FFmpeg Bash Scripts

01 Download

• scripts/01_download/download_hls_stream.sh Purpose: Download an HLS URL (.m3u8) to a local file using stream copy. Usage: ./scripts/01_download/download_hls_stream.sh <hls_url> <output_file>

02 Tagging

• scripts/02_tagging/tag_mov_prores_apcn.sh Purpose: Tag MOV video stream as ProRes apcn while copying all tracks. Usage: ./scripts/02_tagging/tag_mov_prores_apcn.sh <input.mov> [output.mov]

• scripts/02_tagging/tag_mov_avc1_bt709.sh Purpose: Tag MOV video as avc1 and attach BT.709 color metadata. Usage: ./scripts/02_tagging/tag_mov_avc1_bt709.sh <input.mov> [output.mov]

• scripts/02_tagging/tag_mp4_hvc1_bt709.sh Purpose: Tag MP4 video as hvc1 and attach BT.709 color metadata. Usage: ./scripts/02_tagging/tag_mp4_hvc1_bt709.sh <input.mp4> [output.mp4]

03 Metrics

• scripts/03_metrics/metrics_compare_ssim_psnr.sh Purpose: Compute SSIM and PSNR between reference and test videos. Usage: ./scripts/03_metrics/metrics_compare_ssim_psnr.sh <reference> <test> [width] [height]

• scripts/03_metrics/metrics_compare_ssim_psnr_log.sh Purpose: Compute SSIM/PSNR and write SSIM stats to a log file. Usage: ./scripts/03_metrics/metrics_compare_ssim_psnr_log.sh <reference> <test> [stats_file] [width] [height]

• scripts/03_metrics/metrics_compare_vmaf.sh Purpose: Compute VMAF between reference and test videos. Usage: ./scripts/03_metrics/metrics_compare_vmaf.sh <reference> <test> [model] [width] [height]

04 Encode

• scripts/04_encode/hevc/encode_hevc_single_x265.sh Purpose: Re-encode one input to HEVC (libx265) and copy audio/subtitles. Usage: ./scripts/04_encode/hevc/encode_hevc_single_x265.sh <input> [output] [crf] [preset]

• scripts/04_encode/hevc/encode_hevc_batch_x265.sh Purpose: Batch re-encode files matching a glob to HEVC (libx265). Usage: ./scripts/04_encode/hevc/encode_hevc_batch_x265.sh "<glob>" [crf] [preset] [suffix]

• scripts/04_encode/h264/encode_h264_single_x264.sh Purpose: Re-encode one input to H.264 (libx264) and copy audio/subtitles. Usage: ./scripts/04_encode/h264/encode_h264_single_x264.sh <input> [output] [crf] [preset]

• scripts/04_encode/h264/encode_h264_batch_x264.sh Purpose: Batch re-encode files matching a glob to H.264 (libx264). Usage: ./scripts/04_encode/h264/encode_h264_batch_x264.sh "<glob>" [crf] [preset] [suffix]

• scripts/04_encode/av1/encode_av1_single_svt.sh Purpose: Re-encode one input to AV1 with libsvtav1 (popular practical AV1 workflow). Usage: ./scripts/04_encode/av1/encode_av1_single_svt.sh <input> [output] [crf] [preset]

• scripts/04_encode/av1/encode_av1_batch_svt.sh Purpose: Batch re-encode files matching a glob to AV1 with libsvtav1. Usage: ./scripts/04_encode/av1/encode_av1_batch_svt.sh "<glob>" [crf] [preset] [suffix] [container]

• scripts/04_encode/av1/encode_av1_single_aom.sh Purpose: Re-encode one input to AV1 with libaom-av1 (quality-focused, slower). Usage: ./scripts/04_encode/av1/encode_av1_single_aom.sh <input> [output] [crf] [cpu_used]

• scripts/04_encode/hardware/encode_hevc_single_videotoolbox.sh Purpose: Hardware HEVC encode using hevc_videotoolbox (macOS). Usage: ./scripts/04_encode/hardware/encode_hevc_single_videotoolbox.sh <input> [output] [bitrate] [realtime]

• scripts/04_encode/hardware/encode_hevc_batch_videotoolbox.sh Purpose: Batch hardware HEVC encode using hevc_videotoolbox. Usage: ./scripts/04_encode/hardware/encode_hevc_batch_videotoolbox.sh "<glob>" [bitrate] [suffix] [container] [realtime]

