]> git.sesse.net Git - casparcg/blobdiff - modules/ffmpeg/producer/util/util.cpp
[ffmpeg_producer] Recognize .mkv and .mxf files as playable without probing
[casparcg] / modules / ffmpeg / producer / util / util.cpp
index f677b491856ae43c8c654647c8a36b33f4e7b9bc..6ed51aa1c8797beb7469dcc3b159871636b52fda 100644 (file)
-#include "../../stdafx.h"\r
-\r
-#include "util.h"\r
-\r
-#include "flv.h"\r
-\r
-#include "../tbb_avcodec.h"\r
-#include "../../ffmpeg_error.h"\r
-\r
-#include <tbb/concurrent_unordered_map.h>\r
-#include <tbb/concurrent_queue.h>\r
-\r
-#include <core/producer/frame/frame_transform.h>\r
-#include <core/producer/frame/frame_factory.h>\r
-#include <core/producer/frame_producer.h>\r
-#include <core/mixer/write_frame.h>\r
-\r
-#include <common/exception/exceptions.h>\r
-#include <common/utility/assert.h>\r
-#include <common/memory/memcpy.h>\r
-\r
-#include <tbb/parallel_for.h>\r
-\r
-#include <boost/filesystem.hpp>\r
-#include <boost/lexical_cast.hpp>\r
-\r
-#if defined(_MSC_VER)\r
-#pragma warning (push)\r
-#pragma warning (disable : 4244)\r
-#endif\r
-extern "C" \r
-{\r
-       #include <libswscale/swscale.h>\r
-       #include <libavcodec/avcodec.h>\r
-       #include <libavformat/avformat.h>\r
-}\r
-#if defined(_MSC_VER)\r
-#pragma warning (pop)\r
-#endif\r
-\r
-namespace caspar { namespace ffmpeg {\r
-               \r
-std::shared_ptr<core::audio_buffer> flush_audio()\r
-{\r
-       static std::shared_ptr<core::audio_buffer> audio(new core::audio_buffer());\r
-       return audio;\r
-}\r
-\r
-std::shared_ptr<core::audio_buffer> empty_audio()\r
-{\r
-       static std::shared_ptr<core::audio_buffer> audio(new core::audio_buffer());\r
-       return audio;\r
-}\r
-\r
-std::shared_ptr<AVFrame>                       flush_video()\r
-{\r
-       static std::shared_ptr<AVFrame> video(avcodec_alloc_frame(), av_free);\r
-       return video;\r
-}\r
-\r
-std::shared_ptr<AVFrame>                       empty_video()\r
-{\r
-       static std::shared_ptr<AVFrame> video(avcodec_alloc_frame(), av_free);\r
-       return video;\r
-}\r
-\r
-core::field_mode::type get_mode(const AVFrame& frame)\r
-{\r
-       if(!frame.interlaced_frame)\r
-               return core::field_mode::progressive;\r
-\r
-       return frame.top_field_first ? core::field_mode::upper : core::field_mode::lower;\r
-}\r
-\r
-core::pixel_format::type get_pixel_format(PixelFormat pix_fmt)\r
-{\r
-       switch(pix_fmt)\r
-       {\r
-       case CASPAR_PIX_FMT_LUMA:       return core::pixel_format::luma;\r
-       case PIX_FMT_GRAY8:                     return core::pixel_format::gray;\r
-       case PIX_FMT_BGRA:                      return core::pixel_format::bgra;\r
-       case PIX_FMT_ARGB:                      return core::pixel_format::argb;\r
-       case PIX_FMT_RGBA:                      return core::pixel_format::rgba;\r
-       case PIX_FMT_ABGR:                      return core::pixel_format::abgr;\r
-       case PIX_FMT_YUV444P:           return core::pixel_format::ycbcr;\r
-       case PIX_FMT_YUV422P:           return core::pixel_format::ycbcr;\r
-       case PIX_FMT_YUV420P:           return core::pixel_format::ycbcr;\r
-       case PIX_FMT_YUV411P:           return core::pixel_format::ycbcr;\r
-       case PIX_FMT_YUV410P:           return core::pixel_format::ycbcr;\r
-       case PIX_FMT_YUVA420P:          return core::pixel_format::ycbcra;\r
-       default:                                        return core::pixel_format::invalid;\r
-       }\r
-}\r
-\r
-core::pixel_format_desc get_pixel_format_desc(PixelFormat pix_fmt, size_t width, size_t height)\r
-{\r
-       // Get linesizes\r
-       AVPicture dummy_pict;   \r
-       avpicture_fill(&dummy_pict, nullptr, pix_fmt == CASPAR_PIX_FMT_LUMA ? PIX_FMT_GRAY8 : pix_fmt, width, height);\r
-\r
-       core::pixel_format_desc desc;\r
-       desc.pix_fmt = get_pixel_format(pix_fmt);\r
-               \r
-       switch(desc.pix_fmt)\r
-       {\r
-       case core::pixel_format::gray:\r
-       case core::pixel_format::luma:\r
-               {\r
-                       desc.planes.push_back(core::pixel_format_desc::plane(dummy_pict.linesize[0], height, 1));                                               \r
-                       return desc;\r
-               }\r
-       case core::pixel_format::bgra:\r
-       case core::pixel_format::argb:\r
-       case core::pixel_format::rgba:\r
-       case core::pixel_format::abgr:\r
-               {\r
-                       desc.planes.push_back(core::pixel_format_desc::plane(dummy_pict.linesize[0]/4, height, 4));                                             \r
-                       return desc;\r
-               }\r
-       case core::pixel_format::ycbcr:\r
-       case core::pixel_format::ycbcra:\r
-               {               \r
-                       // Find chroma height\r
-                       size_t size2 = dummy_pict.data[2] - dummy_pict.data[1];\r
-                       size_t h2 = size2/dummy_pict.linesize[1];                       \r
-\r
-                       desc.planes.push_back(core::pixel_format_desc::plane(dummy_pict.linesize[0], height, 1));\r
-                       desc.planes.push_back(core::pixel_format_desc::plane(dummy_pict.linesize[1], h2, 1));\r
-                       desc.planes.push_back(core::pixel_format_desc::plane(dummy_pict.linesize[2], h2, 1));\r
-\r
-                       if(desc.pix_fmt == core::pixel_format::ycbcra)                                          \r
-                               desc.planes.push_back(core::pixel_format_desc::plane(dummy_pict.linesize[3], height, 1));       \r
-                       return desc;\r
-               }               \r
-       default:                \r
-               desc.pix_fmt = core::pixel_format::invalid;\r
-               return desc;\r
-       }\r
-}\r
-\r
-int make_alpha_format(int format)\r
-{\r
-       switch(get_pixel_format(static_cast<PixelFormat>(format)))\r
-       {\r
-       case core::pixel_format::ycbcr:\r
-       case core::pixel_format::ycbcra:\r
-               return CASPAR_PIX_FMT_LUMA;\r
-       default:\r
-               return format;\r
-       }\r
-}\r
-\r
-safe_ptr<core::write_frame> make_write_frame(const void* tag, const safe_ptr<AVFrame>& decoded_frame, const safe_ptr<core::frame_factory>& frame_factory, int hints)\r
-{                      \r
-       static tbb::concurrent_unordered_map<size_t, tbb::concurrent_queue<std::shared_ptr<SwsContext>>> sws_contexts_;\r
-       \r
