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 if (ctx->device_name[0])
108 av_freep(&ctx->device_name[0]);
109 if (ctx->device_name[1])
110 av_freep(&ctx->device_name[1]);
111 if (ctx->device_unique_name[0])
112 av_freep(&ctx->device_unique_name[0]);
113 if (ctx->device_unique_name[1])
114 av_freep(&ctx->device_unique_name[1]);
117 CloseHandle(ctx->mutex);
119 CloseHandle(ctx->event[0]);
121 CloseHandle(ctx->event[1]);
125 AVPacketList *next = pktl->next;
126 av_packet_unref(&pktl->pkt);
136 static char *dup_wchar_to_utf8(wchar_t *w)
139 int l = WideCharToMultiByte(CP_UTF8, 0, w, -1, 0, 0, 0, 0);
142 WideCharToMultiByte(CP_UTF8, 0, w, -1, s, l, 0, 0);
146 static int shall_we_drop(AVFormatContext *s, int index, enum dshowDeviceType devtype)
148 struct dshow_ctx *ctx = s->priv_data;
149 static const uint8_t dropscore[] = {62, 75, 87, 100};
150 const int ndropscores = FF_ARRAY_ELEMS(dropscore);
151 unsigned int buffer_fullness = (ctx->curbufsize[index]*100)/s->max_picture_buffer;
152 const char *devtypename = (devtype == VideoDevice) ? "video" : "audio";
154 if(dropscore[++ctx->video_frame_num%ndropscores] <= buffer_fullness) {
155 av_log(s, AV_LOG_ERROR,
156 "real-time buffer [%s] [%s input] too full or near too full (%d%% of size: %d [rtbufsize parameter])! frame dropped!\n",
157 ctx->device_name[devtype], devtypename, buffer_fullness, s->max_picture_buffer);
165 callback(void *priv_data, int index, uint8_t *buf, int buf_size, int64_t time, enum dshowDeviceType devtype)
167 AVFormatContext *s = priv_data;
168 struct dshow_ctx *ctx = s->priv_data;
169 AVPacketList **ppktl, *pktl_next;
171 // dump_videohdr(s, vdhdr);
173 WaitForSingleObject(ctx->mutex, INFINITE);
175 if(shall_we_drop(s, index, devtype))
178 pktl_next = av_mallocz(sizeof(AVPacketList));
182 if(av_new_packet(&pktl_next->pkt, buf_size) < 0) {
187 pktl_next->pkt.stream_index = index;
188 pktl_next->pkt.pts = time;
189 memcpy(pktl_next->pkt.data, buf, buf_size);
191 for(ppktl = &ctx->pktl ; *ppktl ; ppktl = &(*ppktl)->next);
193 ctx->curbufsize[index] += buf_size;
195 SetEvent(ctx->event[1]);
196 ReleaseMutex(ctx->mutex);
200 ReleaseMutex(ctx->mutex);
205 * Cycle through available devices using the device enumerator devenum,
206 * retrieve the device with type specified by devtype and return the
207 * pointer to the object found in *pfilter.
208 * If pfilter is NULL, list all device names.
211 dshow_cycle_devices(AVFormatContext *avctx, ICreateDevEnum *devenum,
212 enum dshowDeviceType devtype, enum dshowSourceFilterType sourcetype,
213 IBaseFilter **pfilter, char **device_unique_name)
215 struct dshow_ctx *ctx = avctx->priv_data;
216 IBaseFilter *device_filter = NULL;
217 IEnumMoniker *classenum = NULL;
219 const char *device_name = ctx->device_name[devtype];
220 int skip = (devtype == VideoDevice) ? ctx->video_device_number
221 : ctx->audio_device_number;
224 const GUID *device_guid[2] = { &CLSID_VideoInputDeviceCategory,
225 &CLSID_AudioInputDeviceCategory };
226 const char *devtypename = (devtype == VideoDevice) ? "video" : "audio only";
227 const char *sourcetypename = (sourcetype == VideoSourceDevice) ? "video" : "audio";
229 r = ICreateDevEnum_CreateClassEnumerator(devenum, device_guid[sourcetype],
230 (IEnumMoniker **) &classenum, 0);
232 av_log(avctx, AV_LOG_ERROR, "Could not enumerate %s devices (or none found).\n",
237 while (!device_filter && IEnumMoniker_Next(classenum, 1, &m, NULL) == S_OK) {
238 IPropertyBag *bag = NULL;
239 char *friendly_name = NULL;
240 char *unique_name = NULL;
242 IBindCtx *bind_ctx = NULL;
243 LPOLESTR olestr = NULL;
244 LPMALLOC co_malloc = NULL;
247 r = CoGetMalloc(1, &co_malloc);
250 r = CreateBindCtx(0, &bind_ctx);
253 /* GetDisplayname works for both video and audio, DevicePath doesn't */
254 r = IMoniker_GetDisplayName(m, bind_ctx, NULL, &olestr);
257 unique_name = dup_wchar_to_utf8(olestr);
258 /* replace ':' with '_' since we use : to delineate between sources */
259 for (i = 0; i < strlen(unique_name); i++) {
260 if (unique_name[i] == ':')
261 unique_name[i] = '_';
264 r = IMoniker_BindToStorage(m, 0, 0, &IID_IPropertyBag, (void *) &bag);
269 r = IPropertyBag_Read(bag, L"FriendlyName", &var, NULL);
272 friendly_name = dup_wchar_to_utf8(var.bstrVal);
275 if (strcmp(device_name, friendly_name) && strcmp(device_name, unique_name))
279 r = IMoniker_BindToObject(m, 0, 0, &IID_IBaseFilter, (void *) &device_filter);
281 av_log(avctx, AV_LOG_ERROR, "Unable to BindToObject for %s\n", device_name);
284 *device_unique_name = unique_name;
285 // success, loop will end now
288 av_log(avctx, AV_LOG_INFO, " \"%s\"\n", friendly_name);
289 av_log(avctx, AV_LOG_INFO, " Alternative name \"%s\"\n", unique_name);
290 av_free(unique_name);
294 if (olestr && co_malloc)
295 IMalloc_Free(co_malloc, olestr);
297 IBindCtx_Release(bind_ctx);
298 av_free(friendly_name);
300 IPropertyBag_Release(bag);
304 IEnumMoniker_Release(classenum);
307 if (!device_filter) {
308 av_log(avctx, AV_LOG_ERROR, "Could not find %s device with name [%s] among source devices of type %s.\n",
309 devtypename, device_name, sourcetypename);
312 *pfilter = device_filter;
319 * Cycle through available formats using the specified pin,
320 * try to set parameters specified through AVOptions and if successful
321 * return 1 in *pformat_set.
