Lets a camera's own H.264 or MJPEG go straight onto the network on Windows as
it already could on Linux, so nothing has to be decoded and re-encoded.

Finding out what a camera offers takes rather more than the single v4l2 ioctl
it takes on Linux: the camera has to be instantiated, its output pins found,
and each pin's stream capabilities walked through IAMStreamConfig, reading the
FourCC out of the BITMAPINFOHEADER in every media type. Asking for the result
is spelled differently too - dshow has no input_format option and takes the
choice from AVFormatContext::video_codec_id instead, which is what
"-vcodec mjpeg" ahead of "-i" sets, and which has to be set on a context
allocated by us rather than by avformat_open_input().

Opening now loosens its constraints a step at a time rather than giving up on
the format at the first refusal. Listing a format, producing it at a given
size, and producing it at a given frame rate are three separate claims, and
cameras honour them independently: an integrated webcam here offers MJPEG but
not at 640x480 15fps, and was falling all the way back to re-encoding. The
frame rate is given up first, then the size, and the format last, since the
format is the one that decides whether anything has to be re-encoded at all.
What the camera settles on is what gets announced to the peer, so this changes
what is sent rather than what is claimed about it.

That relaxation exposed an assumption that was already wrong: the encoder was
hard-wired to kFrameRate with a plain incrementing pts, while the pre-existing
"let the camera choose" fallback had always been able to open without a frame
rate at all. A camera delivering 30fps therefore produced a stream whose
timestamps claimed half that. It now takes the rate from the opened stream,
and the keyframe interval with it, so two seconds between keyframes stays two
seconds - kGopSize is gone, a fixed frame count no longer meaning a fixed
duration.

Every FourCC a camera reports is logged, mapped or not. An integrated camera
offering nothing but uncompressed types while advertising MJPEG under v4l2
would have meant the Windows Camera Frame Server had already decoded it, with
nothing to be done from here - worth being able to tell apart from a FourCC we
simply failed to map, which is what this turned out not to be.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CBqDgvq3Ty6C6BjLFH16bw

Capture already went through FFmpeg's "dshow" demuxer, which does the job -
unlike the audio case there is nothing here the platform offers that we cannot
reach any other way, so there is no reason to drive DirectShow ourselves. Two
things were missing to make it work.

Enumeration, because the demuxer has none: it can only print device names to
the log with -list_devices, which is no use to a settings dialog. The list is
built in videodshow.cpp instead, walking the same CLSID_VideoInputDeviceCategory
the demuxer walks, so the names handed back are the ones it matches on. Where a
friendly name is not unique - two identical cameras - later duplicates fall
back to the device path, which the demuxer also accepts.

And the "video=" prefix, since DirectShow namespaces its devices by type and
the same name behind "audio=" would be a microphone. The name is stored
unprefixed and decorated at open time, so what the settings dialog shows and
writes stays readable.

Also adds a last-resort open with no constraints at all. DirectShow refuses a
video_size or framerate it cannot produce exactly rather than settling for the
nearest, so a camera with an unusual set of modes would otherwise give no
picture; whatever comes back is scaled by the transcode loop anyway.

The failure message on Windows now names the likely cause. A camera that
enumerated but will not open is usually not a broken camera: Windows lists
devices it then refuses to instantiate, which is what the "Let desktop apps
access your camera" privacy setting being off looks like from here, and what
another application holding the camera looks like too. None of them are
distinguishable from a real fault at that level, and the bare "I/O error" sent
me looking at the wrong layer entirely.

compressedFormats is left empty on Windows, so "prefer compressed formats" has
no effect there - finding out what a DirectShow camera can deliver means
walking its output pin's media types through IAMStreamConfig, which is a good
deal more than the one v4l2 ioctl that fills this in on Linux.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CBqDgvq3Ty6C6BjLFH16bw