• scripts/04_encode/hardware/encode_av1_single_videotoolbox.sh Purpose: Hardware AV1 encode using av1_videotoolbox when available. Usage: ./scripts/04_encode/hardware/encode_av1_single_videotoolbox.sh <input> [output] [bitrate] [realtime]

• scripts/04_encode/hardware/encode_av1_batch_videotoolbox.sh Purpose: Batch hardware AV1 encode using av1_videotoolbox when available. Usage: ./scripts/04_encode/hardware/encode_av1_batch_videotoolbox.sh "<glob>" [bitrate] [suffix] [container] [realtime]

• scripts/04_encode/hdr/convert_sdr_to_hdr10_style.sh Purpose: Convert SDR video to an HDR10-style deliverable using a filter pipeline (gamut remap + tone shaping + detail polish). Usage: ./scripts/04_encode/hdr/convert_sdr_to_hdr10_style.sh <input> [output] [profile] [codec] [quality] [preset] Notes: Profiles are natural, gentle, balanced, and punchy (natural is default). Use HDR_TRANSFER=bt2020-10 for a safer non-dark result, or HDR_TRANSFER=smpte2084 for stronger HDR signaling.

• scripts/04_encode/prores/encode_prores4444_minterpolate.sh Purpose: Normalize one source to target FPS/size and encode ProRes 4444. Usage: ./scripts/04_encode/prores/encode_prores4444_minterpolate.sh <input> <output> [fps] [width] [height]

• scripts/04_encode/prores/encode_prores4444xq_alpha_denoise_minterpolate.sh Purpose: Denoise and motion-interpolate to ProRes 4444 XQ with alpha. Usage: ./scripts/04_encode/prores/encode_prores4444xq_alpha_denoise_minterpolate.sh <input> <output> [fps]

• scripts/04_encode/prores/encode_prores4444xq_no_alpha_denoise_minterpolate.sh Purpose: Denoise and motion-interpolate to ProRes 4444 XQ without alpha. Usage: ./scripts/04_encode/prores/encode_prores4444xq_no_alpha_denoise_minterpolate.sh <input> <output> [fps]

• scripts/04_encode/upscale/upscale_2x_60fps_prores422.sh Purpose: Upscale 2x, interpolate to 60 fps, and output ProRes 422. Usage: ./scripts/04_encode/upscale/upscale_2x_60fps_prores422.sh <input_video> [output_dir]

• scripts/04_encode/upscale/upscale_video_lanczos_plain.sh Purpose: Plain non-AI, non-tiling Lanczos upscale only (no interpolation, no ProRes-specific output path). Usage: ./scripts/04_encode/upscale/upscale_video_lanczos_plain.sh <input> [output] [scale_factor] [codec] [quality] [preset] Notes: Default output uses libx264. Set codec to libx265 or *_videotoolbox if desired.

• scripts/04_encode/upscale/upscale_video_realesrgan_coreml_x2.sh Purpose: CoreML upscale in non-tiled mode for supported models (RealESRGAN x2/x4, SwinIR x2), model-native path. Usage: ./scripts/04_encode/upscale/upscale_video_realesrgan_coreml_x2.sh <input> [output] [model_path] [work_dir] [crf] [preset] Notes: Set MODEL_COLOR_ORDER=bgr if a model expects BGR channel order. Set TARGET_UPSCALE=2 to force 2x output from an x4 model.

• scripts/04_encode/upscale/upscale_video_realesrgan_coreml_x2_tiled.sh Purpose: Split source into .mkv tile videos, upscale each tile via supported CoreML model (RealESRGAN x2/x4, SwinIR x2), then reassemble seam-free. Usage: ./scripts/04_encode/upscale/upscale_video_realesrgan_coreml_x2_tiled.sh <input> [output] [model_path] [work_dir] [overlap] [crf] [preset] [rows] [cols] Notes: Default intermediates are lossless ffv1 with gbrp to reduce color shifts. Set MODEL_COLOR_ORDER=bgr for BGR models. Set TARGET_UPSCALE=2 to force 2x output from an x4 model. Set TILE_JOBS to run multiple tile videos in parallel. If grid resolves to 1x1, the script now skips split/reassemble and runs direct tiled inference for speed.