322 * If pformat_set is NULL, list all pin capabilities.
325 dshow_cycle_formats(AVFormatContext *avctx, enum dshowDeviceType devtype,
326 IPin *pin, int *pformat_set)
328 struct dshow_ctx *ctx = avctx->priv_data;
329 IAMStreamConfig *config = NULL;
330 AM_MEDIA_TYPE *type = NULL;
335 if (IPin_QueryInterface(pin, &IID_IAMStreamConfig, (void **) &config) != S_OK)
337 if (IAMStreamConfig_GetNumberOfCapabilities(config, &n, &size) != S_OK)
340 caps = av_malloc(size);
344 for (i = 0; i < n && !format_set; i++) {
345 r = IAMStreamConfig_GetStreamCaps(config, i, &type, (void *) caps);
349 ff_print_AM_MEDIA_TYPE(type);
352 if (devtype == VideoDevice) {
353 VIDEO_STREAM_CONFIG_CAPS *vcaps = caps;
354 BITMAPINFOHEADER *bih;
356 const AVCodecTag *const tags[] = { avformat_get_riff_video_tags(), NULL };
358 ff_print_VIDEO_STREAM_CONFIG_CAPS(vcaps);
360 if (IsEqualGUID(&type->formattype, &FORMAT_VideoInfo)) {
361 VIDEOINFOHEADER *v = (void *) type->pbFormat;
362 fr = &v->AvgTimePerFrame;
364 } else if (IsEqualGUID(&type->formattype, &FORMAT_VideoInfo2)) {
365 VIDEOINFOHEADER2 *v = (void *) type->pbFormat;
366 fr = &v->AvgTimePerFrame;
372 enum AVPixelFormat pix_fmt = dshow_pixfmt(bih->biCompression, bih->biBitCount);
373 if (pix_fmt == AV_PIX_FMT_NONE) {
374 enum AVCodecID codec_id = av_codec_get_id(tags, bih->biCompression);
375 AVCodec *codec = avcodec_find_decoder(codec_id);
376 if (codec_id == AV_CODEC_ID_NONE || !codec) {
377 av_log(avctx, AV_LOG_INFO, " unknown compression type 0x%X", (int) bih->biCompression);
379 av_log(avctx, AV_LOG_INFO, " vcodec=%s", codec->name);
382 av_log(avctx, AV_LOG_INFO, " pixel_format=%s", av_get_pix_fmt_name(pix_fmt));
384 av_log(avctx, AV_LOG_INFO, " min s=%ldx%ld fps=%g max s=%ldx%ld fps=%g\n",
385 vcaps->MinOutputSize.cx, vcaps->MinOutputSize.cy,
386 1e7 / vcaps->MaxFrameInterval,
387 vcaps->MaxOutputSize.cx, vcaps->MaxOutputSize.cy,
388 1e7 / vcaps->MinFrameInterval);
391 if (ctx->video_codec_id != AV_CODEC_ID_RAWVIDEO) {
392 if (ctx->video_codec_id != av_codec_get_id(tags, bih->biCompression))
395 if (ctx->pixel_format != AV_PIX_FMT_NONE &&
396 ctx->pixel_format != dshow_pixfmt(bih->biCompression, bih->biBitCount)) {
399 if (ctx->framerate) {
400 int64_t framerate = ((int64_t) ctx->requested_framerate.den*10000000)
401 / ctx->requested_framerate.num;
402 if (framerate > vcaps->MaxFrameInterval ||
403 framerate < vcaps->MinFrameInterval)
407 if (ctx->requested_width && ctx->requested_height) {
408 if (ctx->requested_width > vcaps->MaxOutputSize.cx ||
409 ctx->requested_width < vcaps->MinOutputSize.cx ||
410 ctx->requested_height > vcaps->MaxOutputSize.cy ||
411 ctx->requested_height < vcaps->MinOutputSize.cy)
413 bih->biWidth = ctx->requested_width;
414 bih->biHeight = ctx->requested_height;
417 AUDIO_STREAM_CONFIG_CAPS *acaps = caps;
420 ff_print_AUDIO_STREAM_CONFIG_CAPS(acaps);
422 if (IsEqualGUID(&type->formattype, &FORMAT_WaveFormatEx)) {
423 fx = (void *) type->pbFormat;
428 av_log(avctx, AV_LOG_INFO, " min ch=%lu bits=%lu rate=%6lu max ch=%lu bits=%lu rate=%6lu\n",
429 acaps->MinimumChannels, acaps->MinimumBitsPerSample, acaps->MinimumSampleFrequency,
430 acaps->MaximumChannels, acaps->MaximumBitsPerSample, acaps->MaximumSampleFrequency);
433 if (ctx->sample_rate) {
434 if (ctx->sample_rate > acaps->MaximumSampleFrequency ||
435 ctx->sample_rate < acaps->MinimumSampleFrequency)
437 fx->nSamplesPerSec = ctx->sample_rate;
439 if (ctx->sample_size) {
440 if (ctx->sample_size > acaps->MaximumBitsPerSample ||
441 ctx->sample_size < acaps->MinimumBitsPerSample)
443 fx->wBitsPerSample = ctx->sample_size;
446 if (ctx->channels > acaps->MaximumChannels ||
447 ctx->channels < acaps->MinimumChannels)
449 fx->nChannels = ctx->channels;
452 if (IAMStreamConfig_SetFormat(config, type) != S_OK)
457 CoTaskMemFree(type->pbFormat);
461 IAMStreamConfig_Release(config);
464 *pformat_set = format_set;
468 * Set audio device buffer size in milliseconds (which can directly impact
469 * latency, depending on the device).