-       if(decoded_frame->width < 1 || decoded_frame->height < 1)\r
-               return make_safe<core::write_frame>(tag);\r
-\r
-       const auto width  = decoded_frame->width;\r
-       const auto height = decoded_frame->height;\r
-       auto desc                 = get_pixel_format_desc(static_cast<PixelFormat>(decoded_frame->format), width, height);\r
-       \r
-       if(hints & core::frame_producer::ALPHA_HINT)\r
-               desc = get_pixel_format_desc(static_cast<PixelFormat>(make_alpha_format(decoded_frame->format)), width, height);\r
-\r
-       std::shared_ptr<core::write_frame> write;\r
-\r
-       if(desc.pix_fmt == core::pixel_format::invalid)\r
-       {\r
-               auto pix_fmt = static_cast<PixelFormat>(decoded_frame->format);\r
-\r
-               write = frame_factory->create_frame(tag, get_pixel_format_desc(PIX_FMT_BGRA, width, height));\r
-               write->set_type(get_mode(*decoded_frame));\r
-\r
-               std::shared_ptr<SwsContext> sws_context;\r
-\r
-               //CASPAR_LOG(warning) << "Hardware accelerated color transform not supported.";\r
-\r
-               size_t key = width << 20 | height << 8 | pix_fmt;\r
-                       \r
-               auto& pool = sws_contexts_[key];\r
-                                               \r
-               if(!pool.try_pop(sws_context))\r
-               {\r
-                       double param;\r
-                       sws_context.reset(sws_getContext(width, height, pix_fmt, width, height, PIX_FMT_BGRA, SWS_BILINEAR, nullptr, nullptr, &param), sws_freeContext);\r
-               }\r
-                       \r
-               if(!sws_context)\r
-               {\r
-                       BOOST_THROW_EXCEPTION(operation_failed() << msg_info("Could not create software scaling context.") << \r
-                                                                       boost::errinfo_api_function("sws_getContext"));\r
-               }       \r
-\r
-               // Use sws_scale when provided colorspace has no hw-accel.\r
-               safe_ptr<AVFrame> av_frame(avcodec_alloc_frame(), av_free);     \r
-               avcodec_get_frame_defaults(av_frame.get());                     \r
-               auto size = avpicture_fill(reinterpret_cast<AVPicture*>(av_frame.get()), write->image_data().begin(), PIX_FMT_BGRA, width, height);\r
-               CASPAR_VERIFY(size == write->image_data().size()); \r
-\r
-               sws_scale(sws_context.get(), decoded_frame->data, decoded_frame->linesize, 0, height, av_frame->data, av_frame->linesize);      \r
-               pool.push(sws_context);\r
-\r
-               write->commit();\r
-       }\r
-       else\r
-       {\r
-               write = frame_factory->create_frame(tag, desc);\r
-               write->set_type(get_mode(*decoded_frame));\r
-\r
-               for(int n = 0; n < static_cast<int>(desc.planes.size()); ++n)\r
-               {\r
-                       auto plane            = desc.planes[n];\r
-                       auto result           = write->image_data(n).begin();\r
-                       auto decoded          = decoded_frame->data[n];\r
-                       auto decoded_linesize = decoded_frame->linesize[n];\r
-                       \r
-                       CASPAR_ASSERT(decoded);\r
-                       CASPAR_ASSERT(write->image_data(n).begin());\r
-\r
-                       if(decoded_linesize != static_cast<int>(plane.width))\r
-                       {\r
-                               // Copy line by line since ffmpeg sometimes pads each line.\r
-                               tbb::parallel_for<size_t>(0, desc.planes[n].height, [&](size_t y)\r
-                               {\r
-                                       fast_memcpy(result + y*plane.linesize, decoded + y*decoded_linesize, plane.linesize);\r
-                               });\r
-                       }\r
-                       else\r
-                       {\r
-                               fast_memcpy(result, decoded, plane.size);\r
-                       }\r
-\r
-                       write->commit(n);\r
-               }\r
-       }\r
-\r
-       if(decoded_frame->height == 480) // NTSC DV\r
-       {\r
-               write->get_frame_transform().fill_translation[1] += 2.0/static_cast<double>(frame_factory->get_video_format_desc().height);\r
-               write->get_frame_transform().fill_scale[1] = 1.0 - 6.0*1.0/static_cast<double>(frame_factory->get_video_format_desc().height);\r
-       }\r
-       \r
-       // Fix field-order if needed\r
-       if(write->get_type() == core::field_mode::lower && frame_factory->get_video_format_desc().field_mode == core::field_mode::upper)\r
-               write->get_frame_transform().fill_translation[1] += 1.0/static_cast<double>(frame_factory->get_video_format_desc().height);\r
-       else if(write->get_type() == core::field_mode::upper && frame_factory->get_video_format_desc().field_mode == core::field_mode::lower)\r
-               write->get_frame_transform().fill_translation[1] -= 1.0/static_cast<double>(frame_factory->get_video_format_desc().height);\r
-\r
-       return make_safe_ptr(write);\r
-}\r
-\r
-bool is_sane_fps(AVRational time_base)\r
-{\r
-       double fps = static_cast<double>(time_base.den) / static_cast<double>(time_base.num);\r
-       return fps > 20.0 && fps < 65.0;\r
-}\r
-\r
-AVRational fix_time_base(AVRational time_base)\r
-{\r
-       if(time_base.num == 1)\r
-               time_base.num = static_cast<int>(std::pow(10.0, static_cast<int>(std::log10(static_cast<float>(time_base.den)))-1));    \r
-                       \r
-       if(!is_sane_fps(time_base))\r
-       {\r
-               auto tmp = time_base;\r
-               tmp.den /= 2;\r
-               if(is_sane_fps(tmp))\r
-                       time_base = tmp;\r
-       }\r
-\r
-       return time_base;\r
-}\r
-\r
-double read_fps(AVFormatContext& context, double fail_value)\r
-{                                              \r
-       auto video_index = av_find_best_stream(&context, AVMEDIA_TYPE_VIDEO, -1, -1, 0, 0);\r
-       auto audio_index = av_find_best_stream(&context, AVMEDIA_TYPE_AUDIO, -1, -1, 0, 0);\r
-       \r
-       if(video_index > -1)\r
-       {\r
-               const auto video_context = context.streams[video_index]->codec;\r
-               const auto video_stream  = context.streams[video_index];\r
-                                               \r
