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 "libavformat/internal.h"
34 #include "libavutil/internal.h"
35 #include "libavutil/time.h"
38 static const int avf_time_base = 1000000;
40 static const AVRational avf_time_base_q = {
45 struct AVFPixelFormatSpec {
46 enum AVPixelFormat ff_id;
50 static const struct AVFPixelFormatSpec avf_pixel_formats[] = {
51 { AV_PIX_FMT_MONOBLACK, kCVPixelFormatType_1Monochrome },
52 { AV_PIX_FMT_RGB555BE, kCVPixelFormatType_16BE555 },
53 { AV_PIX_FMT_RGB555LE, kCVPixelFormatType_16LE555 },
54 { AV_PIX_FMT_RGB565BE, kCVPixelFormatType_16BE565 },
55 { AV_PIX_FMT_RGB565LE, kCVPixelFormatType_16LE565 },
56 { AV_PIX_FMT_RGB24, kCVPixelFormatType_24RGB },
57 { AV_PIX_FMT_BGR24, kCVPixelFormatType_24BGR },
58 { AV_PIX_FMT_0RGB, kCVPixelFormatType_32ARGB },
59 { AV_PIX_FMT_BGR0, kCVPixelFormatType_32BGRA },
60 { AV_PIX_FMT_0BGR, kCVPixelFormatType_32ABGR },
61 { AV_PIX_FMT_RGB0, kCVPixelFormatType_32RGBA },
62 { AV_PIX_FMT_BGR48BE, kCVPixelFormatType_48RGB },
63 { AV_PIX_FMT_UYVY422, kCVPixelFormatType_422YpCbCr8 },
64 { AV_PIX_FMT_YUVA444P, kCVPixelFormatType_4444YpCbCrA8R },
65 { AV_PIX_FMT_YUVA444P16LE, kCVPixelFormatType_4444AYpCbCr16 },
66 { AV_PIX_FMT_YUV444P, kCVPixelFormatType_444YpCbCr8 },
67 { AV_PIX_FMT_YUV422P16, kCVPixelFormatType_422YpCbCr16 },
68 { AV_PIX_FMT_YUV422P10, kCVPixelFormatType_422YpCbCr10 },
69 { AV_PIX_FMT_YUV444P10, kCVPixelFormatType_444YpCbCr10 },
70 { AV_PIX_FMT_YUV420P, kCVPixelFormatType_420YpCbCr8Planar },
71 { AV_PIX_FMT_NV12, kCVPixelFormatType_420YpCbCr8BiPlanarVideoRange },
72 { AV_PIX_FMT_YUYV422, kCVPixelFormatType_422YpCbCr8_yuvs },
73 #if __MAC_OS_X_VERSION_MIN_REQUIRED >= 1080
74 { AV_PIX_FMT_GRAY8, kCVPixelFormatType_OneComponent8 },
76 { AV_PIX_FMT_NONE, 0 }
86 pthread_mutex_t frame_lock;
87 pthread_cond_t frame_wait_cond;
91 int video_device_index;
92 int video_stream_index;
93 enum AVPixelFormat pixel_format;
95 AVCaptureSession *capture_session;
96 AVCaptureVideoDataOutput *video_output;
97 CMSampleBufferRef current_frame;
100 static void lock_frames(AVFContext* ctx)
102 pthread_mutex_lock(&ctx->frame_lock);
105 static void unlock_frames(AVFContext* ctx)
107 pthread_mutex_unlock(&ctx->frame_lock);
110 /** FrameReciever class - delegate for AVCaptureSession
112 @interface AVFFrameReceiver : NSObject
114 AVFContext* _context;
117 - (id)initWithContext:(AVFContext*)context;
119 - (void) captureOutput:(AVCaptureOutput *)captureOutput
120 didOutputSampleBuffer:(CMSampleBufferRef)videoFrame
121 fromConnection:(AVCaptureConnection *)connection;
125 @implementation AVFFrameReceiver
127 - (id)initWithContext:(AVFContext*)context
129 if (self = [super init]) {
135 - (void) captureOutput:(AVCaptureOutput *)captureOutput
136 didOutputSampleBuffer:(CMSampleBufferRef)videoFrame
137 fromConnection:(AVCaptureConnection *)connection
139 lock_frames(_context);
141 if (_context->current_frame != nil) {
142 CFRelease(_context->current_frame);
145 _context->current_frame = (CMSampleBufferRef)CFRetain(videoFrame);
147 pthread_cond_signal(&_context->frame_wait_cond);
149 unlock_frames(_context);