472 dshow_set_audio_buffer_size(AVFormatContext *avctx, IPin *pin)
474 struct dshow_ctx *ctx = avctx->priv_data;
475 IAMBufferNegotiation *buffer_negotiation = NULL;
476 ALLOCATOR_PROPERTIES props = { -1, -1, -1, -1 };
477 IAMStreamConfig *config = NULL;
478 AM_MEDIA_TYPE *type = NULL;
479 int ret = AVERROR(EIO);
481 if (IPin_QueryInterface(pin, &IID_IAMStreamConfig, (void **) &config) != S_OK)
483 if (IAMStreamConfig_GetFormat(config, &type) != S_OK)
485 if (!IsEqualGUID(&type->formattype, &FORMAT_WaveFormatEx))
488 props.cbBuffer = (((WAVEFORMATEX *) type->pbFormat)->nAvgBytesPerSec)
489 * ctx->audio_buffer_size / 1000;
491 if (IPin_QueryInterface(pin, &IID_IAMBufferNegotiation, (void **) &buffer_negotiation) != S_OK)
493 if (IAMBufferNegotiation_SuggestAllocatorProperties(buffer_negotiation, &props) != S_OK)
499 if (buffer_negotiation)
500 IAMBufferNegotiation_Release(buffer_negotiation);
503 CoTaskMemFree(type->pbFormat);
507 IAMStreamConfig_Release(config);
513 * Pops up a user dialog allowing them to adjust properties for the given filter, if possible.
516 dshow_show_filter_properties(IBaseFilter *device_filter, AVFormatContext *avctx) {
517 ISpecifyPropertyPages *property_pages = NULL;
518 IUnknown *device_filter_iunknown = NULL;
520 FILTER_INFO filter_info = {0}; /* a warning on this line is false positive GCC bug 53119 AFAICT */
521 CAUUID ca_guid = {0};
523 hr = IBaseFilter_QueryInterface(device_filter, &IID_ISpecifyPropertyPages, (void **)&property_pages);
525 av_log(avctx, AV_LOG_WARNING, "requested filter does not have a property page to show");
528 hr = IBaseFilter_QueryFilterInfo(device_filter, &filter_info);
532 hr = IBaseFilter_QueryInterface(device_filter, &IID_IUnknown, (void **)&device_filter_iunknown);
536 hr = ISpecifyPropertyPages_GetPages(property_pages, &ca_guid);
540 hr = OleCreatePropertyFrame(NULL, 0, 0, filter_info.achName, 1, &device_filter_iunknown, ca_guid.cElems,
541 ca_guid.pElems, 0, 0, NULL);
547 av_log(avctx, AV_LOG_ERROR, "Failure showing property pages for filter");
550 ISpecifyPropertyPages_Release(property_pages);
551 if (device_filter_iunknown)
552 IUnknown_Release(device_filter_iunknown);
553 if (filter_info.pGraph)
554 IFilterGraph_Release(filter_info.pGraph);
556 CoTaskMemFree(ca_guid.pElems);
560 * Cycle through available pins using the device_filter device, of type
561 * devtype, retrieve the first output pin and return the pointer to the
562 * object found in *ppin.
563 * If ppin is NULL, cycle through all pins listing audio/video capabilities.