• scripts/04_encode/upscale/upscale_video_realesrgan_coreml_vram_turbo.sh Purpose: macOS turbo wrapper with auto-tuned parallelism, GPU-focused CoreML compute, and fast hardware final encode. Usage: ./scripts/04_encode/upscale/upscale_video_realesrgan_coreml_vram_turbo.sh <input> [output] [model_path] [work_dir] [rows] [cols] Notes: Defaults now target quality-safe speed (ffv1 lossless intermediate, large auto tile size for direct path on smaller sources). Set ALLOW_LOSSY_INTERMEDIATE=1 to use hardware intermediate encode for extra speed. Set V_ALLOW_SW=1 (default) to allow VideoToolbox software fallback when hardware sessions are busy.

• scripts/04_encode/upscale/enhance_video_look_secondary_flow.sh Purpose: Secondary post-upscale visual polish flow using FFmpeg enhancement filters (cas, denoise, deband, detail shaping) with profile presets. Usage: ./scripts/04_encode/upscale/enhance_video_look_secondary_flow.sh <input> [output] [profile] [codec] [quality] [preset] Notes: Profile presets are subtle, balanced, crisp, and smooth. This script is standalone and does not modify the core upscale pipeline.

• scripts/04_encode/upscale/upscale_video_realesrgan_coreml_vram_turbo_with_enhance.sh Purpose: One-command chain that runs turbo CoreML upscale first, then runs the secondary FFmpeg enhancement flow. Usage: ./scripts/04_encode/upscale/upscale_video_realesrgan_coreml_vram_turbo_with_enhance.sh <input> [output] [model_path] [enhance_profile] [work_dir] [rows] [cols] Notes: Enhancement profile presets are subtle, balanced, crisp, and smooth. Uses env ENHANCE_CODEC, ENHANCE_QUALITY, ENHANCE_PRESET for stage-2 encode tuning.

05 Blend and Stack

• scripts/05_blend_stack/normalize/normalize_two_to_prores422.sh Purpose: Normalize two inputs to matching fps/size and output ProRes 422 files. Usage: ./scripts/05_blend_stack/normalize/normalize_two_to_prores422.sh <input_a> <input_b> [out_dir] [fps] [width] [height]

• scripts/05_blend_stack/normalize/normalize_three_to_prores4444.sh Purpose: Normalize three inputs to matching fps/size and output ProRes 4444 files. Usage: ./scripts/05_blend_stack/normalize/normalize_three_to_prores4444.sh <input_a> <input_b> <input_c> [out_dir] [fps] [width] [height]

• scripts/05_blend_stack/blend/blend_average_two_hevc.sh Purpose: Average-blend two videos and encode result to HEVC. Usage: ./scripts/05_blend_stack/blend/blend_average_two_hevc.sh <input_a> <input_b> [output] [crf] [preset]

• scripts/05_blend_stack/blend/blend_difference_two.sh Purpose: Difference-blend two videos for alignment/debug comparisons. Usage: ./scripts/05_blend_stack/blend/blend_difference_two.sh <input_a> <input_b> [output]

• scripts/05_blend_stack/blend/blend_weighted_three_prores4444.sh Purpose: Weighted 3-way blend to ProRes 4444. Usage: ./scripts/05_blend_stack/blend/blend_weighted_three_prores4444.sh <input_a> <input_b> <input_c> [output]

• scripts/05_blend_stack/blend/finish_weighted_merge_4k_prores4444.sh Purpose: Finalize merged input to 4K ProRes 4444 with finishing filters. Usage: ./scripts/05_blend_stack/blend/finish_weighted_merge_4k_prores4444.sh <input> <output> [width] [height]

• scripts/05_blend_stack/blend/blend_edge_aware_hevc.sh Purpose: Edge-aware 2-source merge and HEVC encode. Usage: ./scripts/05_blend_stack/blend/blend_edge_aware_hevc.sh <input_a> <input_b> [output] [crf] [preset]

• scripts/05_blend_stack/stack/stack_2x2_hstack_vstack.sh Purpose: Stack four inputs into a 2x2 layout using hstack + vstack. Usage: ./scripts/05_blend_stack/stack/stack_2x2_hstack_vstack.sh <input0> <input1> <input2> <input3> <output> [width] [height]

• scripts/05_blend_stack/stack/stack_2x2_xstack.sh Purpose: Stack four inputs into a 2x2 layout using xstack. Usage: ./scripts/05_blend_stack/stack/stack_2x2_xstack.sh <input0> <input1> <input2> <input3> <output> [width] [height]

• scripts/05_blend_stack/pipelines/pipeline_parallel_blend_hevc.sh Purpose: Parallel upscale/blend/denoise pipeline to HEVC. Usage: ./scripts/05_blend_stack/pipelines/pipeline_parallel_blend_hevc.sh <input> <output> [crf] [preset]

