2 * Directshow capture interface
3 * Copyright (c) 2010 Ramiro Polla
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
22 #include "dshow_capture.h"
23 #include "libavutil/parseutils.h"
24 #include "libavutil/pixdesc.h"
25 #include "libavutil/opt.h"
26 #include "libavformat/internal.h"
27 #include "libavformat/riff.h"
29 #include "libavcodec/raw.h"
34 static enum AVPixelFormat dshow_pixfmt(DWORD biCompression, WORD biBitCount)
36 switch(biCompression) {
39 switch(biBitCount) { /* 1-8 are untested */
41 return AV_PIX_FMT_MONOWHITE;
43 return AV_PIX_FMT_RGB4;
45 return AV_PIX_FMT_RGB8;
47 return AV_PIX_FMT_RGB555;
49 return AV_PIX_FMT_BGR24;
51 return AV_PIX_FMT_0RGB32;
54 return avpriv_find_pix_fmt(avpriv_get_raw_pix_fmt_tags(), biCompression); // all others
58 dshow_read_close(AVFormatContext *s)
60 struct dshow_ctx *ctx = s->priv_data;
64 IMediaControl_Stop(ctx->control);
65 IMediaControl_Release(ctx->control);
69 IMediaEvent_Release(ctx->media_event);
74 r = IGraphBuilder_EnumFilters(ctx->graph, &fenum);
77 IEnumFilters_Reset(fenum);
78 while (IEnumFilters_Next(fenum, 1, &f, NULL) == S_OK) {
79 if (IGraphBuilder_RemoveFilter(ctx->graph, f) == S_OK)
80 IEnumFilters_Reset(fenum); /* When a filter is removed,
81 * the list must be reset. */
82 IBaseFilter_Release(f);
84 IEnumFilters_Release(fenum);
86 IGraphBuilder_Release(ctx->graph);
89 if (ctx->capture_pin[VideoDevice])
90 libAVPin_Release(ctx->capture_pin[VideoDevice]);
91 if (ctx->capture_pin[AudioDevice])
92 libAVPin_Release(ctx->capture_pin[AudioDevice]);
93 if (ctx->capture_filter[VideoDevice])
94 libAVFilter_Release(ctx->capture_filter[VideoDevice]);
95 if (ctx->capture_filter[AudioDevice])
96 libAVFilter_Release(ctx->capture_filter[AudioDevice]);
98 if (ctx->device_pin[VideoDevice])
99 IPin_Release(ctx->device_pin[VideoDevice]);
100 if (ctx->device_pin[AudioDevice])
101 IPin_Release(ctx->device_pin[AudioDevice]);
102 if (ctx->device_filter[VideoDevice])
103 IBaseFilter_Release(ctx->device_filter[VideoDevice]);
104 if (ctx->device_filter[AudioDevice])
105 IBaseFilter_Release(ctx->device_filter[AudioDevice]);
107 av_freep(&ctx->device_name[0]);
108 av_freep(&ctx->device_name[1]);
109 av_freep(&ctx->device_unique_name[0]);
110 av_freep(&ctx->device_unique_name[1]);
113 CloseHandle(ctx->mutex);
115 CloseHandle(ctx->event[0]);
117 CloseHandle(ctx->event[1]);
121 AVPacketList *next = pktl->next;
122 av_packet_unref(&pktl->pkt);
132 static char *dup_wchar_to_utf8(wchar_t *w)
135 int l = WideCharToMultiByte(CP_UTF8, 0, w, -1, 0, 0, 0, 0);
138 WideCharToMultiByte(CP_UTF8, 0, w, -1, s, l, 0, 0);
142 static int shall_we_drop(AVFormatContext *s, int index, enum dshowDeviceType devtype)
144 struct dshow_ctx *ctx = s->priv_data;
145 static const uint8_t dropscore[] = {62, 75, 87, 100};
146 const int ndropscores = FF_ARRAY_ELEMS(dropscore);
147 unsigned int buffer_fullness = (ctx->curbufsize[index]*100)/s->max_picture_buffer;
148 const char *devtypename = (devtype == VideoDevice) ? "video" : "audio";
150 if(dropscore[++ctx->video_frame_num%ndropscores] <= buffer_fullness) {
151 av_log(s, AV_LOG_ERROR,
152 "real-time buffer [%s] [%s input] too full or near too full (%d%% of size: %d [rtbufsize parameter])! frame dropped!\n",
153 ctx->device_name[devtype], devtypename, buffer_fullness, s->max_picture_buffer);
161 callback(void *priv_data, int index, uint8_t *buf, int buf_size, int64_t time, enum dshowDeviceType devtype)
163 AVFormatContext *s = priv_data;
164 struct dshow_ctx *ctx = s->priv_data;
165 AVPacketList **ppktl, *pktl_next;
167 // dump_videohdr(s, vdhdr);
169 WaitForSingleObject(ctx->mutex, INFINITE);
171 if(shall_we_drop(s, index, devtype))
174 pktl_next = av_mallocz(sizeof(AVPacketList));
178 if(av_new_packet(&pktl_next->pkt, buf_size) < 0) {
183 pktl_next->pkt.stream_index = index;
184 pktl_next->pkt.pts = time;
185 memcpy(pktl_next->pkt.data, buf, buf_size);
187 for(ppktl = &ctx->pktl ; *ppktl ; ppktl = &(*ppktl)->next);
189 ctx->curbufsize[index] += buf_size;
191 SetEvent(ctx->event[1]);
192 ReleaseMutex(ctx->mutex);
196 ReleaseMutex(ctx->mutex);
201 * Cycle through available devices using the device enumerator devenum,
202 * retrieve the device with type specified by devtype and return the
203 * pointer to the object found in *pfilter.
204 * If pfilter is NULL, list all device names.