566 dshow_cycle_pins(AVFormatContext *avctx, enum dshowDeviceType devtype,
567 enum dshowSourceFilterType sourcetype, IBaseFilter *device_filter, IPin **ppin)
569 struct dshow_ctx *ctx = avctx->priv_data;
571 IPin *device_pin = NULL;
575 const GUID *mediatype[2] = { &MEDIATYPE_Video, &MEDIATYPE_Audio };
576 const char *devtypename = (devtype == VideoDevice) ? "video" : "audio only";
577 const char *sourcetypename = (sourcetype == VideoSourceDevice) ? "video" : "audio";
579 int set_format = (devtype == VideoDevice && (ctx->framerate ||
580 (ctx->requested_width && ctx->requested_height) ||
581 ctx->pixel_format != AV_PIX_FMT_NONE ||
582 ctx->video_codec_id != AV_CODEC_ID_RAWVIDEO))
583 || (devtype == AudioDevice && (ctx->channels || ctx->sample_rate));
585 int should_show_properties = (devtype == VideoDevice) ? ctx->show_video_device_dialog : ctx->show_audio_device_dialog;
587 if (should_show_properties)
588 dshow_show_filter_properties(device_filter, avctx);
590 r = IBaseFilter_EnumPins(device_filter, &pins);
592 av_log(avctx, AV_LOG_ERROR, "Could not enumerate pins.\n");
597 av_log(avctx, AV_LOG_INFO, "DirectShow %s device options (from %s devices)\n",
598 devtypename, sourcetypename);
601 while (!device_pin && IEnumPins_Next(pins, 1, &pin, NULL) == S_OK) {
602 IKsPropertySet *p = NULL;
603 IEnumMediaTypes *types = NULL;
608 char *name_buf = NULL;
609 wchar_t *pin_id = NULL;
610 char *pin_buf = NULL;
611 char *desired_pin_name = devtype == VideoDevice ? ctx->video_pin_name : ctx->audio_pin_name;
613 IPin_QueryPinInfo(pin, &info);
614 IBaseFilter_Release(info.pFilter);
616 if (info.dir != PINDIR_OUTPUT)
618 if (IPin_QueryInterface(pin, &IID_IKsPropertySet, (void **) &p) != S_OK)
620 if (IKsPropertySet_Get(p, &ROPSETID_Pin, AMPROPERTY_PIN_CATEGORY,
621 NULL, 0, &category, sizeof(GUID), &r2) != S_OK)
623 if (!IsEqualGUID(&category, &PIN_CATEGORY_CAPTURE))
625 name_buf = dup_wchar_to_utf8(info.achName);
627 r = IPin_QueryId(pin, &pin_id);
629 av_log(avctx, AV_LOG_ERROR, "Could not query pin id\n");
632 pin_buf = dup_wchar_to_utf8(pin_id);
635 av_log(avctx, AV_LOG_INFO, " Pin \"%s\" (alternative pin name \"%s\")\n", name_buf, pin_buf);
636 dshow_cycle_formats(avctx, devtype, pin, NULL);
640 if (desired_pin_name) {
641 if(strcmp(name_buf, desired_pin_name) && strcmp(pin_buf, desired_pin_name)) {
642 av_log(avctx, AV_LOG_DEBUG, "skipping pin \"%s\" (\"%s\") != requested \"%s\"\n",
643 name_buf, pin_buf, desired_pin_name);
649 dshow_cycle_formats(avctx, devtype, pin, &format_set);
654 if (devtype == AudioDevice && ctx->audio_buffer_size) {
655 if (dshow_set_audio_buffer_size(avctx, pin) < 0) {
656 av_log(avctx, AV_LOG_ERROR, "unable to set audio buffer size %d to pin, using pin anyway...", ctx->audio_buffer_size);
660 if (IPin_EnumMediaTypes(pin, &types) != S_OK)
663 IEnumMediaTypes_Reset(types);
664 /* in case format_set was not called, just verify the majortype */
665 while (!device_pin && IEnumMediaTypes_Next(types, 1, &type, NULL) == S_OK) {
666 if (IsEqualGUID(&type->majortype, mediatype[devtype])) {
668 av_log(avctx, AV_LOG_DEBUG, "Selecting pin %s on %s\n", name_buf, devtypename);
676 IEnumMediaTypes_Release(types);
678 IKsPropertySet_Release(p);
679 if (device_pin != pin)
684 CoTaskMemFree(pin_id);
687 IEnumPins_Release(pins);
690 if (set_format && !format_set) {
691 av_log(avctx, AV_LOG_ERROR, "Could not set %s options\n", devtypename);
695 av_log(avctx, AV_LOG_ERROR,
696 "Could not find output pin from %s capture device.\n", devtypename);
706 * List options for device with type devtype, source filter type sourcetype
708 * @param devenum device enumerator used for accessing the device
711 dshow_list_device_options(AVFormatContext *avctx, ICreateDevEnum *devenum,
712 enum dshowDeviceType devtype, enum dshowSourceFilterType sourcetype)
714 struct dshow_ctx *ctx = avctx->priv_data;
715 IBaseFilter *device_filter = NULL;
716 char *device_unique_name = NULL;
719 if ((r = dshow_cycle_devices(avctx, devenum, devtype, sourcetype, &device_filter, &device_unique_name)) < 0)