• scripts/05_blend_stack/pipelines/pipeline_serial_hybrid_hevc.sh Purpose: Serial interpolation/upscale/denoise pipeline to HEVC. Usage: ./scripts/05_blend_stack/pipelines/pipeline_serial_hybrid_hevc.sh <input> <output> [crf] [preset] [fps]

• scripts/05_blend_stack/metrics/metrics_ssim_psnr.sh Purpose: Workflow-local SSIM/PSNR comparison script. Usage: ./scripts/05_blend_stack/metrics/metrics_ssim_psnr.sh <ref> <test> [width] [height]

• scripts/05_blend_stack/metrics/metrics_ssim_psnr_stats.sh Purpose: Workflow-local SSIM/PSNR with stats file output. Usage: ./scripts/05_blend_stack/metrics/metrics_ssim_psnr_stats.sh <ref> <test> [stats_file] [width] [height]

• scripts/05_blend_stack/metrics/metrics_vmaf.sh Purpose: Workflow-local VMAF comparison script. Usage: ./scripts/05_blend_stack/metrics/metrics_vmaf.sh <ref> <test> [model] [width] [height]

• scripts/05_blend_stack/contact_sheets/contact_sheet_video_8x8.sh Purpose: Build an 8x8 contact sheet image from one video. Usage: ./scripts/05_blend_stack/contact_sheets/contact_sheet_video_8x8.sh <input> <output> [fps] [scale_width] [tile]

• scripts/05_blend_stack/contact_sheets/contact_sheet_combine_three.sh Purpose: Combine three contact-sheet images into one horizontal comparison image. Usage: ./scripts/05_blend_stack/contact_sheets/contact_sheet_combine_three.sh <img1> <img2> <img3> <output> [title]

• scripts/05_blend_stack/tagging/tag_mov_avc1_bt709.sh Purpose: Apply avc1 + BT.709 metadata tags to a MOV. Usage: ./scripts/05_blend_stack/tagging/tag_mov_avc1_bt709.sh <input> [output]

06 Visualization

• scripts/06_visualization/contact_sheets/contact_sheet_video_tile8x8.sh Purpose: Build a labeled or unlabeled 8x8 tiled contact sheet from one video. Usage: ./scripts/06_visualization/contact_sheets/contact_sheet_video_tile8x8.sh <input_video> <output_image> [fps] [scale_width] [tile]

• scripts/06_visualization/contact_sheets/contact_sheet_three_videos_3wide.sh Purpose: Build one contact sheet per input video, then a 3-wide combined image. Usage: ./scripts/06_visualization/contact_sheets/contact_sheet_three_videos_3wide.sh <video1> <video2> <video3> [out_dir]

• scripts/06_visualization/contact_sheets/contact_sheet_three_videos_multi_filters.sh Purpose: Build contact sheets for multiple analysis filters and combined 3-wide outputs. Usage: ./scripts/06_visualization/contact_sheets/contact_sheet_three_videos_multi_filters.sh <video1> <video2> <video3> [out_dir]

• scripts/06_visualization/contact_sheets/contact_sheet_three_videos_multi_filters_compat.sh Purpose: Compatibility wrapper that forwards to the maintained multi-filter script. Usage: ./scripts/06_visualization/contact_sheets/contact_sheet_three_videos_multi_filters_compat.sh <video1> <video2> <video3> [out_dir]

• scripts/06_visualization/gifs/gif_three_video_filter_suite.sh Purpose: Generate single, triptych, and sweep GIF comparisons across common filter looks. Usage: ./scripts/06_visualization/gifs/gif_three_video_filter_suite.sh <video1> <video2> <video3> [out_dir]

07 Muxing

• scripts/07_muxing/remux_container_copy.sh Purpose: Remux media into a new container without re-encoding streams. Usage: ./scripts/07_muxing/remux_container_copy.sh <input> <output>

• scripts/07_muxing/convert_gif_to_video_mp4.sh Purpose: Convert a GIF animation into MP4 video using compatibility-safe H.264 settings. Usage: ./scripts/07_muxing/convert_gif_to_video_mp4.sh <input.gif> [output.mp4] [fps] [crf] [preset]

• scripts/07_muxing/convert_gif_to_video_batch_mp4.sh Purpose: Batch convert multiple GIF files to MP4 using a glob pattern. Usage: ./scripts/07_muxing/convert_gif_to_video_batch_mp4.sh "<glob>" [out_dir] [fps] [crf] [preset]