-               AVRational time_base = video_context->time_base;\r
-\r
-               if(boost::filesystem2::path(context.filename).extension() == ".flv")\r
-               {\r
-                       try\r
-                       {\r
-                               auto meta = read_flv_meta_info(context.filename);\r
-                               return boost::lexical_cast<double>(meta["framerate"]);\r
-                       }\r
-                       catch(...)\r
-                       {\r
-                               return 0.0;\r
-                       }\r
-               }\r
-               else\r
-               {\r
-                       time_base.num *= video_context->ticks_per_frame;\r
-\r
-                       if(!is_sane_fps(time_base))\r
-                       {                       \r
-                               time_base = fix_time_base(time_base);\r
-\r
-                               if(!is_sane_fps(time_base) && audio_index > -1)\r
-                               {\r
-                                       auto& audio_context = *context.streams[audio_index]->codec;\r
-                                       auto& audio_stream  = *context.streams[audio_index];\r
-\r
-                                       double duration_sec = audio_stream.duration / static_cast<double>(audio_context.sample_rate);\r
-                                                               \r
-                                       time_base.num = static_cast<int>(duration_sec*100000.0);\r
-                                       time_base.den = static_cast<int>(video_stream->nb_frames*100000);\r
-                               }\r
-                       }\r
-               }\r
-               \r
-               double fps = static_cast<double>(time_base.den) / static_cast<double>(time_base.num);\r
-\r
-               double closest_fps = 0.0;\r
-               for(int n = 0; n < core::video_format::count; ++n)\r
-               {\r
-                       auto format = core::video_format_desc::get(static_cast<core::video_format::type>(n));\r
-\r
-                       double diff1 = std::abs(format.fps - fps);\r
-                       double diff2 = std::abs(closest_fps - fps);\r
-\r
-                       if(diff1 < diff2)\r
-                               closest_fps = format.fps;\r
-               }\r
-       \r
-               return closest_fps;\r
-       }\r
-\r
-       return fail_value;      \r
-}\r
-\r
-void fix_meta_data(AVFormatContext& context)\r
-{\r
-       auto video_index = av_find_best_stream(&context, AVMEDIA_TYPE_VIDEO, -1, -1, 0, 0);\r
-\r
-       if(video_index > -1)\r
-       {\r
-               auto video_stream   = context.streams[video_index];\r
-               auto video_context  = context.streams[video_index]->codec;\r
-                                               \r
-               if(boost::filesystem2::path(context.filename).extension() == ".flv")\r
-               {\r
-                       try\r
-                       {\r
-                               auto meta = read_flv_meta_info(context.filename);\r
-                               double fps = boost::lexical_cast<double>(meta["framerate"]);\r
-                               video_stream->nb_frames = static_cast<int64_t>(boost::lexical_cast<double>(meta["duration"])*fps);\r
-                       }\r
-                       catch(...){}\r
-               }\r
-               else\r
-               {\r
-                       video_stream->nb_frames = (video_stream->duration*video_stream->time_base.num*video_context->time_base.den)/(video_stream->time_base.den*video_context->time_base.num*video_context->ticks_per_frame);          }\r
-       }\r
-}\r
-\r
-safe_ptr<AVPacket> create_packet()\r
-{\r
-       safe_ptr<AVPacket> packet(new AVPacket, [](AVPacket* p)\r
-       {\r
-               av_free_packet(p);\r
-               delete p;\r
-       });\r
-       \r
-       av_init_packet(packet.get());\r
-       return packet;\r
-}\r
-\r
-safe_ptr<AVCodecContext> open_codec(AVFormatContext& context, enum AVMediaType type, int& index)\r
-{      \r
-       AVCodec* decoder;\r
-       index = THROW_ON_ERROR2(av_find_best_stream(&context, type, -1, -1, &decoder, 0), "");\r
-       //if(strcmp(decoder->name, "prores") == 0 && decoder->next && strcmp(decoder->next->name, "prores_lgpl") == 0)\r
-       //      decoder = decoder->next;\r
-\r
-       THROW_ON_ERROR2(tbb_avcodec_open(context.streams[index]->codec, decoder), "");\r
-       return safe_ptr<AVCodecContext>(context.streams[index]->codec, tbb_avcodec_close);\r
-}\r
-\r
-safe_ptr<AVFormatContext> open_input(const std::wstring& filename)\r
-{\r
-       AVFormatContext* weak_context = nullptr;\r
-       THROW_ON_ERROR2(avformat_open_input(&weak_context, narrow(filename).c_str(), nullptr, nullptr), filename);\r
-       safe_ptr<AVFormatContext> context(weak_context, av_close_input_file);                   \r
-       THROW_ON_ERROR2(avformat_find_stream_info(weak_context, nullptr), filename);\r
-       fix_meta_data(*context);\r
-       return context;\r
-}\r
-//\r
-//void av_dup_frame(AVFrame* frame)\r
-//{\r
-//     AVFrame* new_frame = avcodec_alloc_frame();\r
-//\r
-//\r
-//     const uint8_t *src_data[4] = {0};\r
-//     memcpy(const_cast<uint8_t**>(&src_data[0]), frame->data, 4);\r
-//     const int src_linesizes[4] = {0};\r
-//     memcpy(const_cast<int*>(&src_linesizes[0]), frame->linesize, 4);\r
-//\r
-//     av_image_alloc(new_frame->data, new_frame->linesize, new_frame->width, new_frame->height, frame->format, 16);\r
-//\r
-//     av_image_copy(new_frame->data, new_frame->linesize, src_data, src_linesizes, frame->format, new_frame->width, new_frame->height);\r
-//\r
-//     frame =\r
-//}\r
-\r
-}}
\ No newline at end of file
+/*
+* Copyright (c) 2011 Sveriges Television AB <info@casparcg.com>
+*
+* This file is part of CasparCG (www.casparcg.com).
+*
+* CasparCG is free software: you can redistribute it and/or modify
+* it under the terms of the GNU General Public License as published by
+* the Free Software Foundation, either version 3 of the License, or
+* (at your option) any later version.
+*
+* CasparCG is distributed in the hope that it will be useful,
+* but WITHOUT ANY WARRANTY; without even the implied warranty of
+* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+* GNU General Public License for more details.
+*
+* You should have received a copy of the GNU General Public License
+* along with CasparCG. If not, see <http://www.gnu.org/licenses/>.