207 dshow_cycle_devices(AVFormatContext *avctx, ICreateDevEnum *devenum,
208 enum dshowDeviceType devtype, enum dshowSourceFilterType sourcetype,
209 IBaseFilter **pfilter, char **device_unique_name)
211 struct dshow_ctx *ctx = avctx->priv_data;
212 IBaseFilter *device_filter = NULL;
213 IEnumMoniker *classenum = NULL;
215 const char *device_name = ctx->device_name[devtype];
216 int skip = (devtype == VideoDevice) ? ctx->video_device_number
217 : ctx->audio_device_number;
220 const GUID *device_guid[2] = { &CLSID_VideoInputDeviceCategory,
221 &CLSID_AudioInputDeviceCategory };
222 const char *devtypename = (devtype == VideoDevice) ? "video" : "audio only";
223 const char *sourcetypename = (sourcetype == VideoSourceDevice) ? "video" : "audio";
225 r = ICreateDevEnum_CreateClassEnumerator(devenum, device_guid[sourcetype],
226 (IEnumMoniker **) &classenum, 0);
228 av_log(avctx, AV_LOG_ERROR, "Could not enumerate %s devices (or none found).\n",
233 while (!device_filter && IEnumMoniker_Next(classenum, 1, &m, NULL) == S_OK) {
234 IPropertyBag *bag = NULL;
235 char *friendly_name = NULL;
236 char *unique_name = NULL;
238 IBindCtx *bind_ctx = NULL;
239 LPOLESTR olestr = NULL;
240 LPMALLOC co_malloc = NULL;
243 r = CoGetMalloc(1, &co_malloc);
246 r = CreateBindCtx(0, &bind_ctx);
249 /* GetDisplayname works for both video and audio, DevicePath doesn't */
250 r = IMoniker_GetDisplayName(m, bind_ctx, NULL, &olestr);
253 unique_name = dup_wchar_to_utf8(olestr);
254 /* replace ':' with '_' since we use : to delineate between sources */
255 for (i = 0; i < strlen(unique_name); i++) {
256 if (unique_name[i] == ':')
257 unique_name[i] = '_';
260 r = IMoniker_BindToStorage(m, 0, 0, &IID_IPropertyBag, (void *) &bag);
265 r = IPropertyBag_Read(bag, L"FriendlyName", &var, NULL);
268 friendly_name = dup_wchar_to_utf8(var.bstrVal);
271 if (strcmp(device_name, friendly_name) && strcmp(device_name, unique_name))
275 r = IMoniker_BindToObject(m, 0, 0, &IID_IBaseFilter, (void *) &device_filter);
277 av_log(avctx, AV_LOG_ERROR, "Unable to BindToObject for %s\n", device_name);
280 *device_unique_name = unique_name;
281 // success, loop will end now
284 av_log(avctx, AV_LOG_INFO, " \"%s\"\n", friendly_name);
285 av_log(avctx, AV_LOG_INFO, " Alternative name \"%s\"\n", unique_name);
286 av_free(unique_name);
290 if (olestr && co_malloc)
291 IMalloc_Free(co_malloc, olestr);
293 IBindCtx_Release(bind_ctx);
294 av_free(friendly_name);
296 IPropertyBag_Release(bag);
300 IEnumMoniker_Release(classenum);
303 if (!device_filter) {
304 av_log(avctx, AV_LOG_ERROR, "Could not find %s device with name [%s] among source devices of type %s.\n",
305 devtypename, device_name, sourcetypename);
308 *pfilter = device_filter;
315 * Cycle through available formats using the specified pin,
316 * try to set parameters specified through AVOptions and if successful
317 * return 1 in *pformat_set.
318 * If pformat_set is NULL, list all pin capabilities.
321 dshow_cycle_formats(AVFormatContext *avctx, enum dshowDeviceType devtype,
322 IPin *pin, int *pformat_set)
324 struct dshow_ctx *ctx = avctx->priv_data;
325 IAMStreamConfig *config = NULL;
326 AM_MEDIA_TYPE *type = NULL;
331 if (IPin_QueryInterface(pin, &IID_IAMStreamConfig, (void **) &config) != S_OK)
333 if (IAMStreamConfig_GetNumberOfCapabilities(config, &n, &size) != S_OK)
336 caps = av_malloc(size);
340 for (i = 0; i < n && !format_set; i++) {
341 r = IAMStreamConfig_GetStreamCaps(config, i, &type, (void *) caps);
345 ff_print_AM_MEDIA_TYPE(type);
348 if (devtype == VideoDevice) {
349 VIDEO_STREAM_CONFIG_CAPS *vcaps = caps;
350 BITMAPINFOHEADER *bih;
352 const AVCodecTag *const tags[] = { avformat_get_riff_video_tags(), NULL };
354 ff_print_VIDEO_STREAM_CONFIG_CAPS(vcaps);
356 if (IsEqualGUID(&type->formattype, &FORMAT_VideoInfo)) {
357 VIDEOINFOHEADER *v = (void *) type->pbFormat;
358 fr = &v->AvgTimePerFrame;
360 } else if (IsEqualGUID(&type->formattype, &FORMAT_VideoInfo2)) {
361 VIDEOINFOHEADER2 *v = (void *) type->pbFormat;
362 fr = &v->AvgTimePerFrame;
368 enum AVPixelFormat pix_fmt = dshow_pixfmt(bih->biCompression, bih->biBitCount);
369 if (pix_fmt == AV_PIX_FMT_NONE) {
370 enum AVCodecID codec_id = av_codec_get_id(tags, bih->biCompression);
371 AVCodec *codec = avcodec_find_decoder(codec_id);
372 if (codec_id == AV_CODEC_ID_NONE || !codec) {
373 av_log(avctx, AV_LOG_INFO, " unknown compression type 0x%X", (int) bih->biCompression);
375 av_log(avctx, AV_LOG_INFO, " vcodec=%s", codec->name);
378 av_log(avctx, AV_LOG_INFO, " pixel_format=%s", av_get_pix_fmt_name(pix_fmt));
380 av_log(avctx, AV_LOG_INFO, " min s=%ldx%ld fps=%g max s=%ldx%ld fps=%g\n",
381 vcaps->MinOutputSize.cx, vcaps->MinOutputSize.cy,
382 1e7 / vcaps->MaxFrameInterval,
