@type option() ::
  {:output_path, Path.t()}
  | {:s3_upload_config, keyword()}
  | {:download_path, Path.t()}
  | {:s3_download_config, keyword()}
  | {:thumbnails_ctx, thumbnails_ctx()}
  | {:only_rids, [ExWebRTC.MediaStreamTrack.rid() | nil]}
  | {:reorder_buffer_size, pos_integer()}
  | {:reencode_ctx, reencode_ctx()}

Options that can be passed to convert!/2.

• :output_path - Directory where Converter will save its artifacts. ./converter/output by default.
• :s3_upload_config - If passed, processed recordings will be uploaded to S3-compatible storage. See ExWebRTC.Recorder.S3.upload_config/0 for more info.
• :download_path - Directory where Converter will save files fetched from S3. ./converter/download by default.
• :s3_download_config - Optional S3 config overrides used when fetching files. See ExWebRTC.Recorder.S3.override_config/0 for more info.
• :thumbnails_ctx - If passed, Converter will generate thumbnails for the output files. See thumbnails_ctx/0 for more info.
• :only_rids - By default, when processing a video track with multiple layers (i.e. simulcast), Converter generates multiple output files, one per layer. If passed, Converter will only process the layers with RIDs present in this list. E.g. if you want to receive a single video file from the layer "h", pass ["h"]. For single-layer tracks RID is set to nil, so if you want to handle both simulcast and regular tracks, pass ["h", nil].
• :reorder_buffer_size - Size of the buffer used for reordering late packets. 100 by default. Increasing this value may help with "Decoded late RTP packet" warnings, but keep in mind that larger values slow the conversion process considerably.
• :reencode_ctx - If passed, Converter will reencode the video using FFmpeg. The keyframe interval of video tracks sent over WebRTC may vary, so this is helpful when you want to generate additional ones, to facilitate accurate seeking in the result file during playback. Keep in mind that reenncoding is slow and resource-intensive. See reencode_ctx/0 for more info.

@type reencode_ctx() :: %{
  optional(:threads) => pos_integer(),
  optional(:bitrate) => String.t(),
  optional(:crf) => 4..63,
  optional(:gop_size) => pos_integer(),
  optional(:resolution) => resolution(),
  optional(:cues_to_front) => boolean()
}

Context for the video reencoding. See man ffmpeg for more details.

• :threads - How many threads to use. Unlimited by default.
• :bitrate - Video bitrate the VP8 encoder shall strive for. 1.5M by default.
• :crf - Output video quality (lower is better). 10 by default.
• :gop_size - Keyframe interval. 125 by default.
• :resolution - Output video resolution. Inferred from the first video frame by default. See resolution/0 for more info.
• :cues_to_front - Whether the muxer should put MKV Cues element at the front of the file, to aid with seeking e.g. when streaming the result file. true by default.

@type resolution() :: %{
  optional(:width) => pos_integer() | -1,
  optional(:height) => pos_integer() | -1
}

Resolution of reencoded video and generated thumbnails.

Make sure to provide at least one of the dimensions. Setting the other dimension to -1 will fit the image to the aspect ratio.

@type thumbnails_ctx() :: %{resolution: resolution()}

Context for the thumbnail generation.

• :resolution (required) - Thumbnail size. See resolution/0 for more info.