+*
+* Author: Robert Nagy, ronag89@gmail.com
+*/
+
+#include "../../StdAfx.h"
+
+#include "util.h"
+
+#include "flv.h"
+
+#include "../tbb_avcodec.h"
+#include "../../ffmpeg_error.h"
+#include "../../ffmpeg.h"
+
+#include <tbb/concurrent_unordered_map.h>
+#include <tbb/concurrent_queue.h>
+
+#include <core/frame/frame_transform.h>
+#include <core/frame/frame_factory.h>
+#include <core/frame/frame.h>
+#include <core/frame/audio_channel_layout.h>
+#include <core/producer/frame_producer.h>
+
+#include <common/except.h>
+#include <common/array.h>
+#include <common/os/filesystem.h>
+#include <common/memcpy.h>
+
+#include <tbb/parallel_for.h>
+
+#include <common/assert.h>
+#include <boost/filesystem.hpp>
+#include <boost/lexical_cast.hpp>
+#include <boost/rational.hpp>
+
+#include <fstream>
+
+#include <asmlib.h>
+
+#if defined(_MSC_VER)
+#pragma warning (push)
+#pragma warning (disable : 4244)
+#endif
+extern "C"
+{
+       #include <libswscale/swscale.h>
+       #include <libavcodec/avcodec.h>
+       #include <libavformat/avformat.h>
+}
+#if defined(_MSC_VER)
+#pragma warning (pop)
+#endif
+
+namespace caspar { namespace ffmpeg {
+
+core::field_mode get_mode(const AVFrame& frame)
+{
+       if(!frame.interlaced_frame)
+               return core::field_mode::progressive;
+
+       return frame.top_field_first ? core::field_mode::upper : core::field_mode::lower;
+}
+
+core::pixel_format get_pixel_format(PixelFormat pix_fmt)
+{
+       switch(pix_fmt)
+       {
+       case PIX_FMT_GRAY8:                     return core::pixel_format::gray;
+       case PIX_FMT_RGB24:                     return core::pixel_format::rgb;
+       case PIX_FMT_BGR24:                     return core::pixel_format::bgr;
+       case PIX_FMT_BGRA:                      return core::pixel_format::bgra;
+       case PIX_FMT_ARGB:                      return core::pixel_format::argb;
+       case PIX_FMT_RGBA:                      return core::pixel_format::rgba;
+       case PIX_FMT_ABGR:                      return core::pixel_format::abgr;
+       case PIX_FMT_YUV444P:           return core::pixel_format::ycbcr;
+       case PIX_FMT_YUV422P:           return core::pixel_format::ycbcr;
+       case PIX_FMT_YUV420P:           return core::pixel_format::ycbcr;
+       case PIX_FMT_YUV411P:           return core::pixel_format::ycbcr;
+       case PIX_FMT_YUV410P:           return core::pixel_format::ycbcr;
+       case PIX_FMT_YUVA420P:          return core::pixel_format::ycbcra;
+       default:                                        return core::pixel_format::invalid;
+       }
+}
+
+core::pixel_format_desc pixel_format_desc(PixelFormat pix_fmt, int width, int height)
+{
+       // Get linesizes
+       AVPicture dummy_pict;
+       avpicture_fill(&dummy_pict, nullptr, pix_fmt, width, height);
+
+       core::pixel_format_desc desc = get_pixel_format(pix_fmt);
+
+       switch(desc.format)
+       {
+       case core::pixel_format::gray:
+       case core::pixel_format::luma:
+               {
+                       desc.planes.push_back(core::pixel_format_desc::plane(dummy_pict.linesize[0], height, 1));
+                       return desc;
+               }
+       case core::pixel_format::bgr:
+       case core::pixel_format::rgb:
+               {
+                       desc.planes.push_back(core::pixel_format_desc::plane(dummy_pict.linesize[0]/3, height, 3));
+                       return desc;
+               }
+       case core::pixel_format::bgra:
+       case core::pixel_format::argb:
+       case core::pixel_format::rgba:
+       case core::pixel_format::abgr:
+               {
+                       desc.planes.push_back(core::pixel_format_desc::plane(dummy_pict.linesize[0]/4, height, 4));
+                       return desc;
+               }
+       case core::pixel_format::ycbcr:
+       case core::pixel_format::ycbcra:
+               {
+                       // Find chroma height
+                       int size2 = static_cast<int>(dummy_pict.data[2] - dummy_pict.data[1]);
+                       int h2 = size2/dummy_pict.linesize[1];
+
+                       desc.planes.push_back(core::pixel_format_desc::plane(dummy_pict.linesize[0], height, 1));
+                       desc.planes.push_back(core::pixel_format_desc::plane(dummy_pict.linesize[1], h2, 1));
+                       desc.planes.push_back(core::pixel_format_desc::plane(dummy_pict.linesize[2], h2, 1));
+
+                       if(desc.format == core::pixel_format::ycbcra)
+                               desc.planes.push_back(core::pixel_format_desc::plane(dummy_pict.linesize[3], height, 1));
+                       return desc;
+               }
+       default:
+               desc.format = core::pixel_format::invalid;
+               return desc;
+       }
+}
+
+core::mutable_frame make_frame(const void* tag, const spl::shared_ptr<AVFrame>& decoded_frame, core::frame_factory& frame_factory, const core::audio_channel_layout& channel_layout)
+{
+       static tbb::concurrent_unordered_map<int64_t, tbb::concurrent_queue<std::shared_ptr<SwsContext>>> sws_contvalid_exts_;
+
+       if(decoded_frame->width < 1 || decoded_frame->height < 1)
+               return frame_factory.create_frame(tag, core::pixel_format_desc(core::pixel_format::invalid), core::audio_channel_layout::invalid());
+
+       const auto width  = decoded_frame->width;
+       const auto height = decoded_frame->height;
+       auto desc                 = pixel_format_desc(static_cast<PixelFormat>(decoded_frame->format), width, height);
+
+       if(desc.format == core::pixel_format::invalid)
+       {
+               auto pix_fmt = static_cast<PixelFormat>(decoded_frame->format);
+               auto target_pix_fmt = PIX_FMT_BGRA;
+
+               if(pix_fmt == PIX_FMT_UYVY422)
+                       target_pix_fmt = PIX_FMT_YUV422P;
+               else if(pix_fmt == PIX_FMT_YUYV422)
+                       target_pix_fmt = PIX_FMT_YUV422P;
+               else if(pix_fmt == PIX_FMT_UYYVYY411)
+                       target_pix_fmt = PIX_FMT_YUV411P;
+               else if(pix_fmt == PIX_FMT_YUV420P10)
+                       target_pix_fmt = PIX_FMT_YUV420P;
+               else if(pix_fmt == PIX_FMT_YUV422P10)
+                       target_pix_fmt = PIX_FMT_YUV422P;
+               else if(pix_fmt == PIX_FMT_YUV444P10)
+                       target_pix_fmt = PIX_FMT_YUV444P;
+
+               auto target_desc = pixel_format_desc(target_pix_fmt, width, height);
+
+               auto write = frame_factory.create_frame(tag, target_desc, channel_layout);
+
+               std::shared_ptr<SwsContext> sws_context;
+
+               //CASPAR_LOG(warning) << "Hardware accelerated color transform not supported.";
+
+               int64_t key = ((static_cast<int64_t>(width)                      << 32) & 0xFFFF00000000) |
+                                         ((static_cast<int64_t>(height)                 << 16) & 0xFFFF0000) |
+                                         ((static_cast<int64_t>(pix_fmt)                <<  8) & 0xFF00) |
+                                         ((static_cast<int64_t>(target_pix_fmt) <<  0) & 0xFF);
+
+               auto& pool = sws_contvalid_exts_[key];
+
+               if(!pool.try_pop(sws_context))
+               {
+                       double param;
+                       sws_context.reset(sws_getContext(width, height, pix_fmt, width, height, target_pix_fmt, SWS_BILINEAR, nullptr, nullptr, &param), sws_freeContext);
+               }
+
+               if(!sws_context)
+               {
+                       CASPAR_THROW_EXCEPTION(operation_failed() << msg_info("Could not create software scaling context.") <<
