omtsndcli_v01

omtsndcli_v01 is the current documented Linux OMT sender CLI snapshot for the sender PC project.

It targets a direct capture path:

Linux sender PC -> Magewell Pro Capture HDMI -> V4L2 / ALSA -> OMT sender

The current source snapshot is based on the proven working implementation from omtcli06_fix.cpp, renamed and documented for repository use.

Current validated state

Validated in the project handoff:

• Debian 13 sender PC
• host qduv1
• 4× Magewell Pro Capture HDMI visible as /dev/video0 to /dev/video3
• OMT stack already built: libvmx, libomtnet, libomt
• verified against vMix:
  • discovery works
  • test video works
  • test audio works
  • live V4L2 video works
  • live ALSA HDMI audio works

What the program does

• sends video from:
  • built-in testcard
  • Linux V4L2 capture device
• sends audio from:
  • built-in test noise generator
  • ALSA capture device
• exposes module help for video/audio backends
• classifies V4L2 formats as passthrough, conversion-only, or unsupported
• maps user bitrate options onto OMT quality enums
• prints runtime video/audio statistics every 5 seconds
• prints sender health roughly every 60 seconds
• keeps rolling interval data and prints a last-24h summary on exit
• stops cleanly on SIGINT / SIGTERM

Source file

Primary repo source:

• omtsndcli_v01.cpp

Lineage source used for this snapshot:

• omtcli06_fix.cpp

Dependencies

System / project dependencies expected by the current source:

• C++17 compiler
• ALSA development headers (alsa/asoundlib.h)
• Linux V4L2 headers (linux/videodev2.h)
• OMT C/C++ wrapper header: libomt.h
• linkable libomt
• linkable libasound

Build

Example build command used by the current source layout:

clang++ -O3 -std=c++17 -o omtsndcli_v01 omtsndcli_v01.cpp \
    -L. -lomt -lasound -Wl,-rpath,'$ORIGIN'

Reference OMT library build sequence from the validated environment:

cd ~/omt/libvmx/build && bash ./buildlinuxx64.sh
cd ~/omt/libomtnet     && dotnet build -c Release
cd ~/omt/libomt/build  && bash ./buildlinuxx64.sh

Usage

./omtsndcli_v01 [-t video_source] [-s audio_source]

General help

./omtsndcli_v01 -h

Video modules

./omtsndcli_v01 -t help
./omtsndcli_v01 -t testcard:help
./omtsndcli_v01 -t v4l2:help

Audio modules

./omtsndcli_v01 -s help
./omtsndcli_v01 -s testcard:help
./omtsndcli_v01 -s alsa:help

Examples

Testcard video only:

./omtsndcli_v01 -t testcard

Testcard video + testcard audio:

./omtsndcli_v01 -t testcard -s testcard

Magewell / V4L2 video only:

./omtsndcli_v01 -t v4l2:device=/dev/video0

Magewell video with explicit name / format / quality:

./omtsndcli_v01 -t v4l2:device=/dev/video0:name=Cam1:format=YUY2:b=mid

Magewell video + ALSA audio:

./omtsndcli_v01 -t v4l2:device=/dev/video0 -s alsa:hw:2,0

V4L2 with control override example:

./omtsndcli_v01 -t v4l2:device=/dev/video0:ctrl.brightness=128

CLI parameter model

Style:

driver:param=value:param=value

Notable aliases:

• device= or d=
• format= or fmt=
• bitrate= or b=

Supported source modules

Video

testcard

Built-in moving test image. When compatible with 1920x1080 UYVY, it can also try to load:

• california-1080-uyvy.yuv

Syntax:

-t testcard[:name=Testcard][:width=1920][:height=1080][:fps=60][:format=UYVY][:file=california-1080-uyvy.yuv]

Supported testcard output formats in the current code:

• UYVY
• YUY2

v4l2

Linux V4L2 capture input, intended for Magewell cards.