383 vcaps->MaxOutputSize.cx, vcaps->MaxOutputSize.cy,
384 1e7 / vcaps->MinFrameInterval);
387 if (ctx->video_codec_id != AV_CODEC_ID_RAWVIDEO) {
388 if (ctx->video_codec_id != av_codec_get_id(tags, bih->biCompression))
391 if (ctx->pixel_format != AV_PIX_FMT_NONE &&
392 ctx->pixel_format != dshow_pixfmt(bih->biCompression, bih->biBitCount)) {
395 if (ctx->framerate) {
396 int64_t framerate = ((int64_t) ctx->requested_framerate.den*10000000)
397 / ctx->requested_framerate.num;
398 if (framerate > vcaps->MaxFrameInterval ||
399 framerate < vcaps->MinFrameInterval)
403 if (ctx->requested_width && ctx->requested_height) {
404 if (ctx->requested_width > vcaps->MaxOutputSize.cx ||
405 ctx->requested_width < vcaps->MinOutputSize.cx ||
406 ctx->requested_height > vcaps->MaxOutputSize.cy ||
407 ctx->requested_height < vcaps->MinOutputSize.cy)
409 bih->biWidth = ctx->requested_width;
410 bih->biHeight = ctx->requested_height;
413 AUDIO_STREAM_CONFIG_CAPS *acaps = caps;
416 ff_print_AUDIO_STREAM_CONFIG_CAPS(acaps);
418 if (IsEqualGUID(&type->formattype, &FORMAT_WaveFormatEx)) {
419 fx = (void *) type->pbFormat;
424 av_log(avctx, AV_LOG_INFO, " min ch=%lu bits=%lu rate=%6lu max ch=%lu bits=%lu rate=%6lu\n",
425 acaps->MinimumChannels, acaps->MinimumBitsPerSample, acaps->MinimumSampleFrequency,
426 acaps->MaximumChannels, acaps->MaximumBitsPerSample, acaps->MaximumSampleFrequency);
429 if (ctx->sample_rate) {
430 if (ctx->sample_rate > acaps->MaximumSampleFrequency ||
431 ctx->sample_rate < acaps->MinimumSampleFrequency)
433 fx->nSamplesPerSec = ctx->sample_rate;
435 if (ctx->sample_size) {
436 if (ctx->sample_size > acaps->MaximumBitsPerSample ||
437 ctx->sample_size < acaps->MinimumBitsPerSample)
439 fx->wBitsPerSample = ctx->sample_size;
442 if (ctx->channels > acaps->MaximumChannels ||
443 ctx->channels < acaps->MinimumChannels)
445 fx->nChannels = ctx->channels;
448 if (IAMStreamConfig_SetFormat(config, type) != S_OK)
453 CoTaskMemFree(type->pbFormat);
457 IAMStreamConfig_Release(config);
460 *pformat_set = format_set;
464 * Set audio device buffer size in milliseconds (which can directly impact
465 * latency, depending on the device).
468 dshow_set_audio_buffer_size(AVFormatContext *avctx, IPin *pin)
470 struct dshow_ctx *ctx = avctx->priv_data;
471 IAMBufferNegotiation *buffer_negotiation = NULL;
472 ALLOCATOR_PROPERTIES props = { -1, -1, -1, -1 };
473 IAMStreamConfig *config = NULL;
474 AM_MEDIA_TYPE *type = NULL;
475 int ret = AVERROR(EIO);
477 if (IPin_QueryInterface(pin, &IID_IAMStreamConfig, (void **) &config) != S_OK)
479 if (IAMStreamConfig_GetFormat(config, &type) != S_OK)
481 if (!IsEqualGUID(&type->formattype, &FORMAT_WaveFormatEx))
484 props.cbBuffer = (((WAVEFORMATEX *) type->pbFormat)->nAvgBytesPerSec)
485 * ctx->audio_buffer_size / 1000;
487 if (IPin_QueryInterface(pin, &IID_IAMBufferNegotiation, (void **) &buffer_negotiation) != S_OK)
489 if (IAMBufferNegotiation_SuggestAllocatorProperties(buffer_negotiation, &props) != S_OK)
495 if (buffer_negotiation)
496 IAMBufferNegotiation_Release(buffer_negotiation);
499 CoTaskMemFree(type->pbFormat);
503 IAMStreamConfig_Release(config);
509 * Pops up a user dialog allowing them to adjust properties for the given filter, if possible.
512 dshow_show_filter_properties(IBaseFilter *device_filter, AVFormatContext *avctx) {
513 ISpecifyPropertyPages *property_pages = NULL;
514 IUnknown *device_filter_iunknown = NULL;
516 FILTER_INFO filter_info = {0}; /* a warning on this line is false positive GCC bug 53119 AFAICT */
517 CAUUID ca_guid = {0};
519 hr = IBaseFilter_QueryInterface(device_filter, &IID_ISpecifyPropertyPages, (void **)&property_pages);
521 av_log(avctx, AV_LOG_WARNING, "requested filter does not have a property page to show");
524 hr = IBaseFilter_QueryFilterInfo(device_filter, &filter_info);
528 hr = IBaseFilter_QueryInterface(device_filter, &IID_IUnknown, (void **)&device_filter_iunknown);
532 hr = ISpecifyPropertyPages_GetPages(property_pages, &ca_guid);
536 hr = OleCreatePropertyFrame(NULL, 0, 0, filter_info.achName, 1, &device_filter_iunknown, ca_guid.cElems,
537 ca_guid.pElems, 0, 0, NULL);
543 av_log(avctx, AV_LOG_ERROR, "Failure showing property pages for filter");
546 ISpecifyPropertyPages_Release(property_pages);
547 if (device_filter_iunknown)
548 IUnknown_Release(device_filter_iunknown);
549 if (filter_info.pGraph)
550 IFilterGraph_Release(filter_info.pGraph);
552 CoTaskMemFree(ca_guid.pElems);
556 * Cycle through available pins using the device_filter device, of type
557 * devtype, retrieve the first output pin and return the pointer to the
558 * object found in *ppin.
559 * If ppin is NULL, cycle through all pins listing audio/video capabilities.