721 ctx->device_filter[devtype] = device_filter;
722 if ((r = dshow_cycle_pins(avctx, devtype, sourcetype, device_filter, NULL)) < 0)
724 av_freep(&device_unique_name);
729 dshow_open_device(AVFormatContext *avctx, ICreateDevEnum *devenum,
730 enum dshowDeviceType devtype, enum dshowSourceFilterType sourcetype)
732 struct dshow_ctx *ctx = avctx->priv_data;
733 IBaseFilter *device_filter = NULL;
734 char *device_filter_unique_name = NULL;
735 IGraphBuilder *graph = ctx->graph;
736 IPin *device_pin = NULL;
737 libAVPin *capture_pin = NULL;
738 libAVFilter *capture_filter = NULL;
739 ICaptureGraphBuilder2 *graph_builder2 = NULL;
740 int ret = AVERROR(EIO);
742 IStream *ifile_stream = NULL;
743 IStream *ofile_stream = NULL;
744 IPersistStream *pers_stream = NULL;
745 enum dshowDeviceType otherDevType = (devtype == VideoDevice) ? AudioDevice : VideoDevice;
747 const wchar_t *filter_name[2] = { L"Audio capture filter", L"Video capture filter" };
750 if ( ((ctx->audio_filter_load_file) && (strlen(ctx->audio_filter_load_file)>0) && (sourcetype == AudioSourceDevice)) ||
751 ((ctx->video_filter_load_file) && (strlen(ctx->video_filter_load_file)>0) && (sourcetype == VideoSourceDevice)) ) {
753 char *filename = NULL;
755 if (sourcetype == AudioSourceDevice)
756 filename = ctx->audio_filter_load_file;
758 filename = ctx->video_filter_load_file;
760 hr = SHCreateStreamOnFile ((LPCSTR) filename, STGM_READ, &ifile_stream);
762 av_log(avctx, AV_LOG_ERROR, "Could not open capture filter description file.\n");
766 hr = OleLoadFromStream(ifile_stream, &IID_IBaseFilter, (void **) &device_filter);
768 av_log(avctx, AV_LOG_ERROR, "Could not load capture filter from file.\n");
772 if (sourcetype == AudioSourceDevice)
773 av_log(avctx, AV_LOG_INFO, "Audio-");
775 av_log(avctx, AV_LOG_INFO, "Video-");
776 av_log(avctx, AV_LOG_INFO, "Capture filter loaded successfully from file \"%s\".\n", filename);
779 if ((r = dshow_cycle_devices(avctx, devenum, devtype, sourcetype, &device_filter, &device_filter_unique_name)) < 0) {
784 if (ctx->device_filter[otherDevType]) {
785 // avoid adding add two instances of the same device to the graph, one for video, one for audio
786 // a few devices don't support this (could also do this check earlier to avoid double crossbars, etc. but they seem OK)
787 if (strcmp(device_filter_unique_name, ctx->device_unique_name[otherDevType]) == 0) {
788 av_log(avctx, AV_LOG_DEBUG, "reusing previous graph capture filter... %s\n", device_filter_unique_name);
789 IBaseFilter_Release(device_filter);
790 device_filter = ctx->device_filter[otherDevType];
791 IBaseFilter_AddRef(ctx->device_filter[otherDevType]);
793 av_log(avctx, AV_LOG_DEBUG, "not reusing previous graph capture filter %s != %s\n", device_filter_unique_name, ctx->device_unique_name[otherDevType]);
797 ctx->device_filter [devtype] = device_filter;
798 ctx->device_unique_name [devtype] = device_filter_unique_name;
800 r = IGraphBuilder_AddFilter(graph, device_filter, NULL);
802 av_log(avctx, AV_LOG_ERROR, "Could not add device filter to graph.\n");
806 if ((r = dshow_cycle_pins(avctx, devtype, sourcetype, device_filter, &device_pin)) < 0) {
811 ctx->device_pin[devtype] = device_pin;
813 capture_filter = libAVFilter_Create(avctx, callback, devtype);
814 if (!capture_filter) {
815 av_log(avctx, AV_LOG_ERROR, "Could not create grabber filter.\n");
818 ctx->capture_filter[devtype] = capture_filter;
820 if ( ((ctx->audio_filter_save_file) && (strlen(ctx->audio_filter_save_file)>0) && (sourcetype == AudioSourceDevice)) ||
821 ((ctx->video_filter_save_file) && (strlen(ctx->video_filter_save_file)>0) && (sourcetype == VideoSourceDevice)) ) {
824 char *filename = NULL;
826 if (sourcetype == AudioSourceDevice)
827 filename = ctx->audio_filter_save_file;
829 filename = ctx->video_filter_save_file;
831 hr = SHCreateStreamOnFile ((LPCSTR) filename, STGM_CREATE | STGM_READWRITE, &ofile_stream);
833 av_log(avctx, AV_LOG_ERROR, "Could not create capture filter description file.\n");