151 ++_context->frames_captured;
156 static void destroy_context(AVFContext* ctx)
158 [ctx->capture_session stopRunning];
160 [ctx->capture_session release];
161 [ctx->video_output release];
162 [ctx->avf_delegate release];
164 ctx->capture_session = NULL;
165 ctx->video_output = NULL;
166 ctx->avf_delegate = NULL;
168 pthread_mutex_destroy(&ctx->frame_lock);
169 pthread_cond_destroy(&ctx->frame_wait_cond);
171 if (ctx->current_frame) {
172 CFRelease(ctx->current_frame);
176 static int add_video_device(AVFormatContext *s, AVCaptureDevice *video_device)
178 AVFContext *ctx = (AVFContext*)s->priv_data;
179 NSError *error = nil;
180 AVCaptureDeviceInput* capture_dev_input = [[[AVCaptureDeviceInput alloc] initWithDevice:video_device error:&error] autorelease];
182 if (!capture_dev_input) {
183 av_log(s, AV_LOG_ERROR, "Failed to create AV capture input device: %s\n",
184 [[error localizedDescription] UTF8String]);
188 if ([ctx->capture_session canAddInput:capture_dev_input]) {
189 [ctx->capture_session addInput:capture_dev_input];
191 av_log(s, AV_LOG_ERROR, "can't add video input to capture session\n");
196 ctx->video_output = [[AVCaptureVideoDataOutput alloc] init];
198 if (!ctx->video_output) {
199 av_log(s, AV_LOG_ERROR, "Failed to init AV video output\n");
203 // select pixel format
204 struct AVFPixelFormatSpec pxl_fmt_spec;
205 pxl_fmt_spec.ff_id = AV_PIX_FMT_NONE;
207 for (int i = 0; avf_pixel_formats[i].ff_id != AV_PIX_FMT_NONE; i++) {
208 if (ctx->pixel_format == avf_pixel_formats[i].ff_id) {
209 pxl_fmt_spec = avf_pixel_formats[i];
214 // check if selected pixel format is supported by AVFoundation
215 if (pxl_fmt_spec.ff_id == AV_PIX_FMT_NONE) {
216 av_log(s, AV_LOG_ERROR, "Selected pixel format (%s) is not supported by AVFoundation.\n",
217 av_get_pix_fmt_name(pxl_fmt_spec.ff_id));
221 // check if the pixel format is available for this device
222 if ([[ctx->video_output availableVideoCVPixelFormatTypes] indexOfObject:[NSNumber numberWithInt:pxl_fmt_spec.avf_id]] == NSNotFound) {
223 av_log(s, AV_LOG_ERROR, "Selected pixel format (%s) is not supported by the input device.\n",
224 av_get_pix_fmt_name(pxl_fmt_spec.ff_id));
226 pxl_fmt_spec.ff_id = AV_PIX_FMT_NONE;
228 av_log(s, AV_LOG_ERROR, "Supported pixel formats:\n");
229 for (NSNumber *pxl_fmt in [ctx->video_output availableVideoCVPixelFormatTypes]) {
230 struct AVFPixelFormatSpec pxl_fmt_dummy;
231 pxl_fmt_dummy.ff_id = AV_PIX_FMT_NONE;
232 for (int i = 0; avf_pixel_formats[i].ff_id != AV_PIX_FMT_NONE; i++) {
233 if ([pxl_fmt intValue] == avf_pixel_formats[i].avf_id) {
234 pxl_fmt_dummy = avf_pixel_formats[i];
239 if (pxl_fmt_dummy.ff_id != AV_PIX_FMT_NONE) {
240 av_log(s, AV_LOG_ERROR, " %s\n", av_get_pix_fmt_name(pxl_fmt_dummy.ff_id));
242 // select first supported pixel format instead of user selected (or default) pixel format
243 if (pxl_fmt_spec.ff_id == AV_PIX_FMT_NONE) {
244 pxl_fmt_spec = pxl_fmt_dummy;
249 // fail if there is no appropriate pixel format or print a warning about overriding the pixel format
250 if (pxl_fmt_spec.ff_id == AV_PIX_FMT_NONE) {
253 av_log(s, AV_LOG_WARNING, "Overriding selected pixel format to use %s instead.\n",