562 dshow_cycle_pins(AVFormatContext *avctx, enum dshowDeviceType devtype,
563 enum dshowSourceFilterType sourcetype, IBaseFilter *device_filter, IPin **ppin)
565 struct dshow_ctx *ctx = avctx->priv_data;
567 IPin *device_pin = NULL;
571 const GUID *mediatype[2] = { &MEDIATYPE_Video, &MEDIATYPE_Audio };
572 const char *devtypename = (devtype == VideoDevice) ? "video" : "audio only";
573 const char *sourcetypename = (sourcetype == VideoSourceDevice) ? "video" : "audio";
575 int set_format = (devtype == VideoDevice && (ctx->framerate ||
576 (ctx->requested_width && ctx->requested_height) ||
577 ctx->pixel_format != AV_PIX_FMT_NONE ||
578 ctx->video_codec_id != AV_CODEC_ID_RAWVIDEO))
579 || (devtype == AudioDevice && (ctx->channels || ctx->sample_rate));
581 int should_show_properties = (devtype == VideoDevice) ? ctx->show_video_device_dialog : ctx->show_audio_device_dialog;
583 if (should_show_properties)
584 dshow_show_filter_properties(device_filter, avctx);
586 r = IBaseFilter_EnumPins(device_filter, &pins);
588 av_log(avctx, AV_LOG_ERROR, "Could not enumerate pins.\n");
593 av_log(avctx, AV_LOG_INFO, "DirectShow %s device options (from %s devices)\n",
594 devtypename, sourcetypename);
597 while (!device_pin && IEnumPins_Next(pins, 1, &pin, NULL) == S_OK) {
598 IKsPropertySet *p = NULL;
599 IEnumMediaTypes *types = NULL;
604 char *name_buf = NULL;
605 wchar_t *pin_id = NULL;
606 char *pin_buf = NULL;
607 char *desired_pin_name = devtype == VideoDevice ? ctx->video_pin_name : ctx->audio_pin_name;
609 IPin_QueryPinInfo(pin, &info);
610 IBaseFilter_Release(info.pFilter);
612 if (info.dir != PINDIR_OUTPUT)
614 if (IPin_QueryInterface(pin, &IID_IKsPropertySet, (void **) &p) != S_OK)
616 if (IKsPropertySet_Get(p, &ROPSETID_Pin, AMPROPERTY_PIN_CATEGORY,
617 NULL, 0, &category, sizeof(GUID), &r2) != S_OK)
619 if (!IsEqualGUID(&category, &PIN_CATEGORY_CAPTURE))
621 name_buf = dup_wchar_to_utf8(info.achName);
623 r = IPin_QueryId(pin, &pin_id);
625 av_log(avctx, AV_LOG_ERROR, "Could not query pin id\n");
628 pin_buf = dup_wchar_to_utf8(pin_id);
631 av_log(avctx, AV_LOG_INFO, " Pin \"%s\" (alternative pin name \"%s\")\n", name_buf, pin_buf);
632 dshow_cycle_formats(avctx, devtype, pin, NULL);
636 if (desired_pin_name) {
637 if(strcmp(name_buf, desired_pin_name) && strcmp(pin_buf, desired_pin_name)) {
638 av_log(avctx, AV_LOG_DEBUG, "skipping pin \"%s\" (\"%s\") != requested \"%s\"\n",
639 name_buf, pin_buf, desired_pin_name);
645 dshow_cycle_formats(avctx, devtype, pin, &format_set);
650 if (devtype == AudioDevice && ctx->audio_buffer_size) {
651 if (dshow_set_audio_buffer_size(avctx, pin) < 0) {
652 av_log(avctx, AV_LOG_ERROR, "unable to set audio buffer size %d to pin, using pin anyway...", ctx->audio_buffer_size);
656 if (IPin_EnumMediaTypes(pin, &types) != S_OK)
659 IEnumMediaTypes_Reset(types);
660 /* in case format_set was not called, just verify the majortype */
661 while (!device_pin && IEnumMediaTypes_Next(types, 1, &type, NULL) == S_OK) {
662 if (IsEqualGUID(&type->majortype, mediatype[devtype])) {
664 av_log(avctx, AV_LOG_DEBUG, "Selecting pin %s on %s\n", name_buf, devtypename);
672 IEnumMediaTypes_Release(types);
674 IKsPropertySet_Release(p);
675 if (device_pin != pin)
680 CoTaskMemFree(pin_id);
683 IEnumPins_Release(pins);
686 if (set_format && !format_set) {
687 av_log(avctx, AV_LOG_ERROR, "Could not set %s options\n", devtypename);
691 av_log(avctx, AV_LOG_ERROR,
692 "Could not find output pin from %s capture device.\n", devtypename);
702 * List options for device with type devtype, source filter type sourcetype
704 * @param devenum device enumerator used for accessing the device
707 dshow_list_device_options(AVFormatContext *avctx, ICreateDevEnum *devenum,
708 enum dshowDeviceType devtype, enum dshowSourceFilterType sourcetype)
710 struct dshow_ctx *ctx = avctx->priv_data;
711 IBaseFilter *device_filter = NULL;
712 char *device_unique_name = NULL;
715 if ((r = dshow_cycle_devices(avctx, devenum, devtype, sourcetype, &device_filter, &device_unique_name)) < 0)
717 ctx->device_filter[devtype] = device_filter;
718 if ((r = dshow_cycle_pins(avctx, devtype, sourcetype, device_filter, NULL)) < 0)
720 av_freep(&device_unique_name);
725 dshow_open_device(AVFormatContext *avctx, ICreateDevEnum *devenum,
726 enum dshowDeviceType devtype, enum dshowSourceFilterType sourcetype)
728 struct dshow_ctx *ctx = avctx->priv_data;
729 IBaseFilter *device_filter = NULL;
730 char *device_filter_unique_name = NULL;
731 IGraphBuilder *graph = ctx->graph;
732 IPin *device_pin = NULL;
733 libAVPin *capture_pin = NULL;
734 libAVFilter *capture_filter = NULL;
