USAGE.md

TOP-LEVEL

Automatic Thumbnail Generation

AlabamaEncoder will gather multiple statistics about each frame,
to then pick nine frames that will make the most appealing thumbnail.
Generated thumbnails will be saved in the same directory as [INPUT FILE]

Usage:

alabamaEncoder autothumbnailer [INPUT FILE] [flags]

Additional Thumbnail Generation Flags

Parameter	Description
--detect_faces	Detect faces in video

Worker

Commands for interfacing with celery, the multi-system part of AlabamaEncoder

Start a worker:

alabamaEncoder worker [FLAGS]

Additional flags:

Parameter	Description
--workers int	Amount of multiprocess workers

Clear the celery queue:

alabamaEncoder clear

Encoder

Encode video using AlabamaEncoder!

Usage:

alabamaEncoder encode [-h] [INPUT FILE] [OUTPUT FILE] [flags]

Encode flags

Parameter	Description
-h, --help	Show the help message and exit
--dont_encode_audio	Do not mux audio
--audio_params AUDIO_PARAMS	Audio params
--celery	Encode on a Celery cluster, which is at localhost
--autocrop	Automatically crop the video
--video_filters VIDEO_FILTERS	Override the crop, put your vf ffmpeg filter there (e.g., scale=-2:1080:flags=lanczos,zscale=t=linear...) Make sure ffmpeg on all workers has support for the filters you use
--bitrate BITRATE	Bitrate to use, auto for auto bitrate selection
--overshoot BITRATE_OVERSHOOT	How much the vbr_perchunk_optimisation is allowed to overshoot
--undershoot BITRATE_UNDERSHOOT	How much the vbr_perchunk_optimisation is allowed to undershoot
--vbr_perchunk_optimisation	Enable automatic bitrate optimisation per chunk
--multiprocess_workers MULTIPROCESS_WORKERS	Number of workers to use for multiprocessing, if -1 the program will auto-scale
--ssim-db-target SSIM_DB_TARGET	What SSIM dB to target when using auto bitrate, not recommended to set manually, otherwise 21.2 is a good starting point
--crf CRF	What CRF (Constant Rate Factor) to use (must be in range 0..=255)
--encoder {X265,SVT_AV1,AOMENC,X264,VPX_VP9,VPX_VP8,VAAPI_H265,VAAPI_H264,RAV1E,NVENC_H264}	What encoder to use
--grain GRAIN	Manually give the grainsynth value, 0 to disable, -1 for auto, -2 for auto per scene, -3 for new fast per scene grain
--vmaf_target VMAF_TARGET	What VMAF (Video Multi-method Assessment Fusion) to target when using bitrate auto
--max_scene_length MAX_SCENE_LENGTH	If a scene is longer than this, it will recursively cut it in the middle until each chunk is within the max
--statically_sized_scenes	Instead of performing scene detection, do statically sized scenes at about 30 seconds
--scene_merge	Merge scenes until they meet the max scene length
--no_crf_based_vmaf_targeting	Per chunk, find a CRF that hits the target quality and encode using that
--auto_crf	Find a CRF that hits the target VMAF, calculate a peak bitrate cap, and encode using that
--chunk_order {random,sequential,length_desc,length_asc,reverse,even}	Encode chunks in a specific order
--start_offset START_OFFSET	Offset from the beginning of the video (in seconds), useful for cutting intros, etc.
--end_offset END_OFFSET	Offset from the end of the video (in seconds), useful for cutting end credits, outtros, etc.
--bitrate_adjust_mode {chunk,global}	Do a complexity analysis on each chunk individually and adjust bitrate based on that (can overshoot/undershoot a lot), or do complexity analysis on all chunks ahead of time and budget it to hit the target by normalizing the bitrate
--log_level LOG_LEVEL	Set the log level, 0 for silent, 1 for verbose
--generate_previews, -previews	Do not generate previews for the encoded file
