]> Git — Sourcephile - julm/camera.git/blob - svt-av1.txt
index: use 1200x and JPG for downloads
[julm/camera.git] / svt-av1.txt
1 DOC: https://gitlab.com/AOMediaCodec/SVT-AV1/-/blob/master/Docs/CommonQuestions.md
2 DOC: https://gitlab.com/AOMediaCodec/SVT-AV1/-/blob/master/Docs/Parameters.md
3 DOC: https://gist.github.com/BlueSwordM/86dfcb6ab38a93a524472a0cbe4c4100
4 DOC: https://www.streamingmediaglobal.com/Articles/ReadArticle.aspx?ArticleID=154488
5 DOC: https://www.reddit.com/r/AV1/comments/s7yyf9/help_me_understand_svtav1_parameters/
6
7 -crf
8 Range is 0-63, with the default being 50.
9 Lower values correspond to higher quality and greater file size
10 Equivalent to: --rc 0 --aq-mode 2 --qp <range>
11 For the same quality, you should decrease the CRF as you lower the resolution
12 (and increase it for higher resolutions).
13
14 enable-overlays=1
15 Higher quality keyframes and better references overall.
16
17 enable-tf=0
18 Disable ALT-REF (temporally filtered) frames (slightly different B-frames)
19 and increases their compressibilities somewhat,
20 at the cost of slight detail loss at higher qualities.
21
22 fast-decode=3
23 Trade compression efficiency for less CPU time when decoding
24 for preset 5-7: levels 1-3 (3 being faster) are supported.
25 FIXME: set fast-decode to 3 available on newer svt-av1
26
27 film-grain=10
28 {0..50} Enables film grain synthesis.
29 The encoder denoises the source (higher value means stronger denoising)
30 and saves noise parameters in a look-up table.
31 The decoder can later use that table to recreate the grain during playback.
32 This reduces the necessary bitrate drastically for grainy footage.
33 As you increase the strength, you increase the strength of the denoiser
34 and how much SVT-AV1 decides to replace the original noise with synthesized noise
35 (ie. if you increase the number, you get more grain and a bit more denoising).
36 It is recommended to not use Film Grain for presets greater than 6
37 as it produces a significant compute overhead.
38 This combination should only be used for debug purposes.
39
40 -preset
41 Range from 0 to 13, with higher numbers providing a higher encoding speed;
42 Reasonable time/quality ratios are in the range 4-8,
43 with 6 being a good starting point.
44 >=8 is good for real time encoding (livestreaming),
45 <4 is rarely worth it.
46
47 scd=1
48 Biases the encoders' scene detection to insert more intra refreshes if needed,
49 and reduce temporal dependencies around scene-changes.
50
51 scm=0
52 Screen content mode decision making. 0 is best for live-action.
53
54 tune=0
55 Subjective mode often results in an image with greater
56 sharpness and is intended to produce a result that appears to humans
57 to be of high quality (as opposed to doing well on basic objective measures,
58 such as PSNR)
59
60 yuv420p
61 yuv420p10le
62 Can represent more shades of grey and colors
63 and is less prone to certain artifacts, such as color
64 banding and loss of detail in low luma areas,
65 cost in terms of resulting file size (~5%),
66 can also be more compute-intensive than 8-bit in some decoders
67 10 bit playback performance is not reliable enough on average consumer hardware.