735 ICaptureGraphBuilder2 *graph_builder2 = NULL;
736 int ret = AVERROR(EIO);
738 IStream *ifile_stream = NULL;
739 IStream *ofile_stream = NULL;
740 IPersistStream *pers_stream = NULL;
741 enum dshowDeviceType otherDevType = (devtype == VideoDevice) ? AudioDevice : VideoDevice;
743 const wchar_t *filter_name[2] = { L"Audio capture filter", L"Video capture filter" };
746 if ( ((ctx->audio_filter_load_file) && (strlen(ctx->audio_filter_load_file)>0) && (sourcetype == AudioSourceDevice)) ||
747 ((ctx->video_filter_load_file) && (strlen(ctx->video_filter_load_file)>0) && (sourcetype == VideoSourceDevice)) ) {
749 char *filename = NULL;
751 if (sourcetype == AudioSourceDevice)
752 filename = ctx->audio_filter_load_file;
754 filename = ctx->video_filter_load_file;
756 hr = SHCreateStreamOnFile ((LPCSTR) filename, STGM_READ, &ifile_stream);
758 av_log(avctx, AV_LOG_ERROR, "Could not open capture filter description file.\n");
762 hr = OleLoadFromStream(ifile_stream, &IID_IBaseFilter, (void **) &device_filter);
764 av_log(avctx, AV_LOG_ERROR, "Could not load capture filter from file.\n");
768 if (sourcetype == AudioSourceDevice)
769 av_log(avctx, AV_LOG_INFO, "Audio-");
771 av_log(avctx, AV_LOG_INFO, "Video-");
772 av_log(avctx, AV_LOG_INFO, "Capture filter loaded successfully from file \"%s\".\n", filename);
775 if ((r = dshow_cycle_devices(avctx, devenum, devtype, sourcetype, &device_filter, &device_filter_unique_name)) < 0) {
780 if (ctx->device_filter[otherDevType]) {
781 // avoid adding add two instances of the same device to the graph, one for video, one for audio
782 // a few devices don't support this (could also do this check earlier to avoid double crossbars, etc. but they seem OK)
783 if (strcmp(device_filter_unique_name, ctx->device_unique_name[otherDevType]) == 0) {
784 av_log(avctx, AV_LOG_DEBUG, "reusing previous graph capture filter... %s\n", device_filter_unique_name);
785 IBaseFilter_Release(device_filter);
786 device_filter = ctx->device_filter[otherDevType];
787 IBaseFilter_AddRef(ctx->device_filter[otherDevType]);
789 av_log(avctx, AV_LOG_DEBUG, "not reusing previous graph capture filter %s != %s\n", device_filter_unique_name, ctx->device_unique_name[otherDevType]);
793 ctx->device_filter [devtype] = device_filter;
794 ctx->device_unique_name [devtype] = device_filter_unique_name;
796 r = IGraphBuilder_AddFilter(graph, device_filter, NULL);
798 av_log(avctx, AV_LOG_ERROR, "Could not add device filter to graph.\n");
802 if ((r = dshow_cycle_pins(avctx, devtype, sourcetype, device_filter, &device_pin)) < 0) {
807 ctx->device_pin[devtype] = device_pin;
809 capture_filter = libAVFilter_Create(avctx, callback, devtype);
810 if (!capture_filter) {
811 av_log(avctx, AV_LOG_ERROR, "Could not create grabber filter.\n");
814 ctx->capture_filter[devtype] = capture_filter;
816 if ( ((ctx->audio_filter_save_file) && (strlen(ctx->audio_filter_save_file)>0) && (sourcetype == AudioSourceDevice)) ||
817 ((ctx->video_filter_save_file) && (strlen(ctx->video_filter_save_file)>0) && (sourcetype == VideoSourceDevice)) ) {
820 char *filename = NULL;
822 if (sourcetype == AudioSourceDevice)
823 filename = ctx->audio_filter_save_file;
825 filename = ctx->video_filter_save_file;
827 hr = SHCreateStreamOnFile ((LPCSTR) filename, STGM_CREATE | STGM_READWRITE, &ofile_stream);
829 av_log(avctx, AV_LOG_ERROR, "Could not create capture filter description file.\n");
833 hr = IBaseFilter_QueryInterface(device_filter, &IID_IPersistStream, (void **) &pers_stream);
835 av_log(avctx, AV_LOG_ERROR, "Query for IPersistStream failed.\n");
839 hr = OleSaveToStream(pers_stream, ofile_stream);
841 av_log(avctx, AV_LOG_ERROR, "Could not save capture filter \n");
845 hr = IStream_Commit(ofile_stream, STGC_DEFAULT);
847 av_log(avctx, AV_LOG_ERROR, "Could not commit capture filter data to file.\n");
851 if (sourcetype == AudioSourceDevice)
852 av_log(avctx, AV_LOG_INFO, "Audio-");
854 av_log(avctx, AV_LOG_INFO, "Video-");
855 av_log(avctx, AV_LOG_INFO, "Capture filter saved successfully to file \"%s\".\n", filename);
858 r = IGraphBuilder_AddFilter(graph, (IBaseFilter *) capture_filter,
859 filter_name[devtype]);
861 av_log(avctx, AV_LOG_ERROR, "Could not add capture filter to graph\n");
865 libAVPin_AddRef(capture_filter->pin);
866 capture_pin = capture_filter->pin;
867 ctx->capture_pin[devtype] = capture_pin;
869 r = CoCreateInstance(&CLSID_CaptureGraphBuilder2, NULL, CLSCTX_INPROC_SERVER,
870 &IID_ICaptureGraphBuilder2, (void **) &graph_builder2);
872 av_log(avctx, AV_LOG_ERROR, "Could not create CaptureGraphBuilder2\n");