837 hr = IBaseFilter_QueryInterface(device_filter, &IID_IPersistStream, (void **) &pers_stream);
839 av_log(avctx, AV_LOG_ERROR, "Query for IPersistStream failed.\n");
843 hr = OleSaveToStream(pers_stream, ofile_stream);
845 av_log(avctx, AV_LOG_ERROR, "Could not save capture filter \n");
849 hr = IStream_Commit(ofile_stream, STGC_DEFAULT);
851 av_log(avctx, AV_LOG_ERROR, "Could not commit capture filter data to file.\n");
855 if (sourcetype == AudioSourceDevice)
856 av_log(avctx, AV_LOG_INFO, "Audio-");
858 av_log(avctx, AV_LOG_INFO, "Video-");
859 av_log(avctx, AV_LOG_INFO, "Capture filter saved successfully to file \"%s\".\n", filename);
862 r = IGraphBuilder_AddFilter(graph, (IBaseFilter *) capture_filter,
863 filter_name[devtype]);
865 av_log(avctx, AV_LOG_ERROR, "Could not add capture filter to graph\n");
869 libAVPin_AddRef(capture_filter->pin);
870 capture_pin = capture_filter->pin;
871 ctx->capture_pin[devtype] = capture_pin;
873 r = CoCreateInstance(&CLSID_CaptureGraphBuilder2, NULL, CLSCTX_INPROC_SERVER,
874 &IID_ICaptureGraphBuilder2, (void **) &graph_builder2);
876 av_log(avctx, AV_LOG_ERROR, "Could not create CaptureGraphBuilder2\n");
879 ICaptureGraphBuilder2_SetFiltergraph(graph_builder2, graph);
881 av_log(avctx, AV_LOG_ERROR, "Could not set graph for CaptureGraphBuilder2\n");
885 r = ICaptureGraphBuilder2_RenderStream(graph_builder2, NULL, NULL, (IUnknown *) device_pin, NULL /* no intermediate filter */,
886 (IBaseFilter *) capture_filter); /* connect pins, optionally insert intermediate filters like crossbar if necessary */
889 av_log(avctx, AV_LOG_ERROR, "Could not RenderStream to connect pins\n");
893 r = dshow_try_setup_crossbar_options(graph_builder2, device_filter, devtype, avctx);
896 av_log(avctx, AV_LOG_ERROR, "Could not setup CrossBar\n");
903 if (graph_builder2 != NULL)
904 ICaptureGraphBuilder2_Release(graph_builder2);
907 IPersistStream_Release(pers_stream);
910 IStream_Release(ifile_stream);
913 IStream_Release(ofile_stream);
918 static enum AVCodecID waveform_codec_id(enum AVSampleFormat sample_fmt)
920 switch (sample_fmt) {
921 case AV_SAMPLE_FMT_U8: return AV_CODEC_ID_PCM_U8;
922 case AV_SAMPLE_FMT_S16: return AV_CODEC_ID_PCM_S16LE;
923 case AV_SAMPLE_FMT_S32: return AV_CODEC_ID_PCM_S32LE;
924 default: return AV_CODEC_ID_NONE; /* Should never happen. */
928 static enum AVSampleFormat sample_fmt_bits_per_sample(int bits)
931 case 8: return AV_SAMPLE_FMT_U8;
932 case 16: return AV_SAMPLE_FMT_S16;
933 case 32: return AV_SAMPLE_FMT_S32;
934 default: return AV_SAMPLE_FMT_NONE; /* Should never happen. */
939 dshow_add_device(AVFormatContext *avctx,
940 enum dshowDeviceType devtype)
942 struct dshow_ctx *ctx = avctx->priv_data;
944 AVCodecParameters *par;
946 int ret = AVERROR(EIO);
948 st = avformat_new_stream(avctx, NULL);
950 ret = AVERROR(ENOMEM);
955 ctx->capture_filter[devtype]->stream_index = st->index;
957 libAVPin_ConnectionMediaType(ctx->capture_pin[devtype], &type);
960 if (devtype == VideoDevice) {
961 BITMAPINFOHEADER *bih = NULL;
962 AVRational time_base;
964 if (IsEqualGUID(&type.formattype, &FORMAT_VideoInfo)) {
965 VIDEOINFOHEADER *v = (void *) type.pbFormat;
966 time_base = (AVRational) { v->AvgTimePerFrame, 10000000 };
968 } else if (IsEqualGUID(&type.formattype, &FORMAT_VideoInfo2)) {
969 VIDEOINFOHEADER2 *v = (void *) type.pbFormat;
970 time_base = (AVRational) { v->AvgTimePerFrame, 10000000 };
974 av_log(avctx, AV_LOG_ERROR, "Could not get media type.\n");
978 st->avg_frame_rate = av_inv_q(time_base);
979 st->r_frame_rate = av_inv_q(time_base);
981 par->codec_type = AVMEDIA_TYPE_VIDEO;
982 par->width = bih->biWidth;
983 par->height = bih->biHeight;
984 par->codec_tag = bih->biCompression;
985 par->format = dshow_pixfmt(bih->biCompression, bih->biBitCount);
986 if (bih->biCompression == MKTAG('H', 'D', 'Y', 'C')) {
987 av_log(avctx, AV_LOG_DEBUG, "attempt to use full range for HDYC...\n");
988 par->color_range = AVCOL_RANGE_MPEG; // just in case it needs this...