254 av_get_pix_fmt_name(pxl_fmt_spec.ff_id));
258 ctx->pixel_format = pxl_fmt_spec.ff_id;
259 NSNumber *pixel_format = [NSNumber numberWithUnsignedInt:pxl_fmt_spec.avf_id];
260 NSDictionary *capture_dict = [NSDictionary dictionaryWithObject:pixel_format
261 forKey:(id)kCVPixelBufferPixelFormatTypeKey];
263 [ctx->video_output setVideoSettings:capture_dict];
264 [ctx->video_output setAlwaysDiscardsLateVideoFrames:YES];
266 ctx->avf_delegate = [[AVFFrameReceiver alloc] initWithContext:ctx];
268 dispatch_queue_t queue = dispatch_queue_create("avf_queue", NULL);
269 [ctx->video_output setSampleBufferDelegate:ctx->avf_delegate queue:queue];
270 dispatch_release(queue);
272 if ([ctx->capture_session canAddOutput:ctx->video_output]) {
273 [ctx->capture_session addOutput:ctx->video_output];
275 av_log(s, AV_LOG_ERROR, "can't add video output to capture session\n");
282 static int get_video_config(AVFormatContext *s)
284 AVFContext *ctx = (AVFContext*)s->priv_data;
286 // Take stream info from the first frame.
287 while (ctx->frames_captured < 1) {
288 CFRunLoopRunInMode(kCFRunLoopDefaultMode, 0.1, YES);
293 AVStream* stream = avformat_new_stream(s, NULL);
299 ctx->video_stream_index = stream->index;
301 avpriv_set_pts_info(stream, 64, 1, avf_time_base);
303 CVImageBufferRef image_buffer = CMSampleBufferGetImageBuffer(ctx->current_frame);
304 CGSize image_buffer_size = CVImageBufferGetEncodedSize(image_buffer);
306 stream->codec->codec_id = AV_CODEC_ID_RAWVIDEO;
307 stream->codec->codec_type = AVMEDIA_TYPE_VIDEO;
308 stream->codec->width = (int)image_buffer_size.width;
309 stream->codec->height = (int)image_buffer_size.height;
310 stream->codec->pix_fmt = ctx->pixel_format;
312 CFRelease(ctx->current_frame);
313 ctx->current_frame = nil;
320 static int avf_read_header(AVFormatContext *s)
322 NSAutoreleasePool *pool = [[NSAutoreleasePool alloc] init];
323 AVFContext *ctx = (AVFContext*)s->priv_data;
324 ctx->first_pts = av_gettime();
326 pthread_mutex_init(&ctx->frame_lock, NULL);
327 pthread_cond_init(&ctx->frame_wait_cond, NULL);
329 // List devices if requested
330 if (ctx->list_devices) {
331 av_log(ctx, AV_LOG_INFO, "AVFoundation video devices:\n");
332 NSArray *devices = [AVCaptureDevice devicesWithMediaType:AVMediaTypeVideo];
333 for (AVCaptureDevice *device in devices) {
334 const char *name = [[device localizedName] UTF8String];
335 int index = [devices indexOfObject:device];
336 av_log(ctx, AV_LOG_INFO, "[%d] %s\n", index, name);
341 // Find capture device
342 AVCaptureDevice *video_device = nil;
344 // check for device index given in filename
345 if (ctx->video_device_index == -1) {
346 sscanf(s->filename, "%d", &ctx->video_device_index);
349 if (ctx->video_device_index >= 0) {
350 NSArray *devices = [AVCaptureDevice devicesWithMediaType:AVMediaTypeVideo];
352 if (ctx->video_device_index >= [devices count]) {
353 av_log(ctx, AV_LOG_ERROR, "Invalid device index\n");
357 video_device = [devices objectAtIndex:ctx->video_device_index];
358 } else if (strncmp(s->filename, "", 1) &&
359 strncmp(s->filename, "default", 7)) {
