Hilbert: Hey Rosy, I noticed your release includes subtitles. What’s the deal with them?
Rosy: Oh, good question! For this release, I’ve included two types of subtitles.
Hilbert: Two types? What’s the difference?
Rosy: Some of the subtitles are human-translated, mostly sourced from OpenSubtitles, so they’re usually very accurate.
Hilbert: And the others?
Rosy: The others are machine-translated using SubtitleCat. They’re great for less common languages, but they might not be perfect.
Hilbert: So, you combine them to cover more audiences?
Rosy: Exactly! It’s about giving everyone options, even if the machine-translated ones aren’t as polished as the human ones.
Hilbert: That’s smart. Thanks for explaining!
____________________________________________________________
Encoding
Ab-av1 20 samples
prompt → sample-encode -i "C:\file.mkv" --pix-format yuv420p10le --svt fast-decode=1 --svt tune=0 --preset 5 --min-samples 20 --crf 25 --vfilter "scale=1920x960"
crf 25 VMAF 95.13 predicted video stream size 739.91 MiB (12%) taking 46 minutes
FFmetrics full stream
VMAF → 92.62
Handbrake 1.8.2 (2024081000)
Video.svt-av1 10 bit fast decode Preset 5 2-pass vbr 1818 kbps targeting 800mb
Audio.2.0 opus 128 kbps
____________________________________________________________
File
Video: AV01 1920x960 23.976fps [V: av1 main, yuv420p10le, 1920x960 [default]]
Audio: Opus 48000Hz stereo 3072kbps [A: Stereo [eng] (opus, 48000 Hz, stereo) [default]]
Subtitle: Advanced SubStation Alpha [S: English [SDH] [eng] (ass)]
Subtitle: UTF-8 [S: opensubtitles [por] (subrip) [default]]
Subtitle: UTF-8 [S: opensubtitles [srp] (subrip) [default]]
Subtitle: UTF-8 [S: opensubtitles [bul] (subrip) [default]]
Subtitle: UTF-8 [S: opensubtitles [spa] (subrip) [default]]
Subtitle: UTF-8 [S: opensubtitles [tur] (subrip) [default]]
Subtitle: UTF-8 [S: opensubtitles [fin] (subrip) [default]]
Subtitle: UTF-8 [S: opensubtitles [ara] (subrip) [default]]
Subtitle: UTF-8 [S: opensubtitles [eng] (subrip) [default]]
Subtitle: UTF-8 [S: opensubtitles [ger] (subrip) [default]]
Subtitle: UTF-8 [S: opensubtitles [dut] (subrip) [default]]
Subtitle: UTF-8 [S: opensubtitles [rum] (subrip) [default]]
Subtitle: UTF-8 [S: opensubtitles [rum] (subrip) [default]]
Subtitle: UTF-8 [S: opensubtitles [pol] (subrip) [default]]
Subtitle: UTF-8 [S: opensubtitles [rus] (subrip) [default]]
Subtitle: UTF-8 [S: opensubtitles [dan] (subrip) [default]]
Subtitle: UTF-8 [S: opensubtitles [swe] (subrip) [default]]
Subtitle: UTF-8 [S: opensubtitles [eng] (subrip) [default]]
Subtitle: UTF-8 [S: subtitlecat [fre] (subrip) [default]]
Subtitle: UTF-8 [S: subtitlecat [hin] (subrip) [default]]
Subtitle: UTF-8 [S: subtitlecat [ita] (subrip) [default]]
Subtitle: UTF-8 [S: subtitlecat [jpn] (subrip) [default]]
Subtitle: UTF-8 [S: subtitlecat [kor] (subrip) [default]]
Subtitle: UTF-8 [S: No subtitles]
General
Unique ID : 243398441555870647955881714261125217173 (0xB71CD5AEF65D639388E00B9A662FBB95)
Complete name : Landman.S01E01.web.sdr.2160p.av1.ds4k.subs.vmaf95-Rosy.mkv
Format : Matroska
Format version : Version 4
File size : 774 MiB
Duration : 56 min 3 s
Overall bit rate : 1 930 kb/s