• scripts/07_muxing/mux_replace_audio_from_second_input.sh Purpose: Keep video from input #1 and replace its audio with input #2 audio. Usage: ./scripts/07_muxing/mux_replace_audio_from_second_input.sh <video_input> <audio_input> [output]

• scripts/07_muxing/mux_add_external_audio_track.sh Purpose: Add an external audio track while keeping existing streams. Usage: ./scripts/07_muxing/mux_add_external_audio_track.sh <video_input> <audio_input> [output]

• scripts/07_muxing/mux_add_srt_subtitle_to_mp4.sh Purpose: Add an SRT subtitle track to MP4 (subtitle codec mov_text). Usage: ./scripts/07_muxing/mux_add_srt_subtitle_to_mp4.sh <video_input> <subtitle.srt> [output.mp4]

• scripts/07_muxing/mux_concat_copy_from_args.sh Purpose: Concatenate multiple matching files using concat demuxer and stream copy. Usage: ./scripts/07_muxing/mux_concat_copy_from_args.sh <output> <input1> <input2> [inputN...]

• scripts/07_muxing/mux_video_audio_subtitle_mp4.sh Purpose: Mux video + external audio + SRT subtitle into one MP4. Usage: ./scripts/07_muxing/mux_video_audio_subtitle_mp4.sh <video_input> <audio_input> <subtitle.srt> [output.mp4]

08 Compression Profiles

• scripts/08_compression_profiles/compress_h264_high_compat.sh Purpose: Standard high-compatibility H.264 compression profile for broad playback. Usage: ./scripts/08_compression_profiles/compress_h264_high_compat.sh <input> [output] [crf] [preset]

• scripts/08_compression_profiles/compress_hevc_high_efficiency.sh Purpose: Standard HEVC high-efficiency profile (10-bit) for smaller files at similar quality. Usage: ./scripts/08_compression_profiles/compress_hevc_high_efficiency.sh <input> [output] [crf] [preset]

• scripts/08_compression_profiles/compress_av1_high_efficiency.sh Purpose: Standard AV1 high-efficiency profile with libsvtav1. Usage: ./scripts/08_compression_profiles/compress_av1_high_efficiency.sh <input> [output] [crf] [preset]

• scripts/08_compression_profiles/compress_h264_two_pass_target_size.sh Purpose: Two-pass H.264 encode targeting a specific output size. Usage: ./scripts/08_compression_profiles/compress_h264_two_pass_target_size.sh <input> <target_mb> [output] [preset] [audio_kbps]

• scripts/08_compression_profiles/compress_h264_ladder_abr.sh Purpose: Build a 3-rung ABR ladder (1080p/720p/480p) for streaming-ready renditions. Usage: ./scripts/08_compression_profiles/compress_h264_ladder_abr.sh <input> [output_dir]

09 Tiling

• scripts/09_tiling/split_video_into_tiles_high_quality.sh Purpose: Split a video frame into an N×M grid of spatial tiles with a manifest for exact reconstruction. Usage: ./scripts/09_tiling/split_video_into_tiles_high_quality.sh <input> <rows> <cols> <out_dir> [crf] [preset]

• scripts/09_tiling/reassemble_video_from_tiles_blackout.sh Purpose: Reassemble tiled pieces seam-free from manifest, with optional blacked-out spatial blocks. Usage: ./scripts/09_tiling/reassemble_video_from_tiles_blackout.sh <manifest> <output> [black_tiles_csv] [audio_source] [tiles_dir] Notes: For tile_count=1 and no blackout list, the script uses a stream-copy fast path.

• scripts/09_tiling/upscale_video_by_tiles_and_reassemble.sh Purpose: Split into spatial tiles, upscale each tile, then reassemble the upscaled tiles seam-free. Usage: ./scripts/09_tiling/upscale_video_by_tiles_and_reassemble.sh <input> <rows> <cols> <scale_int> <output> [work_dir] [crf] [preset]

10 Audio

• scripts/10_audio/enhance_audio_track_ffmpeg.sh Purpose: Enhance audio in audio/video files with denoise, tone shaping, dynamics control, and optional loudness normalization. Usage: ./scripts/10_audio/enhance_audio_track_ffmpeg.sh <input> [output] [profile] [audio_codec] [audio_quality] Notes: Profiles are subtle, balanced, voice, and strong. Video streams are copied while audio is replaced with the enhanced track.

Attribution

Thanks to FFmpeg https://github.com/FFmpeg/FFmpeg