Weekly pre-release

Improvements

• Faster (21-26% on 1080p content with 8 tiles).
• Tracing support via hawktracer
• Basic in-bitstream timing info support
• ASM building is parallel now.

Changes

• CDEF damping is adjusted based on quantizer

Fixes

• Builds cleanly on rust 1.40
• Docs.rs generation is working

Known issues

• #1858 `Tune PSNR may cause desyncs on 4k content"

v0.1.0: Made in Tokyo

First official release, published during the Video Dev Days 2019 in Tokyo.

Features

• Intra and inter frames
• 64x64 superblocks
• 4x4 to 64x64 RDO-selected square and 2:1/1:2 rectangular blocks
• DC, H, V, Paeth, smooth, and a subset of directional prediction modes
• DCT, (FLIP-)ADST and identity transforms (up to 64x64, 16x16 and 32x32 respectively)
• 8-, 10- and 12-bit depth color
• 4:2:0 (full support), 4:2:2 and 4:4:4 (limited) chroma sampling
• 11 speed settings (0-10)
• Near real-time encoding at high speed levels
• Rate control (single-pass and two-pass)
• Temporal RDO
• Scene cut detection
• CLI tool and C API

Weekly pre-release

Improvements

• More x86 assembly from dav1d
• Validate the frame size to reject impossibly small frames.
• Initial SSE4.1 support

Changes

• serde is now optional
  • the fastrdo training feature is optional as well.

Fixes

• CDEF RDO could infiniloop in certain occasions.

Weekly pre-release

Improvements

• SSSE3 support
• Aarch64 neon support
• FlipADST
• Additional asm tests (CDEF, Inverse transform).
• Make sure all the dependencies are using the same version of syn, reducing the overall build time.

Changes

• The nasm feature had been renamed asm.

Fixes

• Unbreak C-API building on rust 1.38 and later.

Weekly pre-release

Improvements

• More API documentation
• Better error reporting
• Nicer error messages in the cli application
• Explicit validation of the encoder configuration
• The input Frame type can be expressed
• Added cargo-fuzz support and documented its usage
• Better rate-control support
• Two-pass rate-control support
• Additional speedups over all the speed levels on x86_64

Changes

• Tiles are now expressed in linear units (before were log2-units)
• The for color config enum variants now use the AV1 spec names
• The C-API is now part of the main repo, use cargo-c to build it
• send_frame can consume bare Frames
• Speed preset overhaul and rebalance

Fixes

• Invalid encoder settings, that would trigger panics on execution, are now caught on validation phase.
• Desync caused by a discrepancy of what would be the tile dimension and the loop restoration filter dimension when using certain tile sizes.

Weekly pre-release

New features

• Equally sized tiles support 🥇

Bugfixes

• Unbreak building on non-x86
• Make sure avx2 code is not called if only avx is present

API Changes

• set_limit() has been dropped. Use flush() instead.

Internals

• More coding lint enabled
• Additional speedups
• More documentation added

Weekly pre-release

New features:

• Enabled assembly for Windows x86_64 builds.
• Speed-ups in Motion Estimation, with new x86/x86_64 simd optimizations.
• Diamond Search in motion estimation.
• RDO Speedups

Bugfixes:

• Fix the memory allocation alignment issues found on windows x86_64.
• Fix container_sequence_header()
• Additional fixes to the 422 codepaths.
• More smaller and larger bugfixes

API Changes:

• Large simplification: now it is possible to encode by using onlysend_frame() and receive_packet().
• Update and document EncoderStatus variants.

Internal changes:

• Lots of refactoring to prepare for new features (tile support, tile and segment threading)
• Improvements to debug builds and tests to accelerate development
• Additional tests

Weekly pre-release

New Features:

• Fast 8bit codepath
• Shell auto-completion support
• Add a --skip option

Bugfixes:

• Depend on a y4m version that properly supports odd-dimensioned content
• Fix and enable directional intra prediction modes

API Changes:

• Add an explicit thread field in the Configuration to manage the threadpool using the API
• Cap the default threadpool to 4 maximum threads
• Default the tuning to psy
• Remove the explicit lookahead logic from the API

Weekly pre-release

Features:

• speed 0 is now up to 3x faster with no quality loss due to new early-exit implementation
• support for writing keyframe locations to a file for chunked encoders
• initial threading support (motion estimation only)
• tweaked quantizer modulation for better quality
• one-pass rate control
• 10% faster at speeds 3 and above due to addition of diamond me
• preparations for fast 8-bit mode

Fixes:

• bugfixes for HDR and colorspace obus (#977)

Weekly pre-release

• Workaround #987 pending a proper fix
• Further quality tuning