Frame rate : 23.976 FPS
Movie name : Landman.S01E01.web.sdr.2160p.av1.ds4k.subs.vmaf95-Rosy
Encoded date : 2024-11-25 22:10:33 UTC
Writing application : mkvmerge v88.0 ('All I Know') 64-bit
Writing library : libebml v1.4.5 + libmatroska v1.7.1
Cover : Yes
Attachments : cover.jpg
Video
ID : 1
Format : AV1
Format/Info : AOMedia Video 1
Format profile : [email protected]
Codec ID : V_AV1
Duration : 56 min 3 s
Bit rate : 1 814 kb/s
Width : 1 920 pixels
Height : 960 pixels
Display aspect ratio : 2.000
Frame rate mode : Constant
Frame rate : 23.976 (24000/1001) FPS
Color space : YUV
Chroma subsampling : 4:2:0 (Type 0)
Bit depth : 10 bits
Bits/(Pixel*Frame) : 0.041
Stream size : 728 MiB (94%)
Default : Yes
Forced : No
Color range : Limited
Color primaries : BT.709
Transfer characteristics : BT.709
Matrix coefficients : BT.709
Audio
ID : 2
Format : Opus
Codec ID : A_OPUS
Duration : 56 min 3 s
Bit rate : 109 kb/s
Channel(s) : 2 channels
Channel layout : L R
Sampling rate : 48.0 kHz
Frame rate : 50.000 FPS (960 SPF)
Compression mode : Lossy
Stream size : 43.7 MiB (6%)
Title : Stereo
Language : English
Default : Yes
Forced : No
____________________________________________________________
Source
landman.s01e01.2160p.web.h265-successfulcrab
Video: HVC1 3840x1920 23.976fps [V: hevc main L5.0, yuv420p, 3840x1920 [default]]
Audio: DD+ 48000Hz 6ch 640kbps [A: English [eng] (eac3, 48000 Hz, 5.1(side), 640 kb/s) [default]]
Subtitle: UTF-8 [S: English [SDH] [eng] (subrip)]
Subtitle: UTF-8 [S: No subtitles]
General
Unique ID : 304432051444733883179062824855482538608 (0xE5077AFAE54DCEA6C1C26B8F93E61A70)
Complete name : landman.s01e01.2160p.web.h265-successfulcrab.mkv
Format : Matroska
Format version : Version 4
File size : 6.02 GiB
Duration : 56 min 3 s
Overall bit rate : 15.4 Mb/s
Frame rate : 23.976 FPS
Writing application : mkvmerge v84.0 ('Sleeper') 64-bit
Writing library : libebml v1.4.5 + libmatroska v1.7.1
Video
ID : 1
Format : HEVC
Format/Info : High Efficiency Video Coding
Format profile : Main@L5@Main
Codec ID : V_MPEGH/ISO/HEVC
Duration : 56 min 3 s
Bit rate : 14.7 Mb/s
Width : 3 840 pixels
Height : 1 920 pixels
Display aspect ratio : 2.000
Frame rate mode : Constant
Frame rate : 23.976 (24000/1001) FPS
Color space : YUV
Chroma subsampling : 4:2:0
Bit depth : 8 bits
Bits/(Pixel*Frame) : 0.083
Stream size : 5.77 GiB (96%)
Default : Yes
Forced : No
Color range : Limited
Color primaries : BT.709
Transfer characteristics : BT.709
Matrix coefficients : BT.709
Audio
ID : 2
Format : E-AC-3
Format/Info : Enhanced AC-3
Commercial name : Dolby Digital Plus
Codec ID : A_EAC3
Duration : 56 min 3 s
Bit rate mode : Constant
Bit rate : 640 kb/s
Channel(s) : 6 channels
Channel layout : L R C LFE Ls Rs
Sampling rate : 48.0 kHz
Frame rate : 31.250 FPS (1536 SPF)
Compression mode : Lossy
Stream size : 257 MiB (4%)
Language : English
Service kind : Complete Main
Default : Yes
Forced : No
Dialog Normalization : -31 dB
compr : -0.28 dB
dialnorm_Average : -31 dB
dialnorm_Minimum : -31 dB
dialnorm_Maximum : -31 dB
____________________________________________________________
Tools
Ab-av1
FFBitrateViewer
FFmpeg
FFmetrics
Freefilesync