875 ICaptureGraphBuilder2_SetFiltergraph(graph_builder2, graph);
877 av_log(avctx, AV_LOG_ERROR, "Could not set graph for CaptureGraphBuilder2\n");
881 r = ICaptureGraphBuilder2_RenderStream(graph_builder2, NULL, NULL, (IUnknown *) device_pin, NULL /* no intermediate filter */,
882 (IBaseFilter *) capture_filter); /* connect pins, optionally insert intermediate filters like crossbar if necessary */
885 av_log(avctx, AV_LOG_ERROR, "Could not RenderStream to connect pins\n");
889 r = dshow_try_setup_crossbar_options(graph_builder2, device_filter, devtype, avctx);
892 av_log(avctx, AV_LOG_ERROR, "Could not setup CrossBar\n");
899 if (graph_builder2 != NULL)
900 ICaptureGraphBuilder2_Release(graph_builder2);
903 IPersistStream_Release(pers_stream);
906 IStream_Release(ifile_stream);
909 IStream_Release(ofile_stream);
914 static enum AVCodecID waveform_codec_id(enum AVSampleFormat sample_fmt)
916 switch (sample_fmt) {
917 case AV_SAMPLE_FMT_U8: return AV_CODEC_ID_PCM_U8;
918 case AV_SAMPLE_FMT_S16: return AV_CODEC_ID_PCM_S16LE;
919 case AV_SAMPLE_FMT_S32: return AV_CODEC_ID_PCM_S32LE;
920 default: return AV_CODEC_ID_NONE; /* Should never happen. */
924 static enum AVSampleFormat sample_fmt_bits_per_sample(int bits)
927 case 8: return AV_SAMPLE_FMT_U8;
928 case 16: return AV_SAMPLE_FMT_S16;
929 case 32: return AV_SAMPLE_FMT_S32;
930 default: return AV_SAMPLE_FMT_NONE; /* Should never happen. */
935 dshow_add_device(AVFormatContext *avctx,
936 enum dshowDeviceType devtype)
938 struct dshow_ctx *ctx = avctx->priv_data;
940 AVCodecParameters *par;
942 int ret = AVERROR(EIO);
944 st = avformat_new_stream(avctx, NULL);
946 ret = AVERROR(ENOMEM);
951 ctx->capture_filter[devtype]->stream_index = st->index;
953 libAVPin_ConnectionMediaType(ctx->capture_pin[devtype], &type);
956 if (devtype == VideoDevice) {
957 BITMAPINFOHEADER *bih = NULL;
958 AVRational time_base;
960 if (IsEqualGUID(&type.formattype, &FORMAT_VideoInfo)) {
961 VIDEOINFOHEADER *v = (void *) type.pbFormat;
962 time_base = (AVRational) { v->AvgTimePerFrame, 10000000 };
964 } else if (IsEqualGUID(&type.formattype, &FORMAT_VideoInfo2)) {
965 VIDEOINFOHEADER2 *v = (void *) type.pbFormat;
966 time_base = (AVRational) { v->AvgTimePerFrame, 10000000 };
970 av_log(avctx, AV_LOG_ERROR, "Could not get media type.\n");
974 st->avg_frame_rate = av_inv_q(time_base);
975 st->r_frame_rate = av_inv_q(time_base);
977 par->codec_type = AVMEDIA_TYPE_VIDEO;
978 par->width = bih->biWidth;
979 par->height = bih->biHeight;
980 par->codec_tag = bih->biCompression;
981 par->format = dshow_pixfmt(bih->biCompression, bih->biBitCount);
982 if (bih->biCompression == MKTAG('H', 'D', 'Y', 'C')) {
983 av_log(avctx, AV_LOG_DEBUG, "attempt to use full range for HDYC...\n");
984 par->color_range = AVCOL_RANGE_MPEG; // just in case it needs this...
986 if (par->format == AV_PIX_FMT_NONE) {
987 const AVCodecTag *const tags[] = { avformat_get_riff_video_tags(), NULL };
988 par->codec_id = av_codec_get_id(tags, bih->biCompression);
989 if (par->codec_id == AV_CODEC_ID_NONE) {
990 av_log(avctx, AV_LOG_ERROR, "Unknown compression type. "
991 "Please report type 0x%X.\n", (int) bih->biCompression);
992 return AVERROR_PATCHWELCOME;
994 par->bits_per_coded_sample = bih->biBitCount;
996 par->codec_id = AV_CODEC_ID_RAWVIDEO;
997 if (bih->biCompression == BI_RGB || bih->biCompression == BI_BITFIELDS) {
998 par->bits_per_coded_sample = bih->biBitCount;
999 par->extradata = av_malloc(9 + AV_INPUT_BUFFER_PADDING_SIZE);
1000 if (par->extradata) {
1001 par->extradata_size = 9;
1002 memcpy(par->extradata, "BottomUp", 9);
1007 WAVEFORMATEX *fx = NULL;
1009 if (IsEqualGUID(&type.formattype, &FORMAT_WaveFormatEx)) {
1010 fx = (void *) type.pbFormat;
1013 av_log(avctx, AV_LOG_ERROR, "Could not get media type.\n");
1017 par->codec_type = AVMEDIA_TYPE_AUDIO;
1018 par->format = sample_fmt_bits_per_sample(fx->wBitsPerSample);
1019 par->codec_id = waveform_codec_id(par->format);
1020 par->sample_rate = fx->nSamplesPerSec;
1021 par->channels = fx->nChannels;
1024 avpriv_set_pts_info(st, 64, 1, 10000000);
1032 static int parse_device_name(AVFormatContext *avctx)
1034 struct dshow_ctx *ctx = avctx->priv_data;
1035 char **device_name = ctx->device_name;
1036 char *name = av_strdup(avctx->filename);
1041 while ((type = strtok(tmp, "="))) {
1042 char *token = strtok(NULL, ":");
1045 if (!strcmp(type, "video")) {
1046 device_name[0] = token;
1047 } else if (!strcmp(type, "audio")) {
1048 device_name[1] = token;
1050 device_name[0] = NULL;
1051 device_name[1] = NULL;
1056 if (!device_name[0] && !device_name[1]) {
1060 device_name[0] = av_strdup(device_name[0]);