Syntax:

-t v4l2:device=/dev/video0[:name=Cam1][:width=1920][:height=1080][:fps=50][:format=YUY2]
-t v4l2:device=/dev/video0:ctrl.<control_name>=<value>

Module help enumerates:

• available /dev/video* devices
• formats
• sizes
• frame rates
• normalized control names

The sender classifies formats into:

• OMT passthrough
• conversion-only
• unsupported

Formats currently handled in code include these categories:

• passthrough: YUYV/YUY2, UYVY, NV12, YV12, BGR32/BGRA
• conversion to BGRA: RGB32, RGB24, BGR24
• explicitly not forwarded as-is: YUV420/YU12/I420

Audio modules

testcard

Built-in audio generator using random noise.

Syntax:

-s testcard[:volume=-20][:rate=48000][:channels=2]

Notes:

• default volume is -20 dB
• current implementation expects 2 channels

alsa

ALSA capture device input.

Syntax:

-s alsa:hw:2,0
-s alsa:device=hw:2,0[:volume=0][:rate=48000][:channels=2]

The module help enumerates visible ALSA capture PCM devices.

Quality mapping

The current code accepts both labels and numeric aliases:

• low or 1 -> OMTQuality_Low
• mid, medium, 2, or default -> OMTQuality_Medium
• high or 3 -> OMTQuality_High

Startup prints the resolved OMT quality explicitly.

Runtime output and diagnostics

The sender prints:

• selected video source
• selected audio source
• stream name
• resolved OMT quality
• audio frame size
• capture startup information
• 5-second interval stats
• OMT sender connection / tally / totals
• 60-second sender health line
• final rolling summary for the last 24 hours

Sender health fields:

• captured = frames obtained from capture
• sent = frames accepted by omt_send
• dropped = send failures
• late = omt_send exceeded frame-time budget
• omt_send avg/max = mean and peak send duration in milliseconds

Important interpretation:

• late does not automatically mean dropped
• on the validated 1080p50 path, occasional late events can happen while the pipeline still runs correctly

Known tested observations from handoff

Observed/recorded bitrate behavior in earlier testing:

• low ≈ 75 Mbit/s
• medium ≈ 136 Mbit/s
• high ≈ 136 Mbit/s

Interpretation used in the project notes:

• medium and high looked similar on the tested content
• that was treated as an observation to re-measure, not yet as a confirmed encoder bug

Signal handling and shutdown

The current source handles:

• Ctrl+C / SIGINT
• SIGTERM

On shutdown it:

• stops V4L2 streaming
• unmaps capture buffers
• closes ALSA capture
• frees frame buffers
• prints the rolling summary
• destroys the OMT sender

Recommended next steps in the project

1. treat omtsndcli_v01.cpp as the canonical repo source
2. re-measure medium vs. high quality behavior
3. archive a stable release snapshot
4. then proceed to multi-instance execution and systemd services

Notes

This repository snapshot is meant to preserve the current known-good implementation state in a cleaner, documented form for GitHub. It does not claim that packaging, install scripts, service units, or multi-instance orchestration are finished.

Licensing

This repository can be published under the MIT License for the project’s own code.

Why this looks safe for the current tree:

• the checked-in application source directly depends on libomt.h, ALSA, and Linux V4L2 headers
• the project notes say the sender lineage began from the OMT C++ example sender
• the Open Media Transport organization lists libomtnet, libomt, libvmx, and Examples as MIT-licensed repositories
• alsa-lib is LGPL-2.1
• videodev2.h is part of the Linux V4L2 userspace API and is published under ((GPL-2.0+ WITH Linux-syscall-note) OR BSD-3-Clause)

Practical implications:

• your application source may be MIT
• OMT libraries remain under their own MIT licenses
• ALSA remains under LGPL-2.1; dynamic linking is the normal path on Linux and does not force your own source code to become LGPL, but binary redistribution should still respect LGPL obligations
• if you later commit vendored copies of libomt.h, libomt.so, libvmx.so, or other third-party files into the repository, keep their original license texts/notices with them
• no Magewell SDK headers are used in the current source snapshot; capture is done through generic V4L2/ALSA interfaces

This is a technical license check for the current repository contents, not formal legal advice.