Handbrake
Jackett
Mediainfo
Mkvtoolknix
Qbittorent
Scenerulez
Shutterencoder
VVV-virtualvolumeview
Xnconvert
Xnview mp
===================== AV1 guide v14 ========================
Step 1 - Check VMAF with Ab-av1
1. AV1
Source resolution → crf-search -i "c:\file.mkv" --pix-format yuv420p10le --svt fast-decode=1 --svt tune=0 --preset 5 --min-samples 20 --min-vmaf 98
1080p Downscale → crf-search -i "c:\file.mkv" --pix-format yuv420p10le --svt fast-decode=1 --svt tune=0 --preset 5 --min-samples 20 --min-vmaf 96 --vfilter "scale=-1:1080"
720p Downscale → crf-search -i "c:\file.mkv" --pix-format yuv420p10le --svt fast-decode=1 --svt tune=0 --preset 5 --min-samples 20 --min-vmaf 96 --vfilter "scale=-1:720"
Sample encode → sample-encode -i "c:\file.mkv" --pix-format yuv420p10le --svt fast-decode=1 --svt tune=0 --preset 5 --min-samples 20 --crf 20
2. x265
Source resolution → crf-search -i "c:\file.mkv" --encoder libx265 --pix-format yuv420p10le --preset medium --min-samples 20 --enc x265-params=fastdecode=1 --min-vmaf 98
1080p Downscale → crf-search -i "c:\file.mkv" --encoder libx265 --pix-format yuv420p10le --preset medium --min-samples 20 --enc x265-params=fastdecode=1 --min-vmaf 96 --vfilter "scale=-1:1080"
720p Downscale → crf-search -i "c:\file.mkv" --encoder libx265 --pix-format yuv420p10le --preset medium --min-samples 20 --enc x265-params=fastdecode=1 --min-vmaf 96 --vfilter "scale=-1:720"
____________________________________________________________
Step 2 - HandBrake
Summary:
- Container: MKV
Dimensions:
- Cropping: None
- Anamorphic: None
- Borders: None
Video:
- Encoder: AV1 (10-bit SVT)
- Framerate: Same as source
- Preset: 5
- Fast Decode: ✓
- Encoder Profile: Main
- Encoder Level: Auto
- Quality: CRF determined from "ab-av1"(1), or crf 20-25
Audio:
- DTS to Opus Conversion:
- DTS 7.1 → Opus @ 512 kbps
- DTS 5.1 → Opus @ 384 kbps
- DTS 2.0 → Opus @ 128 kbps
*When passtrough? lossy codec(opus,aac, ac3, eac3,wma) 768-128 kbps
____________________________________________________________
Step 3 - Subtitles
- Original → passtrough
- Subtitle Translator
- Whisper OpenAI
- Opensubtitles
- Subtitlecat
*If size is an issue don`t use pgs, prefer ass/srt.
Whisperai large-v3 2024 (prompts)
!pip install git+https://github.com/openai/whisper.git
!sudo apt update && sudo apt install ffmpeg
!whisper "audio.aac" --model large-v3 --language Japanese --task translate
____________________________________________________________
References
1. Dolby, 2023. What are Dolby Vision profiles and levels? [Online]
2. Ozer, J., 2017. Video Encoding by the Numbers: Eliminate the Guesswork from your Streaming Video. Doceo Publishing.
3. Sayood, K., 2017. Introduction to Data Compression. 5th Edition.
4. Ozer, J., 2021. What is CBR, VBR, CRF, Capped-CRF? Rate Control Modes Explained.
5. Rassool, Reza. VMAF Reproducibility: Validating a Perceptual Practical Video Quality Metric.
6. Kah, A. e al. Fundamental relationships between subjective quality, user acceptance, and the VMAF metric for a quality-based bit rate ladder design for over-the-top video streaming services.
7. Boldrin, M. 2008. Against Intellectual Monopoly.