360 NSArray *devices = [AVCaptureDevice devicesWithMediaType:AVMediaTypeVideo];
362 for (AVCaptureDevice *device in devices) {
363 if (!strncmp(s->filename, [[device localizedName] UTF8String], strlen(s->filename))) {
364 video_device = device;
370 av_log(ctx, AV_LOG_ERROR, "Video device not found\n");
374 video_device = [AVCaptureDevice defaultDeviceWithMediaType:AVMediaTypeVideo];
377 // Video capture device not found, looking for AVMediaTypeVideo
379 video_device = [AVCaptureDevice defaultDeviceWithMediaType:AVMediaTypeVideo];
382 av_log(s, AV_LOG_ERROR, "No AV capture device found\n");
387 av_log(s, AV_LOG_DEBUG, "'%s' opened\n", [[video_device localizedName] UTF8String]);
389 // Initialize capture session
390 ctx->capture_session = [[AVCaptureSession alloc] init];
392 if (add_video_device(s, video_device)) {
396 [ctx->capture_session startRunning];
398 if (get_video_config(s)) {
407 destroy_context(ctx);
411 static int avf_read_packet(AVFormatContext *s, AVPacket *pkt)
413 AVFContext* ctx = (AVFContext*)s->priv_data;
418 CVImageBufferRef image_buffer = CMSampleBufferGetImageBuffer(ctx->current_frame);
420 if (ctx->current_frame != nil) {
421 if (av_new_packet(pkt, (int)CVPixelBufferGetDataSize(image_buffer)) < 0) {
425 pkt->pts = pkt->dts = av_rescale_q(av_gettime() - ctx->first_pts,
428 pkt->stream_index = ctx->video_stream_index;
429 pkt->flags |= AV_PKT_FLAG_KEY;
431 CVPixelBufferLockBaseAddress(image_buffer, 0);
433 void* data = CVPixelBufferGetBaseAddress(image_buffer);
434 memcpy(pkt->data, data, pkt->size);
436 CVPixelBufferUnlockBaseAddress(image_buffer, 0);
437 CFRelease(ctx->current_frame);
438 ctx->current_frame = nil;
441 pthread_cond_wait(&ctx->frame_wait_cond, &ctx->frame_lock);
445 } while (!pkt->data);
450 static int avf_close(AVFormatContext *s)
452 AVFContext* ctx = (AVFContext*)s->priv_data;
453 destroy_context(ctx);
457 static const AVOption options[] = {
458 { "frame_rate", "set frame rate", offsetof(AVFContext, frame_rate), AV_OPT_TYPE_FLOAT, { .dbl = 30.0 }, 0.1, 30.0, AV_OPT_TYPE_VIDEO_RATE, NULL },
459 { "list_devices", "list available devices", offsetof(AVFContext, list_devices), AV_OPT_TYPE_INT, {.i64=0}, 0, 1, AV_OPT_FLAG_DECODING_PARAM, "list_devices" },
460 { "true", "", 0, AV_OPT_TYPE_CONST, {.i64=1}, 0, 0, AV_OPT_FLAG_DECODING_PARAM, "list_devices" },
461 { "false", "", 0, AV_OPT_TYPE_CONST, {.i64=0}, 0, 0, AV_OPT_FLAG_DECODING_PARAM, "list_devices" },
462 { "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 },
463 { "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},
467 static const AVClass avf_class = {
468 .class_name = "AVFoundation input device",
469 .item_name = av_default_item_name,
471 .version = LIBAVUTIL_VERSION_INT,
472 .category = AV_CLASS_CATEGORY_DEVICE_VIDEO_INPUT,
475 AVInputFormat ff_avfoundation_demuxer = {
476 .name = "avfoundation",
477 .long_name = NULL_IF_CONFIG_SMALL("AVFoundation input device"),
478 .priv_data_size = sizeof(AVFContext),
479 .read_header = avf_read_header,
480 .read_packet = avf_read_packet,
481 .read_close = avf_close,
482 .flags = AVFMT_NOFILE,
483 .priv_class = &avf_class,