1062 device_name[1] = av_strdup(device_name[1]);
1069 static int dshow_read_header(AVFormatContext *avctx)
1071 struct dshow_ctx *ctx = avctx->priv_data;
1072 IGraphBuilder *graph = NULL;
1073 ICreateDevEnum *devenum = NULL;
1074 IMediaControl *control = NULL;
1075 IMediaEvent *media_event = NULL;
1076 HANDLE media_event_handle;
1078 int ret = AVERROR(EIO);
1083 if (!ctx->list_devices && !parse_device_name(avctx)) {
1084 av_log(avctx, AV_LOG_ERROR, "Malformed dshow input string.\n");
1088 ctx->video_codec_id = avctx->video_codec_id ? avctx->video_codec_id
1089 : AV_CODEC_ID_RAWVIDEO;
1090 if (ctx->pixel_format != AV_PIX_FMT_NONE) {
1091 if (ctx->video_codec_id != AV_CODEC_ID_RAWVIDEO) {
1092 av_log(avctx, AV_LOG_ERROR, "Pixel format may only be set when "
1093 "video codec is not set or set to rawvideo\n");
1094 ret = AVERROR(EINVAL);
1098 if (ctx->framerate) {
1099 r = av_parse_video_rate(&ctx->requested_framerate, ctx->framerate);
1101 av_log(avctx, AV_LOG_ERROR, "Could not parse framerate '%s'.\n", ctx->framerate);
1106 r = CoCreateInstance(&CLSID_FilterGraph, NULL, CLSCTX_INPROC_SERVER,
1107 &IID_IGraphBuilder, (void **) &graph);
1109 av_log(avctx, AV_LOG_ERROR, "Could not create capture graph.\n");
1114 r = CoCreateInstance(&CLSID_SystemDeviceEnum, NULL, CLSCTX_INPROC_SERVER,
1115 &IID_ICreateDevEnum, (void **) &devenum);
1117 av_log(avctx, AV_LOG_ERROR, "Could not enumerate system devices.\n");
1121 if (ctx->list_devices) {
1122 av_log(avctx, AV_LOG_INFO, "DirectShow video devices (some may be both video and audio devices)\n");
1123 dshow_cycle_devices(avctx, devenum, VideoDevice, VideoSourceDevice, NULL, NULL);
1124 av_log(avctx, AV_LOG_INFO, "DirectShow audio devices\n");
1125 dshow_cycle_devices(avctx, devenum, AudioDevice, AudioSourceDevice, NULL, NULL);
1129 if (ctx->list_options) {
1130 if (ctx->device_name[VideoDevice])
1131 if ((r = dshow_list_device_options(avctx, devenum, VideoDevice, VideoSourceDevice))) {
1135 if (ctx->device_name[AudioDevice]) {
1136 if (dshow_list_device_options(avctx, devenum, AudioDevice, AudioSourceDevice)) {
1137 /* show audio options from combined video+audio sources as fallback */
1138 if ((r = dshow_list_device_options(avctx, devenum, AudioDevice, VideoSourceDevice))) {
1145 if (ctx->device_name[VideoDevice]) {
1146 if ((r = dshow_open_device(avctx, devenum, VideoDevice, VideoSourceDevice)) < 0 ||
1147 (r = dshow_add_device(avctx, VideoDevice)) < 0) {
1152 if (ctx->device_name[AudioDevice]) {
1153 if ((r = dshow_open_device(avctx, devenum, AudioDevice, AudioSourceDevice)) < 0 ||
1154 (r = dshow_add_device(avctx, AudioDevice)) < 0) {
1155 av_log(avctx, AV_LOG_INFO, "Searching for audio device within video devices for %s\n", ctx->device_name[AudioDevice]);
1156 /* see if there's a video source with an audio pin with the given audio name */
1157 if ((r = dshow_open_device(avctx, devenum, AudioDevice, VideoSourceDevice)) < 0 ||
1158 (r = dshow_add_device(avctx, AudioDevice)) < 0) {
1164 if (ctx->list_options) {
1165 /* allow it to list crossbar options in dshow_open_device */
1169 ctx->curbufsize[0] = 0;
1170 ctx->curbufsize[1] = 0;
1171 ctx->mutex = CreateMutex(NULL, 0, NULL);
1173 av_log(avctx, AV_LOG_ERROR, "Could not create Mutex\n");
1176 ctx->event[1] = CreateEvent(NULL, 1, 0, NULL);
1177 if (!ctx->event[1]) {
1178 av_log(avctx, AV_LOG_ERROR, "Could not create Event\n");
1182 r = IGraphBuilder_QueryInterface(graph, &IID_IMediaControl, (void **) &control);
1184 av_log(avctx, AV_LOG_ERROR, "Could not get media control.\n");
1187 ctx->control = control;
1189 r = IGraphBuilder_QueryInterface(graph, &IID_IMediaEvent, (void **) &media_event);
1191 av_log(avctx, AV_LOG_ERROR, "Could not get media event.\n");
1194 ctx->media_event = media_event;
1196 r = IMediaEvent_GetEventHandle(media_event, (void *) &media_event_handle);
1198 av_log(avctx, AV_LOG_ERROR, "Could not get media event handle.\n");
1201 proc = GetCurrentProcess();
1202 r = DuplicateHandle(proc, media_event_handle, proc, &ctx->event[0],
1203 0, 0, DUPLICATE_SAME_ACCESS);
1205 av_log(avctx, AV_LOG_ERROR, "Could not duplicate media event handle.\n");
1209 r = IMediaControl_Run(control);
1212 r = IMediaControl_GetState(control, 0, &pfs);
1215 av_log(avctx, AV_LOG_ERROR, "Could not run graph (sometimes caused by a device already in use by other application)\n");
1224 ICreateDevEnum_Release(devenum);
1227 dshow_read_close(avctx);
1233 * Checks media events from DirectShow and returns -1 on error or EOF. Also
1234 * purges all events that might be in the event queue to stop the trigger
1235 * of event notification.