--generate_stats, -stats	Generate stats for the encoded file
--override_bad_wrong_cache_path	Override the check for input file path matching in scene cache loading
--hdr	Encode in HDR, if not specified and input is HDR, it will automatically tonemap
--crop_string CROP_STRING	Crop string to use (e.g., 1920:1080:0:0, 3840:1600:0:280). Obtained using the cropdetect ffmpeg filter
--scale_string SCALE_STRING	Scale string to use (e.g., 1920:1080, 1280:-2, 1920:1080:force_original_aspect_ratio=decrease)
--dry_run	Do not encode, just print what would be done
--title TITLE	Title of the video
--encoder_flag_override ENCODER_FLAG_OVERRIDE	Override the encoder flags with this string, write all params except paths
--encoder_speed_override ENCODER_SPEED_OVERRIDE	Override the encoder speed parameter (must be in range 0..=10)
--crf_map STRING	Map of CRF <-> chunk index, for debugging purposes only
--color-primaries COLOR_PRIMARIES	Color primaries
--transfer-characteristics TRANSFER_CHARACTERISTICS	Transfer characteristics
--matrix-coefficients MATRIX_COEFFICIENTS	Matrix coefficients
--maximum_content_light_level MAXIMUM_CONTENT_LIGHT_LEVEL	Maximum content light level
--frame-average-light FRAME_AVERAGE_LIGHT	Maximum frame average light level
--chroma-sample-position CHROMA_SAMPLE_POSITION	Chroma sample position
--sub_file SUB_FILE	Subtitles file (e.g., .srt or .vvt)
--vmaf_phone_model	Use VMAF phone model for auto CRF tuning
--vmaf_4k_model	Use VMAF 4K model for auto CRF tuning
--vmaf_no_motion	Use VMAF no motion model for auto CRF tuning
--auto_accept_autocrop	Automatically accept autocrop
--resolution_preset RESOLUTION_PRESET	Preset for the scale filter, possible choices are 4K, 1440p, 1080p, 768p, 720p, 540p, 480p, 360p
--probe_count PROBE_COUNT	Max number of probes for metric targeting, higher is more accurate but slower (must be in range 1..=10)
--vmaf_probe_speed VMAF_PROBE_SPEED	Override the speed for target VMAF probes (must be in range 0..=10)
--vmaf_reference_display {HD,FHD,UHD}	HD, FHD, or UHD
--vmaf_ai_assisted_targeting	Use VMAF AI-assisted targeting
--vmaf_target_repesentation {mean,min,max,harmonic_mean,percentile_1,percentile_5,percentile_10,percentile_25,percentile_50}	VMAF target representation, default is mean
--simple_denoise	Use atadenoise on input, useful for x26 encoding with very noisy inputs and target VMAF, to be automated in the future
--dont_pin_to_cores	Do not pin each chunk to a core
--niceness NICENESS	Nice the encoder process
--print_analysis_logs	Print content analysis logs into the console, like what CRF did VMAF target pick, etc.
--poster_url POSTER_URL	URL of poster for website updates
--offload_server OFFLOAD_SERVER	If filled with a server address, will try to send a serialized job to that server
--metric_to_target {vmaf,ssimu2}	Uses all the VMAF target logic but a different metric
--dynamic_vmaf_target	Target VMAF and weight it against the bitrate, useful for lossy sources that trick VMAF into low scores
--dynamic_vmaf_target_vbr	VMAF targeting but instead of tuning CRF, it tunes the bitrate and uses variable bitrate encoding
--tune {fidelity,appeal,balanced}	Tune the encoder setting for a specific use case
--denoise_vmaf_ref	Denoise the VMAF reference
--calc_final_vmaf	Calculate final VMAF of a chunk and save it in chunks.log
--multi_res_pipeline	Create an optimized multi-bitrate tier stream
--throughput_scaling	Scale the multi-process workers based on throughput

This list will be updated as new features & flags are added.