We're here again with a new SVT-AV1-PSY release.
This time, we've decided to change the way we name these updates as they've been difficult to keep track of versioning against mainline SVT-AV1 and our own. We'll still follow the cadence of mainline SVT-AV1, but with letter releases added in between for additional clarity.
I plan on releasing Linux optimized binaries soon and someone else will step up to provide the all important Windows builds and ideally, an ffmpeg build.
The Linux optimized binaries have been added. They are packaged under a single archive with a generic build, an x86-64-v3 build, an x86-64-v4 build (AVX512 capable CPU required) and an x86-64-v4-AVX512 build that utilizes the native SVT-AV1 AVX512 SIMD assembly code for maximum speed.
Let's get on to the main course, which is discussing the new options.
-> MAIN CHANGES:
-
CRF range: previously capped at 63, it has been extended to a maximum value of 70. It can also be incremented in quarter increments of 0.25. This change allows for greater control of the base quantizer through CRF and as you increase CRF beyond 63, gradually reducing the effect of quantizer scaling.
-
New option: --enable-dlf 2 for a slower, more accurate deblocking loop filter based on scaling deblocker preset pruning by 3 levels (P7 with --enable-dlf 2 = P4 deblocking)
-
New option: --qp-scale-compress-strength (0 to 3) which sets a strength for the QP scale algorithm to compress values across all temporal layers. Higher values result in more consistent video quality, especially with lots of motion or moving features (grain...). Default is 1, SVT-AV1 default is 0.
-
New option: --frame-luma-bias (0 to 100) enables experimental frame-level luma bias to improve quality in dark scenes by adjusting frame-level QP based on average luminance across each frame. Future improvements will revolve around making it more accurate and based on proper lightness to make it work across color spaces. Other general improvements to Tune 3
-> INNOVATIONS:
- Dynamic delta_q_res switching implemented to help reduce signaling overhead, which should improve quality especially at CRF ≥40! This is a big move considering it's a free efficiency increase
-> DOCUMENTATION:
-
PNG images have been replaced with smaller lossless WebP images, resulting in faster loading & repository cloning times.
-
More consistent & thorough PSY Development page, including build instructions