1237 static int dshow_check_event_queue(IMediaEvent *media_event)
1243 while (IMediaEvent_GetEvent(media_event, &code, &p1, &p2, 0) != E_ABORT) {
1244 if (code == EC_COMPLETE || code == EC_DEVICE_LOST || code == EC_ERRORABORT)
1246 IMediaEvent_FreeEventParams(media_event, code, p1, p2);
1252 static int dshow_read_packet(AVFormatContext *s, AVPacket *pkt)
1254 struct dshow_ctx *ctx = s->priv_data;
1255 AVPacketList *pktl = NULL;
1257 while (!ctx->eof && !pktl) {
1258 WaitForSingleObject(ctx->mutex, INFINITE);
1262 ctx->pktl = ctx->pktl->next;
1264 ctx->curbufsize[pkt->stream_index] -= pkt->size;
1266 ResetEvent(ctx->event[1]);
1267 ReleaseMutex(ctx->mutex);
1269 if (dshow_check_event_queue(ctx->media_event) < 0) {
1271 } else if (s->flags & AVFMT_FLAG_NONBLOCK) {
1272 return AVERROR(EAGAIN);
1274 WaitForMultipleObjects(2, ctx->event, 0, INFINITE);
1279 return ctx->eof ? AVERROR(EIO) : pkt->size;
1282 #define OFFSET(x) offsetof(struct dshow_ctx, x)
1283 #define DEC AV_OPT_FLAG_DECODING_PARAM
1284 static const AVOption options[] = {
1285 { "video_size", "set video size given a string such as 640x480 or hd720.", OFFSET(requested_width), AV_OPT_TYPE_IMAGE_SIZE, {.str = NULL}, 0, 0, DEC },
1286 { "pixel_format", "set video pixel format", OFFSET(pixel_format), AV_OPT_TYPE_PIXEL_FMT, {.i64 = AV_PIX_FMT_NONE}, -1, INT_MAX, DEC },
1287 { "framerate", "set video frame rate", OFFSET(framerate), AV_OPT_TYPE_STRING, {.str = NULL}, 0, 0, DEC },
1288 { "sample_rate", "set audio sample rate", OFFSET(sample_rate), AV_OPT_TYPE_INT, {.i64 = 0}, 0, INT_MAX, DEC },
1289 { "sample_size", "set audio sample size", OFFSET(sample_size), AV_OPT_TYPE_INT, {.i64 = 0}, 0, 16, DEC },
1290 { "channels", "set number of audio channels, such as 1 or 2", OFFSET(channels), AV_OPT_TYPE_INT, {.i64 = 0}, 0, INT_MAX, DEC },
1291 { "audio_buffer_size", "set audio device buffer latency size in milliseconds (default is the device's default)", OFFSET(audio_buffer_size), AV_OPT_TYPE_INT, {.i64 = 0}, 0, INT_MAX, DEC },
1292 { "list_devices", "list available devices", OFFSET(list_devices), AV_OPT_TYPE_BOOL, {.i64=0}, 0, 1, DEC },
1293 { "list_options", "list available options for specified device", OFFSET(list_options), AV_OPT_TYPE_BOOL, {.i64=0}, 0, 1, DEC },
1294 { "video_device_number", "set video device number for devices with same name (starts at 0)", OFFSET(video_device_number), AV_OPT_TYPE_INT, {.i64 = 0}, 0, INT_MAX, DEC },
1295 { "audio_device_number", "set audio device number for devices with same name (starts at 0)", OFFSET(audio_device_number), AV_OPT_TYPE_INT, {.i64 = 0}, 0, INT_MAX, DEC },
1296 { "video_pin_name", "select video capture pin by name", OFFSET(video_pin_name),AV_OPT_TYPE_STRING, {.str = NULL}, 0, 0, AV_OPT_FLAG_ENCODING_PARAM },
1297 { "audio_pin_name", "select audio capture pin by name", OFFSET(audio_pin_name),AV_OPT_TYPE_STRING, {.str = NULL}, 0, 0, AV_OPT_FLAG_ENCODING_PARAM },
1298 { "crossbar_video_input_pin_number", "set video input pin number for crossbar device", OFFSET(crossbar_video_input_pin_number), AV_OPT_TYPE_INT, {.i64 = -1}, -1, INT_MAX, DEC },
1299 { "crossbar_audio_input_pin_number", "set audio input pin number for crossbar device", OFFSET(crossbar_audio_input_pin_number), AV_OPT_TYPE_INT, {.i64 = -1}, -1, INT_MAX, DEC },
1300 { "show_video_device_dialog", "display property dialog for video capture device", OFFSET(show_video_device_dialog), AV_OPT_TYPE_BOOL, {.i64 = 0}, 0, 1, DEC },
1301 { "show_audio_device_dialog", "display property dialog for audio capture device", OFFSET(show_audio_device_dialog), AV_OPT_TYPE_BOOL, {.i64 = 0}, 0, 1, DEC },
1302 { "show_video_crossbar_connection_dialog", "display property dialog for crossbar connecting pins filter on video device", OFFSET(show_video_crossbar_connection_dialog), AV_OPT_TYPE_BOOL, {.i64 = 0}, 0, 1, DEC },
1303 { "show_audio_crossbar_connection_dialog", "display property dialog for crossbar connecting pins filter on audio device", OFFSET(show_audio_crossbar_connection_dialog), AV_OPT_TYPE_BOOL, {.i64 = 0}, 0, 1, DEC },
1304 { "show_analog_tv_tuner_dialog", "display property dialog for analog tuner filter", OFFSET(show_analog_tv_tuner_dialog), AV_OPT_TYPE_BOOL, {.i64 = 0}, 0, 1, DEC },
1305 { "show_analog_tv_tuner_audio_dialog", "display property dialog for analog tuner audio filter", OFFSET(show_analog_tv_tuner_audio_dialog), AV_OPT_TYPE_BOOL, {.i64 = 0}, 0, 1, DEC },
1306 { "audio_device_load", "load audio capture filter device (and properties) from file", OFFSET(audio_filter_load_file), AV_OPT_TYPE_STRING, {.str = NULL}, 0, 0, DEC },
1307 { "audio_device_save", "save audio capture filter device (and properties) to file", OFFSET(audio_filter_save_file), AV_OPT_TYPE_STRING, {.str = NULL}, 0, 0, DEC },
1308 { "video_device_load", "load video capture filter device (and properties) from file", OFFSET(video_filter_load_file), AV_OPT_TYPE_STRING, {.str = NULL}, 0, 0, DEC },
1309 { "video_device_save", "save video capture filter device (and properties) to file", OFFSET(video_filter_save_file), AV_OPT_TYPE_STRING, {.str = NULL}, 0, 0, DEC },
1313 static const AVClass dshow_class = {
1314 .class_name = "dshow indev",
1315 .item_name = av_default_item_name,
1317 .version = LIBAVUTIL_VERSION_INT,
1318 .category = AV_CLASS_CATEGORY_DEVICE_VIDEO_INPUT,
1321 AVInputFormat ff_dshow_demuxer = {
1323 .long_name = NULL_IF_CONFIG_SMALL("DirectShow capture"),
1324 .priv_data_size = sizeof(struct dshow_ctx),
1325 .read_header = dshow_read_header,
1326 .read_packet = dshow_read_packet,
1327 .read_close = dshow_read_close,
1328 .flags = AVFMT_NOFILE,
1329 .priv_class = &dshow_class,