990 if (par->format == AV_PIX_FMT_NONE) {
991 const AVCodecTag *const tags[] = { avformat_get_riff_video_tags(), NULL };
992 par->codec_id = av_codec_get_id(tags, bih->biCompression);
993 if (par->codec_id == AV_CODEC_ID_NONE) {
994 av_log(avctx, AV_LOG_ERROR, "Unknown compression type. "
995 "Please report type 0x%X.\n", (int) bih->biCompression);
996 return AVERROR_PATCHWELCOME;
998 par->bits_per_coded_sample = bih->biBitCount;
1000 par->codec_id = AV_CODEC_ID_RAWVIDEO;
1001 if (bih->biCompression == BI_RGB || bih->biCompression == BI_BITFIELDS) {
1002 par->bits_per_coded_sample = bih->biBitCount;
1003 par->extradata = av_malloc(9 + AV_INPUT_BUFFER_PADDING_SIZE);
1004 if (par->extradata) {
1005 par->extradata_size = 9;
1006 memcpy(par->extradata, "BottomUp", 9);
1011 WAVEFORMATEX *fx = NULL;
1013 if (IsEqualGUID(&type.formattype, &FORMAT_WaveFormatEx)) {
1014 fx = (void *) type.pbFormat;
1017 av_log(avctx, AV_LOG_ERROR, "Could not get media type.\n");
1021 par->codec_type = AVMEDIA_TYPE_AUDIO;
1022 par->format = sample_fmt_bits_per_sample(fx->wBitsPerSample);
1023 par->codec_id = waveform_codec_id(par->format);
1024 par->sample_rate = fx->nSamplesPerSec;
1025 par->channels = fx->nChannels;
1028 avpriv_set_pts_info(st, 64, 1, 10000000);
1036 static int parse_device_name(AVFormatContext *avctx)
1038 struct dshow_ctx *ctx = avctx->priv_data;
1039 char **device_name = ctx->device_name;
1040 char *name = av_strdup(avctx->filename);
1045 while ((type = strtok(tmp, "="))) {
1046 char *token = strtok(NULL, ":");
1049 if (!strcmp(type, "video")) {
1050 device_name[0] = token;
1051 } else if (!strcmp(type, "audio")) {
1052 device_name[1] = token;
1054 device_name[0] = NULL;
1055 device_name[1] = NULL;
1060 if (!device_name[0] && !device_name[1]) {
1064 device_name[0] = av_strdup(device_name[0]);
1066 device_name[1] = av_strdup(device_name[1]);
1073 static int dshow_read_header(AVFormatContext *avctx)
1075 struct dshow_ctx *ctx = avctx->priv_data;
1076 IGraphBuilder *graph = NULL;
1077 ICreateDevEnum *devenum = NULL;
1078 IMediaControl *control = NULL;
1079 IMediaEvent *media_event = NULL;
1080 HANDLE media_event_handle;
1082 int ret = AVERROR(EIO);
1087 if (!ctx->list_devices && !parse_device_name(avctx)) {
1088 av_log(avctx, AV_LOG_ERROR, "Malformed dshow input string.\n");
1092 ctx->video_codec_id = avctx->video_codec_id ? avctx->video_codec_id
1093 : AV_CODEC_ID_RAWVIDEO;
1094 if (ctx->pixel_format != AV_PIX_FMT_NONE) {
1095 if (ctx->video_codec_id != AV_CODEC_ID_RAWVIDEO) {
1096 av_log(avctx, AV_LOG_ERROR, "Pixel format may only be set when "
1097 "video codec is not set or set to rawvideo\n");
1098 ret = AVERROR(EINVAL);
1102 if (ctx->framerate) {
1103 r = av_parse_video_rate(&ctx->requested_framerate, ctx->framerate);
1105 av_log(avctx, AV_LOG_ERROR, "Could not parse framerate '%s'.\n", ctx->framerate);
1110 r = CoCreateInstance(&CLSID_FilterGraph, NULL, CLSCTX_INPROC_SERVER,
1111 &IID_IGraphBuilder, (void **) &graph);
1113 av_log(avctx, AV_LOG_ERROR, "Could not create capture graph.\n");
1118 r = CoCreateInstance(&CLSID_SystemDeviceEnum, NULL, CLSCTX_INPROC_SERVER,
1119 &IID_ICreateDevEnum, (void **) &devenum);
1121 av_log(avctx, AV_LOG_ERROR, "Could not enumerate system devices.\n");
1125 if (ctx->list_devices) {
1126 av_log(avctx, AV_LOG_INFO, "DirectShow video devices (some may be both video and audio devices)\n");
1127 dshow_cycle_devices(avctx, devenum, VideoDevice, VideoSourceDevice, NULL, NULL);
1128 av_log(avctx, AV_LOG_INFO, "DirectShow audio devices\n");
1129 dshow_cycle_devices(avctx, devenum, AudioDevice, AudioSourceDevice, NULL, NULL);
1133 if (ctx->list_options) {
1134 if (ctx->device_name[VideoDevice])
1135 if ((r = dshow_list_device_options(avctx, devenum, VideoDevice, VideoSourceDevice))) {
1139 if (ctx->device_name[AudioDevice]) {
1140 if (dshow_list_device_options(avctx, devenum, AudioDevice, AudioSourceDevice)) {
1141 /* show audio options from combined video+audio sources as fallback */
1142 if ((r = dshow_list_device_options(avctx, devenum, AudioDevice, VideoSourceDevice))) {
1149 if (ctx->device_name[VideoDevice]) {
1150 if ((r = dshow_open_device(avctx, devenum, VideoDevice, VideoSourceDevice)) < 0 ||
1151 (r = dshow_add_device(avctx, VideoDevice)) < 0) {
1156 if (ctx->device_name[AudioDevice]) {
1157 if ((r = dshow_open_device(avctx, devenum, AudioDevice, AudioSourceDevice)) < 0 ||
1158 (r = dshow_add_device(avctx, AudioDevice)) < 0) {
1159 av_log(avctx, AV_LOG_INFO, "Searching for audio device within video devices for %s\n", ctx->device_name[AudioDevice]);
1160 /* see if there's a video source with an audio pin with the given audio name */
1161 if ((r = dshow_open_device(avctx, devenum, AudioDevice, VideoSourceDevice)) < 0 ||
1162 (r = dshow_add_device(avctx, AudioDevice)) < 0) {
1168 if (ctx->list_options) {
1169 /* allow it to list crossbar options in dshow_open_device */
1173 ctx->curbufsize[0] = 0;
1174 ctx->curbufsize[1] = 0;
1175 ctx->mutex = CreateMutex(NULL, 0, NULL);
1177 av_log(avctx, AV_LOG_ERROR, "Could not create Mutex\n");
1180 ctx->event[1] = CreateEvent(NULL, 1, 0, NULL);
1181 if (!ctx->event[1]) {
1182 av_log(avctx, AV_LOG_ERROR, "Could not create Event\n");
1186 r = IGraphBuilder_QueryInterface(graph, &IID_IMediaControl, (void **) &control);
1188 av_log(avctx, AV_LOG_ERROR, "Could not get media control.\n");
1191 ctx->control = control;
1193 r = IGraphBuilder_QueryInterface(graph, &IID_IMediaEvent, (void **) &media_event);
1195 av_log(avctx, AV_LOG_ERROR, "Could not get media event.\n");
1198 ctx->media_event = media_event;
1200 r = IMediaEvent_GetEventHandle(media_event, (void *) &media_event_handle);
1202 av_log(avctx, AV_LOG_ERROR, "Could not get media event handle.\n");
1205 proc = GetCurrentProcess();
1206 r = DuplicateHandle(proc, media_event_handle, proc, &ctx->event[0],
1207 0, 0, DUPLICATE_SAME_ACCESS);
1209 av_log(avctx, AV_LOG_ERROR, "Could not duplicate media event handle.\n");
1213 r = IMediaControl_Run(control);
1216 r = IMediaControl_GetState(control, 0, &pfs);
1219 av_log(avctx, AV_LOG_ERROR, "Could not run graph (sometimes caused by a device already in use by other application)\n");
1228 ICreateDevEnum_Release(devenum);
1231 dshow_read_close(avctx);
1237 * Checks media events from DirectShow and returns -1 on error or EOF. Also
1238 * purges all events that might be in the event queue to stop the trigger
1239 * of event notification.
