.i64=0}, 0, 1, AV_OPT_FLAG_DECODING_PARAM },
1219 { "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 },
1220 { "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 },
1221 { "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},
1222 { "framerate", "set frame rate", offsetof(AVFContext, framerate), AV_OPT_TYPE_VIDEO_RATE, {.str = "ntsc"}, 0, INT_MAX, AV_OPT_FLAG_DECODING_PARAM },
1223 { "video_size", "set video size", offsetof(AVFContext, width), AV_OPT_TYPE_IMAGE_SIZE, {.str = NULL}, 0, 0, AV_OPT_FLAG_DECODING_PARAM },
1224 { "capture_cursor", "capture the screen cursor", offsetof(AVFContext, capture_cursor), AV_OPT_TYPE_BOOL, {.i64=0}, 0, 1, AV_OPT_FLAG_DECODING_PARAM },
1225 { "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 },
1226 { "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 },
1227 { "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 },
1232 static const AVClass avf_class = {
1233 .class_name = "AVFoundation indev",
1234 .item_name = av_default_item_name,
1236 .version = LIBAVUTIL_VERSION_INT,
1237 .category = AV_CLASS_CATEGORY_DEVICE_VIDEO_INPUT,
1240 const AVInputFormat ff_avfoundation_demuxer = {
1241 .name = "avfoundation",
1242 .long_name = NULL_IF_CONFIG_SMALL("AVFoundation input device"),
1243 .priv_data_size = sizeof(AVFContext),
1244 .read_header = avf_read_header,
1245 .read_packet = avf_read_packet,
1246 .read_close = avf_close,
1247 .flags = AVFMT_NOFILE,
1248 .priv_class = &avf_class,