+                                                                       boost::errinfo_api_function("sws_getContext"));
+               }
+
+               auto av_frame = create_frame();
+               if(target_pix_fmt == PIX_FMT_BGRA)
+               {
+                       auto size = avpicture_fill(reinterpret_cast<AVPicture*>(av_frame.get()), write.image_data(0).begin(), PIX_FMT_BGRA, width, height);
+                       CASPAR_VERIFY(size == write.image_data(0).size());
+               }
+               else
+               {
+                       av_frame->width  = width;
+                       av_frame->height = height;
+                       for(int n = 0; n < target_desc.planes.size(); ++n)
+                       {
+                               av_frame->data[n]               = write.image_data(n).begin();
+                               av_frame->linesize[n]   = target_desc.planes[n].linesize;
+                       }
+               }
+
+               sws_scale(sws_context.get(), decoded_frame->data, decoded_frame->linesize, 0, height, av_frame->data, av_frame->linesize);
+               pool.push(sws_context);
+
+               return std::move(write);
+       }
+       else
+       {
+               auto write = frame_factory.create_frame(tag, desc, channel_layout);
+
+               for(int n = 0; n < static_cast<int>(desc.planes.size()); ++n)
+               {
+                       auto plane            = desc.planes[n];
+                       auto result           = write.image_data(n).begin();
+                       auto decoded          = decoded_frame->data[n];
+                       auto decoded_linesize = decoded_frame->linesize[n];
+
+                       CASPAR_ASSERT(decoded);
+                       CASPAR_ASSERT(write.image_data(n).begin());
+
+                       if (decoded_linesize != plane.linesize)
+                       {
+                               // Copy line by line since ffmpeg sometimes pads each line.
+                               tbb::affinity_partitioner ap;
+                               tbb::parallel_for(tbb::blocked_range<int>(0, desc.planes[n].height), [&](const tbb::blocked_range<int>& r)
+                               {
+                                       for (int y = r.begin(); y != r.end(); ++y)
+                                               A_memcpy(result + y*plane.linesize, decoded + y*decoded_linesize, plane.linesize);
+                               }, ap);
+                       }
+                       else
+                       {
+                               fast_memcpy(result, decoded, plane.size);
+                       }
+               }
+
+               return std::move(write);
+       }
+}
+
+spl::shared_ptr<AVFrame> make_av_frame(core::mutable_frame& frame)
+{
+       std::array<uint8_t*, 4> data = {};
+       for(int n = 0; n < frame.pixel_format_desc().planes.size(); ++n)
+               data[n] = frame.image_data(n).begin();
+
+       return make_av_frame(data, frame.pixel_format_desc());
+}
+
+spl::shared_ptr<AVFrame> make_av_frame(std::array<uint8_t*, 4> data, const core::pixel_format_desc& pix_desc)
+{
+       auto av_frame = create_frame();
+
+       auto planes              = pix_desc.planes;
+       auto format              = pix_desc.format;
+
+       av_frame->width  = planes[0].width;
+       av_frame->height = planes[0].height;
+       for(int n = 0; n < planes.size(); ++n)
+       {
+               av_frame->data[n]         = data[n];
+               av_frame->linesize[n] = planes[n].linesize;
+       }
+
+       switch(format)
+       {
+       case core::pixel_format::rgb:
+               av_frame->format = PIX_FMT_RGB24;
+               break;
+       case core::pixel_format::bgr:
+               av_frame->format = PIX_FMT_BGR24;
+               break;
+       case core::pixel_format::rgba:
+               av_frame->format = PIX_FMT_RGBA;
+               break;
+       case core::pixel_format::argb:
+               av_frame->format = PIX_FMT_ARGB;
+               break;
+       case core::pixel_format::bgra:
+               av_frame->format = PIX_FMT_BGRA;
+               break;
+       case core::pixel_format::abgr:
+               av_frame->format = PIX_FMT_ABGR;
+               break;
+       case core::pixel_format::gray:
+               av_frame->format = PIX_FMT_GRAY8;
+               break;
+       case core::pixel_format::ycbcr:
+       {
+               int y_w = planes[0].width;
+               int y_h = planes[0].height;
+               int c_w = planes[1].width;
+               int c_h = planes[1].height;
+
+               if(c_h == y_h && c_w == y_w)
+                       av_frame->format = PIX_FMT_YUV444P;
+               else if(c_h == y_h && c_w*2 == y_w)
+                       av_frame->format = PIX_FMT_YUV422P;
+               else if(c_h == y_h && c_w*4 == y_w)
+                       av_frame->format = PIX_FMT_YUV411P;
+               else if(c_h*2 == y_h && c_w*2 == y_w)
+                       av_frame->format = PIX_FMT_YUV420P;
+               else if(c_h*2 == y_h && c_w*4 == y_w)
+                       av_frame->format = PIX_FMT_YUV410P;
+
+               break;
+       }
+       case core::pixel_format::ycbcra:
+               av_frame->format = PIX_FMT_YUVA420P;
+               break;
+       }
+       return av_frame;
+}
+
+bool is_sane_fps(AVRational time_base)
+{
+       double fps = static_cast<double>(time_base.den) / static_cast<double>(time_base.num);
+       return fps > 20.0 && fps < 65.0;
+}
+
+AVRational fix_time_base(AVRational time_base)
+{
+       if(time_base.num == 1)
+               time_base.num = static_cast<int>(std::pow(10.0, static_cast<int>(std::log10(static_cast<float>(time_base.den)))-1));
+
+       if(!is_sane_fps(time_base))
+       {
+               auto tmp = time_base;
+               tmp.den /= 2;
+               if(is_sane_fps(tmp))
+                       time_base = tmp;
+       }
+
+       return time_base;
+}
+
+double read_fps(AVFormatContext& context, double fail_value)
+{
+       auto framerate = read_framerate(context, boost::rational<int>(static_cast<int>(fail_value * 1000000.0), 1000000));
+
+       return static_cast<double>(framerate.numerator()) / static_cast<double>(framerate.denominator());
+}
+
+boost::rational<int> read_framerate(AVFormatContext& context, const boost::rational<int>& fail_value)
+{
+       auto video_index = av_find_best_stream(&context, AVMEDIA_TYPE_VIDEO, -1, -1, 0, 0);
+       auto audio_index = av_find_best_stream(&context, AVMEDIA_TYPE_AUDIO, -1, -1, 0, 0);
+
+       if (video_index > -1)
+       {
+               const auto video_context = context.streams[video_index]->codec;
+               const auto video_stream = context.streams[video_index];
+
+               auto frame_rate_time_base = video_stream->avg_frame_rate;
+               std::swap(frame_rate_time_base.num, frame_rate_time_base.den);
+
+               if (is_sane_fps(frame_rate_time_base))
+               {
+                       return boost::rational<int>(frame_rate_time_base.den, frame_rate_time_base.num);
+               }
+
+               AVRational time_base = video_context->time_base;
+
+               if (boost::filesystem::path(context.filename).extension().string() == ".flv")
+               {
+                       try
+                       {
+                               auto meta = read_flv_meta_info(context.filename);
+                               return boost::rational<int>(static_cast<int>(boost::lexical_cast<double>(meta["framerate"]) * 1000000.0), 1000000);
+                       }
+                       catch (...)
