pat-e/auto-boost-algorithm
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Auto-Boost-Next placeholder

Browse Source
This commit is contained in:
Trix
2025-07-24 12:36:43 +02:00
committed by GitHub
parent f8cb4de139
commit 196483eee0
2 changed files with 35 additions and 36 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
Auto-Boost-Next/README.md Normal file
View File

@@ -0,0 +1 @@
# Auto-Boost-Next

70
README.md
View File

@@ -1,31 +1,24 @@
# Auto-boost Algorithm # Auto-boost Algorithm
## Version 1.0: brightness-based ## Auto-Boost-Next:
> Gets the average brightness of a scene and lowers CQ/CRF/Q the darker the scene is, in a _zones.txt_ file to feed Av1an. _Soon:tm:_
The requirements are Vapoursynth, vstools and LSMASHSource.
__Usage:__ ## Version 2.5: SSIMULACRA2&XPSNR-based
```
python auto-boost_1.0.py "{animu.mkv}" "{scenes.json}" {base CQ/CRF/Q} "{encoder: aom/svt-av1/rav1e (optional)}"
```
__Example:__ The requirements are Vapoursynth, LSMASHSource, Av1an, vszip and a few python library libraries. Optionally: ffmpeg built with XPSNR support, turbo-metrics.
```
python auto-boost_1.0.py "path/to/nice_boat.mkv" "path/to/scenes.json" 30
```
__Advantages:__ Refined 2.0 with the following additions & changes:
- Fast - Proper argument parsing, more control over the script
- No bs - Separated the fast encode, metric calculation and zone creation into three independant, callable stages
- Solves one long-lasting issue of AV1 encoders: low bitrate allocation in dark scenes - Replaced the deprecated `vapoursynth-ssimulacra2` by `vszip`
- Added `turbo-metrics` for GPU-accelerated metrics measurement (Nvidia only)
- Added XPSNR metric and a few zones calculation methods
- Taking advantage of SVT-AV1-PSY quarter-step CRF feature for more granular control
- Possibility to use a more aggressive boosting curve
- And a few other smaller changes...
__Known limitations:__ _Many thanks to R1chterScale, Yiss and Kosaka for iterating on auto-boost and making these amazing contributions!_
- Not every dark scene is made equal, brightness is not a great enough metric to determine whether CRF should be decreased or not
- CRF is boosted to the max during credits
- Script now entirely irrelevant with SVT-AV1-PSY's new frame-luma-bias feature
_Inspiration was drawn from the original Av1an (python) boosting code_
## Version 2.0: SSIMULACRA2-based ## Version 2.0: SSIMULACRA2-based
@@ -53,24 +46,29 @@ __Known limitations:__
_Borrowed some code from Sav1or's SSIMULACRA2 script_ _Borrowed some code from Sav1or's SSIMULACRA2 script_
## Version 2.5: SSIMULACRA2&XPSNR-based ## Version 1.0: brightness-based
The requirements are Vapoursynth, LSMASHSource, Av1an, vszip and a few python library libraries. Optionally: ffmpeg built with XPSNR support, turbo-metrics. > Gets the average brightness of a scene and lowers CQ/CRF/Q the darker the scene is, in a _zones.txt_ file to feed Av1an.
The requirements are Vapoursynth, vstools and LSMASHSource.
Refined 2.0 with the following additions & changes: __Usage:__
- Proper argument parsing, more control over the script ```
- Separated the fast encode, metric calculation and zone creation into three independant, callable stages python auto-boost_1.0.py "{animu.mkv}" "{scenes.json}" {base CQ/CRF/Q} "{encoder: aom/svt-av1/rav1e (optional)}"
- Replaced the deprecated `vapoursynth-ssimulacra2` by `vszip` ```
- Added `turbo-metrics` for GPU-accelerated metrics measurement (Nvidia only)
- Added XPSNR metric and a few zones calculation methods
- Taking advantage of SVT-AV1-PSY quarter-step CRF feature for more granular control
- Possibility to use a more aggressive boosting curve
- And a few other smaller changes...
_Many thanks to R1chterScale, Yiss and Kosaka for iterating on auto-boost and making these amazing contributions!_ __Example:__
```
python auto-boost_1.0.py "path/to/nice_boat.mkv" "path/to/scenes.json" 30
```
## Version 3.0: SSIMULACRA2-based + per-scene grain synthesis strength determination __Advantages:__
- Fast
- No bs
- Solves one long-lasting issue of AV1 encoders: low bitrate allocation in dark scenes
...and a few other improvements. __Known limitations:__
- Not every dark scene is made equal, brightness is not a great enough metric to determine whether CRF should be decreased or not
- CRF is boosted to the max during credits
- Script now entirely irrelevant with SVT-AV1-PSY's new frame-luma-bias feature
_Soon:tm:_ _Inspiration was drawn from the original Av1an (python) boosting code_