ffplayout/assets/ffplayout.yml
2022-03-12 22:18:49 +01:00

152 lines
6.4 KiB
YAML

general:
helptext: Sometimes it can happen, that a file is corrupt but still playable,
this can produce an streaming error over all following files. The only way
in this case is, to stop ffplayout and start it again. Here we only say when
it stops, the starting process is in your hand. Best way is a systemd service
on linux. 'stop_threshold' stop ffplayout, if it is async in time above this
value. A number below 3 can cause unexpected errors.
stop_threshold: 11
mail:
helptext: Send error messages to email address, like missing playlist; invalid
json format; missing clip path. Leave recipient blank, if you don't need this.
'mail_level' can be WARNING or ERROR.
subject: "Playout Error"
smtp_server: "mail.example.org"
starttls: true
sender_addr: "ffplayout@example.org"
sender_pass: "abc123"
recipient:
mail_level: "ERROR"
logging:
helptext: Logging to file, if 'log_to_file' false log to console. 'backup_count'
says how long log files will be saved in days. 'local_time' to false will set
log timestamps to UTC. Path to /var/log/ only if you run this program as daemon.
'log_level' can be DEBUG, INFO, WARNING, ERROR. 'ffmpeg_level' can be info,
warning, error.
log_to_file: false
backup_count: 7
local_time: true
log_path: "/var/log/ffplayout/"
log_level: "DEBUG"
ffmpeg_level: "error"
processing:
helptext: Set playing mode, like playlist; folder, or you own custom one.
Default processing, for all clips that they get prepared in that way,
so the output is unique. 'aspect' must be a float number. 'logo' is only used
if the path exist. 'logo_scale' scale the logo to target size, leave it blank
when no scaling is needed, format is 'number:number', for example '100:-1'
for proportional scaling. With 'logo_opacity' logo can become transparent.
With 'logo_filter' 'overlay=W-w-12:12' you can modify the logo position.
With 'use_loudnorm' you can activate single pass EBU R128 loudness normalization.
'loud_*' can adjust the loudnorm filter. 'output_count' sets the outputs for
the filtering, > 1 gives the option to use the same filters for multiple outputs.
This outputs can be taken in 'stream_param', names will be vout2, vout3;
aout2, aout2 etc.
mode: playlist
width: 1024
height: 576
aspect: 1.778
fps: 25
add_logo: true
logo: "docs/logo.png"
logo_scale:
logo_opacity: 0.7
logo_filter: "overlay=W-w-12:12"
add_loudnorm: false
loud_i: -18
loud_tp: -1.5
loud_lra: 11
output_count: 1
volume: 1
ingest:
helptext: Works not with direct hls output, it always needs full processing! Run a server
for a ingest stream. This stream will override the normal streaming until is done.
There is no authentication, this is up to you. The recommend way is to set address to localhost, stream to a local server with authentication and from there stream to this app.
enable: false
stream_input: [-f, live_flv, -listen, 1, -i, rtmp://localhost:1936/live/stream]
playlist:
helptext: >
'path' can be a path to a single file, or a directory. For directory put
only the root folder, for example '/playlists', subdirectories are read by the
script. Subdirectories needs this structure '/playlists/2018/01'. 'day_start'
means at which time the playlist should start, leave day_start blank when playlist
should always start at the begin. 'length' represent the target length from
playlist, when is blank real length will not consider. 'infinit true' works with
single playlist file and loops it infinitely.
path: "/playlists"
day_start: "5:59:25"
length: "24:00:00"
infinit: false
storage:
helptext: Play ordered or randomly files from path. 'filler_clip' is for fill
the end to reach 24 hours, it will loop when is necessary. 'extensions' search
only files with this extension. Set 'shuffle' to 'True' to pick files randomly.
path: "/mediaStorage"
filler_clip: "/mediaStorage/filler/filler.mp4"
extensions:
- "mp4"
- "mkv"
shuffle: true
text:
helptext: Overlay text in combination with libzmq for remote text manipulation.
On windows fontfile path need to be like this 'C\:/WINDOWS/fonts/DejaVuSans.ttf'.
In a standard environment the filter drawtext node is Parsed_drawtext_2.
'over_pre' if True text will be overlay in pre processing. Continue same text
over multiple files is in that mode not possible. 'text_from_filename' activate the
extraction from text of a filename. With 'style' you can define the drawtext
parameters like position, color, etc. Post Text over API will override this.
With 'regex' you can format file names, to get a title from it.
add_text: false
over_pre: false
bind_address: "127.0.0.1:5555"
fontfile: "/usr/share/fonts/truetype/dejavu/DejaVuSans.ttf"
text_from_filename: false
style: "x=(w-tw)/2:y=(h-line_h)*0.9:fontsize=24:fontcolor=#ffffff:box=1:boxcolor=#000000:boxborderw=4"
regex: "^(.*)_"
out:
helptext: The final playout compression. Set the settings to your needs.
'mode' has the standard options 'desktop', 'hls', 'live_switch', 'stream'. Self made
outputs can be define, by adding script in output folder with an 'output' function
inside. 'preview' works only in streaming output and creates a separate preview stream.
mode: 'stream'
preview: false
preview_param:
[-s, 512x288,
-c:v, libx264,
-crf, 24,
-x264-params, keyint=50:min-keyint=25:scenecut=-1,
-maxrate, 800k,
-bufsize, 1600k,
-preset, ultrafast,
-tune, zerolatency,
-profile:v, Main,
-level, 3.1,
-c:a, aac,
-ar, 44100,
-b:a, 128k,
-flags, +global_header,
-f, flv, rtmp://preview.local/live/stream]
stream_param:
[-c:v, libx264,
-crf, 23,
-x264-params, keyint=50:min-keyint=25:scenecut=-1,
-maxrate, 1300k,
-bufsize, 2600k,
-preset, faster,
-tune, zerolatency,
-profile:v, Main,
-level, 3.1,
-c:a, aac,
-ar, 44100,
-b:a, 128k,
-flags, +global_header,
-f, flv, rtmp://localhost/live/stream]