+                       {
+                               return fail_value;
+                       }
+               }
+               else
+               {
+                       time_base.num *= video_context->ticks_per_frame;
+
+                       if (!is_sane_fps(time_base))
+                       {
+                               time_base = fix_time_base(time_base);
+
+                               if (!is_sane_fps(time_base) && audio_index > -1)
+                               {
+                                       auto& audio_context = *context.streams[audio_index]->codec;
+                                       auto& audio_stream = *context.streams[audio_index];
+
+                                       double duration_sec = audio_stream.duration / static_cast<double>(audio_context.sample_rate);
+
+                                       time_base.num = static_cast<int>(duration_sec*100000.0);
+                                       time_base.den = static_cast<int>(video_stream->nb_frames * 100000);
+                               }
+                       }
+               }
+
+               boost::rational<int> fps(time_base.den, time_base.num);
+               boost::rational<int> closest_fps(0);
+
+               for (auto video_mode : enum_constants<core::video_format>())
+               {
+                       auto format = core::video_format_desc(core::video_format(video_mode));
+
+                       auto diff1 = boost::abs(boost::rational<int>(format.time_scale, format.duration) - fps);
+                       auto diff2 = boost::abs(closest_fps - fps);
+
+                       if (diff1 < diff2)
+                               closest_fps = boost::rational<int>(format.time_scale, format.duration);
+               }
+
+               return closest_fps;
+       }
+
+       return fail_value;
+}
+
+void fix_meta_data(AVFormatContext& context)
+{
+       auto video_index = av_find_best_stream(&context, AVMEDIA_TYPE_VIDEO, -1, -1, 0, 0);
+
+       if(video_index > -1)
+       {
+               auto video_stream   = context.streams[video_index];
+               auto video_context  = context.streams[video_index]->codec;
+
+               if(boost::filesystem::path(context.filename).extension().string() == ".flv")
+               {
+                       try
+                       {
+                               auto meta = read_flv_meta_info(context.filename);
+                               double fps = boost::lexical_cast<double>(meta["framerate"]);
+                               video_stream->nb_frames = static_cast<int64_t>(boost::lexical_cast<double>(meta["duration"])*fps);
+                       }
+                       catch(...){}
+               }
+               else
+               {
+                       auto stream_time = video_stream->time_base;
+                       auto duration    = video_stream->duration;
+                       auto codec_time  = video_context->time_base;
+                       auto ticks               = video_context->ticks_per_frame;
+
+                       if(video_stream->nb_frames == 0)
+                               video_stream->nb_frames = (duration*stream_time.num*codec_time.den)/(stream_time.den*codec_time.num*ticks);
+               }
+       }
+}
+
+spl::shared_ptr<AVPacket> create_packet()
+{
+       spl::shared_ptr<AVPacket> packet(new AVPacket(), [](AVPacket* p)
+       {
+               av_free_packet(p);
+               delete p;
+       });
+
+       av_init_packet(packet.get());
+       return packet;
+}
+
+spl::shared_ptr<AVFrame> create_frame()
+{
+       spl::shared_ptr<AVFrame> frame(av_frame_alloc(), [](AVFrame* p)
+       {
+               av_frame_free(&p);
+       });
+       return frame;
+}
+
+std::shared_ptr<core::mutable_audio_buffer> flush_audio()
+{
+       static std::shared_ptr<core::mutable_audio_buffer> audio(new core::mutable_audio_buffer());
+       return audio;
+}
+
+std::shared_ptr<core::mutable_audio_buffer> empty_audio()
+{
+       static std::shared_ptr<core::mutable_audio_buffer> audio(new core::mutable_audio_buffer());
+       return audio;
+}
+
+std::shared_ptr<AVFrame> flush_video()
+{
+       static auto video = create_frame();
+       return video;
+}
+
+std::shared_ptr<AVFrame> empty_video()
+{
+       static auto video = create_frame();
+       return video;
+}
+
+spl::shared_ptr<AVCodecContext> open_codec(AVFormatContext& context, enum AVMediaType type, int& index, bool single_threaded)
+{
+       AVCodec* decoder;
+       index = THROW_ON_ERROR2(av_find_best_stream(&context, type, index, -1, &decoder, 0), "");
+       //if(strcmp(decoder->name, "prores") == 0 && decoder->next && strcmp(decoder->next->name, "prores_lgpl") == 0)
+       //      decoder = decoder->next;
+
+       THROW_ON_ERROR2(tbb_avcodec_open(context.streams[index]->codec, decoder, single_threaded), "");
+       return spl::shared_ptr<AVCodecContext>(context.streams[index]->codec, tbb_avcodec_close);
+}
+
+spl::shared_ptr<AVFormatContext> open_input(const std::wstring& filename)
+{
+       AVFormatContext* weak_context = nullptr;
+       THROW_ON_ERROR2(avformat_open_input(&weak_context, u8(filename).c_str(), nullptr, nullptr), filename);
+       spl::shared_ptr<AVFormatContext> context(weak_context, [](AVFormatContext* p)
+       {
+               avformat_close_input(&p);
+       });
+       THROW_ON_ERROR2(avformat_find_stream_info(weak_context, nullptr), filename);
+       fix_meta_data(*context);
+       return context;
+}
+
+std::wstring print_mode(int width, int height, double fps, bool interlaced)
+{
+       std::wostringstream fps_ss;
+       fps_ss << std::fixed << std::setprecision(2) << (!interlaced ? fps : 2.0 * fps);
+
+       return boost::lexical_cast<std::wstring>(width) + L"x" + boost::lexical_cast<std::wstring>(height) + (!interlaced ? L"p" : L"i") + fps_ss.str();
+}
+
+bool is_valid_file(const std::wstring& filename, bool only_video)
+{
+       static const auto invalid_exts = {
+               L".png",
