All pull requests

Pull requests

• avx2: improve blend operations in adm

  #1609·anematode opened Sep 21, 2026·

  ·
  1

• tools/vmaf: give model_collection_label a positive bound

  #1606·lusoris opened Sep 19, 2026·

  ·

• tools: read frames with an odd width or height

  #1604·lusoris opened Sep 19, 2026·

  ·

• test/checkasm: fix the source strides in the ADM dwt2 tests

  #1603·lusoris opened Sep 19, 2026·

  ·

• feature/adm: keep contrast masking thresholds in range

  #1602·lusoris opened Sep 19, 2026·

  ·
  2

• feature/adm: fix undefined behaviour in the 16-bit scale 0 dwt2

  #1601·lusoris opened Sep 19, 2026·

  ·

• feature/adm: fix undefined rounding constant in adm_cm for frame widths 17 to 32

  #1600·lusoris opened Sep 19, 2026·

  ·

• feature/adm: fix out-of-bounds read at scale 3 for frame dimensions 17 to 32

  #1599·lusoris opened Sep 19, 2026·

  ·

• .github/ci: compare checkasm benchmarks against the PR's base ref

  #1598·kylophone opened Sep 17, 2026·

  ·
  2

• cuda: build the kernels with clang without the CUDA toolkit headers

  #1596·birkdev opened Sep 17, 2026·

  ·
  4

• cuda: keep kernel parameter structs and accumulators out of local memory

  #1595·birkdev opened Sep 17, 2026·

  ·
  1

• Update link to models documentation

  #1593·forsakendaemon opened Sep 14, 2026·

  ·

• fix(thread): clean up when a worker cannot be created

  #1591·lusoris opened Sep 8, 2026·

  ·

• fix(model): preserve a collection when growth allocation fails

  #1590·lusoris opened Sep 8, 2026·

  ·

• libvmaf: add median and percentile C API pooling

  #1589·lusoris opened Sep 8, 2026·

  ·
  1

• model, feature: clean up option dictionaries on failure

  #1588·lusoris opened Sep 8, 2026·

  ·

• avx2: use vpabsd

  #1586·anematode opened Aug 22, 2026·

  ·
  6

• avx2: vectorize more of vif_statistic_8_avx2/vif_statistic_16_avx2

  #1585·anematode opened Aug 22, 2026·

  ·
  1

• avx2: vectorize a couple scalar parts

  #1584·anematode opened Aug 22, 2026·

  ·
  1

• cuda: fix end-of-stream double flush and unordered motion SAD reset

  #1583·jmsether opened Aug 19, 2026·

  ·
  1

• Fix out-of-bounds boundary reflection for tiny frames in motion feature

  #1582·kjg0724 opened Aug 17, 2026·

  ·
  1

  2

• Keep motion filter mirror() in bounds for tiny frames

  #1581·Hashim1999164 opened Aug 14, 2026·

  ·
  1

• cuda: Add compile test only for CI

  #1579·StormBytePP opened Aug 13, 2026·

  ·
  1

• speed: CUDA chroma SpEED-QA feature extractor

  #1574·samfrazerdutton opened Aug 10, 2026·

  ·

• CUDA: Fix compile, SIGSEGV and tests

  #1573·StormBytePP opened Aug 10, 2026·

  ·