1241 static int dshow_check_event_queue(IMediaEvent *media_event)
1247 while (IMediaEvent_GetEvent(media_event, &code, &p1, &p2, 0) != E_ABORT) {
1248 if (code == EC_COMPLETE || code == EC_DEVICE_LOST || code == EC_ERRORABORT)
1250 IMediaEvent_FreeEventParams(media_event, code, p1, p2);
1256 static int dshow_read_packet(AVFormatContext *s, AVPacket *pkt)
1258 struct dshow_ctx *ctx = s->priv_data;
1259 AVPacketList *pktl = NULL;
1261 while (!ctx->eof && !pktl) {
1262 WaitForSingleObject(ctx->mutex, INFINITE);
1266 ctx->pktl = ctx->pktl->next;
1268 ctx->curbufsize[pkt->stream_index] -= pkt->size;
1270 ResetEvent(ctx->event[1]);
1271 ReleaseMutex(ctx->mutex);
1273 if (dshow_check_event_queue(ctx->media_event) < 0) {
1275 } else if (s->flags & AVFMT_FLAG_NONBLOCK) {
1276 return AVERROR(EAGAIN);
1278 WaitForMultipleObjects(2, ctx->event, 0, INFINITE);
1283 return ctx->eof ? AVERROR(EIO) : pkt->size;
1286 #define OFFSET(x) offsetof(struct dshow_ctx, x)
1287 #define DEC AV_OPT_FLAG_DECODING_PARAM
1288 static const AVOption options[] = {
1289 { "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 },
1290 { "pixel_format", "set video pixel format", OFFSET(pixel_format), AV_OPT_TYPE_PIXEL_FMT, {.i64 = AV_PIX_FMT_NONE}, -1, INT_MAX, DEC },
1291 { "framerate", "set video frame rate", OFFSET(framerate), AV_OPT_TYPE_STRING, {.str = NULL}, 0, 0, DEC },
1292 { "sample_rate", "set audio sample rate", OFFSET(sample_rate), AV_OPT_TYPE_INT, {.i64 = 0}, 0, INT_MAX, DEC },
1293 { "sample_size", "set audio sample size", OFFSET(sample_size), AV_OPT_TYPE_INT, {.i64 = 0}, 0, 16, DEC },
1294 { "channels", "set number of audio channels, such as 1 or 2", OFFSET(channels), AV_OPT_TYPE_INT, {.i64 = 0}, 0, INT_MAX, DEC },
1295 { "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 },
1296 { "list_devices", "list available devices", OFFSET(list_devices), AV_OPT_TYPE_BOOL, {.i64=0}, 0, 1, DEC },
1297 { "list_options", "list available options for specified device", OFFSET(list_options), AV_OPT_TYPE_BOOL, {.i64=0}, 0, 1, DEC },
1298 { "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 },
1299 { "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 },
1300 { "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 },
1301 { "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 },
1302 { "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 },
1303 { "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 },
1304 { "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 },
1305 { "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 },
1306 { "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 },
1307 { "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 },
1308 { "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 },
1309 { "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 },
1310 { "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 },
1311 { "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 },
1312 { "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 },
1313 { "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 },
1317 static const AVClass dshow_class = {
1318 .class_name = "dshow indev",
1319 .item_name = av_default_item_name,
1321 .version = LIBAVUTIL_VERSION_INT,
1322 .category = AV_CLASS_CATEGORY_DEVICE_VIDEO_INPUT,
1325 AVInputFormat ff_dshow_demuxer = {
1327 .long_name = NULL_IF_CONFIG_SMALL("DirectShow capture"),
1328 .priv_data_size = sizeof(struct dshow_ctx),
1329 .read_header = dshow_read_header,
1330 .read_packet = dshow_read_packet,
1331 .read_close = dshow_read_close,
1332 .flags = AVFMT_NOFILE,
1333 .priv_class = &dshow_class,