+               L".tga",
+               L".bmp",
+               L".jpg",
+               L".jpeg",
+               L".gif",
+               L".tiff",
+               L".tif",
+               L".jp2",
+               L".jpx",
+               L".j2k",
+               L".j2c",
+               L".swf",
+               L".ct"
+       };
+       static const auto only_audio = {
+               L".mp3",
+               L".wav",
+               L".wma"
+       };
+       static const auto valid_exts = {
+               L".m2t",
+               L".mov",
+               L".mp4",
+               L".dv",
+               L".flv",
+               L".mpg",
+               L".dnxhd",
+               L".h264",
+               L".prores",
+               L".mkv",
+               L".mxf"
+       };
+
+       auto ext = boost::to_lower_copy(boost::filesystem::path(filename).extension().wstring());
+
+       if(std::find(valid_exts.begin(), valid_exts.end(), ext) != valid_exts.end())
+               return true;
+
+       if (!only_video && std::find(only_audio.begin(), only_audio.end(), ext) != only_audio.end())
+               return true;
+
+       if(std::find(invalid_exts.begin(), invalid_exts.end(), ext) != invalid_exts.end())
+               return false;
+
+       if (only_video && std::find(only_audio.begin(), only_audio.end(), ext) != only_audio.end())
+               return false;
+
+       auto u8filename = u8(filename);
+
+       int score = 0;
+       AVProbeData pb = {};
+       pb.filename = u8filename.c_str();
+
+       if(av_probe_input_format2(&pb, false, &score) != nullptr)
+               return true;
+
+       std::ifstream file(u8filename);
+
+       std::vector<unsigned char> buf;
+       for(auto file_it = std::istreambuf_iterator<char>(file); file_it != std::istreambuf_iterator<char>() && buf.size() < 1024; ++file_it)
+               buf.push_back(*file_it);
+
+       if(buf.empty())
+               return false;
+
+       pb.buf          = buf.data();
+       pb.buf_size = static_cast<int>(buf.size());
+
+       return av_probe_input_format2(&pb, true, &score) != nullptr;
+}
+
+bool try_get_duration(const std::wstring filename, std::int64_t& duration, boost::rational<std::int64_t>& time_base)
+{
+       AVFormatContext* weak_context = nullptr;
+       if (avformat_open_input(&weak_context, u8(filename).c_str(), nullptr, nullptr) < 0)
+               return false;
+
+       std::shared_ptr<AVFormatContext> context(weak_context, [](AVFormatContext* p)
+       {
+               avformat_close_input(&p);
+       });
+
+       context->probesize = context->probesize / 10;
+       context->max_analyze_duration = context->probesize / 10;
+
+       if (avformat_find_stream_info(context.get(), nullptr) < 0)
+               return false;
+
+       const auto fps = read_fps(*context, 1.0);
+
+       const auto rational_fps = boost::rational<std::int64_t>(static_cast<int>(fps * AV_TIME_BASE), AV_TIME_BASE);
+
+       duration = boost::rational_cast<std::int64_t>(context->duration * rational_fps / AV_TIME_BASE);
+
+       if (rational_fps == 0)
+               return false;
+
+       time_base = 1 / rational_fps;
+
+       return true;
+}
+
+std::wstring probe_stem(const std::wstring& stem, bool only_video)
+{
+       auto stem2 = boost::filesystem::path(stem);
+       auto parent = find_case_insensitive(stem2.parent_path().wstring());
+
+       if (!parent)
+               return L"";
+
+       auto dir = boost::filesystem::path(*parent);
+
+       for(auto it = boost::filesystem::directory_iterator(dir); it != boost::filesystem::directory_iterator(); ++it)
+       {
+               if(boost::iequals(it->path().stem().wstring(), stem2.filename().wstring()) && is_valid_file(it->path().wstring(), only_video))
+                       return it->path().wstring();
+       }
+       return L"";
+}
+
+core::audio_channel_layout get_audio_channel_layout(int num_channels, std::uint64_t layout, const std::wstring& channel_layout_spec)
+{
+       if (!channel_layout_spec.empty())
+       {
+               if (boost::contains(channel_layout_spec, L":")) // Custom on the fly layout specified.
+               {
+                       std::vector<std::wstring> type_and_channel_order;
+                       boost::split(type_and_channel_order, channel_layout_spec, boost::is_any_of(L":"), boost::algorithm::token_compress_off);
+                       auto& type                      = type_and_channel_order.at(0);
+                       auto& order                     = type_and_channel_order.at(1);
+
+                       return core::audio_channel_layout(num_channels, std::move(type), order);
+               }
+               else // Preconfigured named channel layout selected.
+               {
+                       auto channel_layout = core::audio_channel_layout_repository::get_default()->get_layout(channel_layout_spec);
+
+                       if (!channel_layout)
+                               CASPAR_THROW_EXCEPTION(user_error() << msg_info(L"No channel layout with name " + channel_layout_spec + L" registered"));
+
+                       channel_layout->num_channels = num_channels;
+
+                       return *channel_layout;
+               }
+       }
+
+       if (!layout)
+       {
+               if (num_channels == 1)
+                       return core::audio_channel_layout(num_channels, L"mono", L"FC");
+               else if (num_channels == 2)
+                       return core::audio_channel_layout(num_channels, L"stereo", L"FL FR");
+               else
+                       return core::audio_channel_layout(num_channels, L"", L""); // Passthru without named channels as is.
+       }
+
+       // What FFmpeg calls "channel layout" is only the "layout type" of a channel layout in
+       // CasparCG where the channel layout supports different orders as well.
+       // The user needs to provide additional mix-configs in casparcg.config to support more
+       // than the most common (5.1, mono and stereo) types.
+
+       // Based on information in https://ffmpeg.org/ffmpeg-utils.html#Channel-Layout
+       switch (layout)
+       {
+       case AV_CH_LAYOUT_MONO:
+               return core::audio_channel_layout(num_channels, L"mono",                        L"FC");
+       case AV_CH_LAYOUT_STEREO:
+               return core::audio_channel_layout(num_channels, L"stereo",                      L"FL FR");
+       case AV_CH_LAYOUT_2POINT1:
+               return core::audio_channel_layout(num_channels, L"2.1",                         L"FL FR LFE");
+       case AV_CH_LAYOUT_SURROUND:
+               return core::audio_channel_layout(num_channels, L"3.0",                         L"FL FR FC");
+       case AV_CH_LAYOUT_2_1:
+               return core::audio_channel_layout(num_channels, L"3.0(back)",           L"FL FR BC");
+       case AV_CH_LAYOUT_4POINT0:
+               return core::audio_channel_layout(num_channels, L"4.0",                         L"FL FR FC BC");
+       case AV_CH_LAYOUT_QUAD:
+               return core::audio_channel_layout(num_channels, L"quad",                        L"FL FR BL BR");
+       case AV_CH_LAYOUT_2_2:
+               return core::audio_channel_layout(num_channels, L"quad(side)",          L"FL FR SL SR");
+       case AV_CH_LAYOUT_3POINT1:
+               return core::audio_channel_layout(num_channels, L"3.1",                         L"FL FR FC LFE");
+       case AV_CH_LAYOUT_5POINT0_BACK:
+               return core::audio_channel_layout(num_channels, L"5.0",                         L"FL FR FC BL BR");
+       case AV_CH_LAYOUT_5POINT0:
+               return core::audio_channel_layout(num_channels, L"5.0(side)",           L"FL FR FC SL SR");
+       case AV_CH_LAYOUT_4POINT1:
+               return core::audio_channel_layout(num_channels, L"4.1",                         L"FL FR FC LFE BC");
+       case AV_CH_LAYOUT_5POINT1_BACK:
+               return core::audio_channel_layout(num_channels, L"5.1",                         L"FL FR FC LFE BL BR");
+       case AV_CH_LAYOUT_5POINT1:
+               return core::audio_channel_layout(num_channels, L"5.1(side)",           L"FL FR FC LFE SL SR");
+       case AV_CH_LAYOUT_6POINT0:
+               return core::audio_channel_layout(num_channels, L"6.0",                         L"FL FR FC BC SL SR");
+       case AV_CH_LAYOUT_6POINT0_FRONT:
+               return core::audio_channel_layout(num_channels, L"6.0(front)",          L"FL FR FLC FRC SL SR");
+       case AV_CH_LAYOUT_HEXAGONAL:
+               return core::audio_channel_layout(num_channels, L"hexagonal",           L"FL FR FC BL BR BC");
+       case AV_CH_LAYOUT_6POINT1:
+               return core::audio_channel_layout(num_channels, L"6.1",                         L"FL FR FC LFE BC SL SR");
+       case AV_CH_LAYOUT_6POINT1_BACK:
+               return core::audio_channel_layout(num_channels, L"6.1(back)",           L"FL FR FC LFE BL BR BC");
+       case AV_CH_LAYOUT_6POINT1_FRONT:
+               return core::audio_channel_layout(num_channels, L"6.1(front)",          L"FL FR LFE FLC FRC SL SR");
+       case AV_CH_LAYOUT_7POINT0:
+               return core::audio_channel_layout(num_channels, L"7.0",                         L"FL FR FC BL BR SL SR");
+       case AV_CH_LAYOUT_7POINT0_FRONT:
+               return core::audio_channel_layout(num_channels, L"7.0(front)",          L"FL FR FC FLC FRC SL SR");
+       case AV_CH_LAYOUT_7POINT1:
+               return core::audio_channel_layout(num_channels, L"7.1",                         L"FL FR FC LFE BL BR SL SR");
+       case AV_CH_LAYOUT_7POINT1_WIDE_BACK:
+               return core::audio_channel_layout(num_channels, L"7.1(wide)",           L"FL FR FC LFE BL BR FLC FRC");
+       case AV_CH_LAYOUT_7POINT1_WIDE:
+               return core::audio_channel_layout(num_channels, L"7.1(wide-side)",      L"FL FR FC LFE FLC FRC SL SR");
+       case AV_CH_LAYOUT_STEREO_DOWNMIX:
+               return core::audio_channel_layout(num_channels, L"downmix",                     L"DL DR");
+       default:
+               // Passthru
+               return core::audio_channel_layout(num_channels, L"", L"");
+       }
+}
+
+// av_get_default_channel_layout does not work for layouts not predefined in ffmpeg. This is needed to support > 8 channels.
+std::uint64_t create_channel_layout_bitmask(int num_channels)
+{
+       if (num_channels > 63)
+               CASPAR_THROW_EXCEPTION(invalid_argument() << msg_info(L"FFmpeg cannot handle more than 63 audio channels"));
+
+       const auto ALL_63_CHANNELS = 0x7FFFFFFFFFFFFFFFULL;
+
+       auto to_shift = 63 - num_channels;
+       auto result = ALL_63_CHANNELS >> to_shift;
+
+       return static_cast<std::uint64_t>(result);
+}
+
+std::string to_string(const boost::rational<int>& framerate)
+{
+       return boost::lexical_cast<std::string>(framerate.numerator())
+               + "/" + boost::lexical_cast<std::string>(framerate.denominator()) + " (" + boost::lexical_cast<std::string>(static_cast<double>(framerate.numerator()) / static_cast<double>(framerate.denominator())) + ") fps";
+}
+
+std::vector<int> find_audio_cadence(const boost::rational<int>& framerate)
+{
+       static std::map<boost::rational<int>, std::vector<int>> CADENCES_BY_FRAMERATE = []
+       {
+               std::map<boost::rational<int>, std::vector<int>> result;
+
+               for (core::video_format format : enum_constants<core::video_format>())
+               {
+                       core::video_format_desc desc(format);
+                       boost::rational<int> format_rate(desc.time_scale, desc.duration);
+
+                       result.insert(std::make_pair(format_rate, desc.audio_cadence));
+               }
+
+               return result;
+       }();
+
+       auto exact_match = CADENCES_BY_FRAMERATE.find(framerate);
+
+       if (exact_match != CADENCES_BY_FRAMERATE.end())
+               return exact_match->second;
+
+       boost::rational<int> closest_framerate_diff = std::numeric_limits<int>::max();
+       boost::rational<int> closest_framerate = 0;
+
+       for (auto format_framerate : CADENCES_BY_FRAMERATE | boost::adaptors::map_keys)
+       {
+               auto diff = boost::abs(framerate - format_framerate);
+
+               if (diff < closest_framerate_diff)
+               {
+                       closest_framerate_diff = diff;
+                       closest_framerate = format_framerate;
+               }
+       }
+
+       if (is_logging_quiet_for_thread())
+               CASPAR_LOG(debug) << "No exact audio cadence match found for framerate " << to_string(framerate)
+               << "\nClosest match is " << to_string(closest_framerate)
+               << "\nwhich is a " << to_string(closest_framerate_diff) << " difference.";
+       else
+               CASPAR_LOG(warning) << "No exact audio cadence match found for framerate " << to_string(framerate)
+               << "\nClosest match is " << to_string(closest_framerate)
+               << "\nwhich is a " << to_string(closest_framerate_diff) << " difference.";
+
+       return CADENCES_BY_FRAMERATE[closest_framerate];
+}
+
+//
+//void av_dup_frame(AVFrame* frame)
+//{
+//     AVFrame* new_frame = avcodec_alloc_frame();
+//
+//
+//     const uint8_t *src_data[4] = {0};
+//     memcpy(const_cast<uint8_t**>(&src_data[0]), frame->data, 4);
+//     const int src_linesizes[4] = {0};
+//     memcpy(const_cast<int*>(&src_linesizes[0]), frame->linesize, 4);
+//
+//     av_image_alloc(new_frame->data, new_frame->linesize, new_frame->width, new_frame->height, frame->format, 16);
+//
+//     av_image_copy(new_frame->data, new_frame->linesize, src_data, src_linesizes, frame->format, new_frame->width, new_frame->height);
+//
+//     frame =
+//}
+
+}}