Skip to content
Graphics, images & media

Graphics, images & media

Graphics, images & media

The 2D socle everything above draws through, and the pixels and frames it moves.

OrganisationReposSiteDocsWhat it holds
go-gfx3sitedocsThe shared 2D socle: an anti-aliased vector rasteriser, a raster surface, colour, geometry, resampling, and SVG to bitmap.
go-images7sitedocsImage processing at scikit-image scope, plus the decoders the standard library does not have or does differently: Photoshop PSD composites, and forks of image/gif, image/jpeg and image/png that upsample four-component chroma the way libjpeg does and can report what a truncated stream did decode, so a half-arrived picture still shows. And one that is not processing at all: depth from a single photograph, and the second eye synthesised from it.
go-avkit7sitedocsAn audio/video toolkit: MP4/ISO-BMFF, Matroska/WebM and MPEG-TS onto one model, demuxed and muxed — and under it the bitstreams themselves, H.264 and H.265 read down to clause 8.2’s picture order counts, VP8 and VP9, over the primitives they share rather than once per codec: RBSP and Exp-Golomb bit reading, NAL framing, and the boolean arithmetic decoder VP8 and VP9 have in common.
go-streamkit1sitedocsSite-agnostic transport for streaming media: a retrying, rate-limited HTTP client, an HLS parser and an MPEG-DASH manifest reader.
go-extractors1sitedocsA plugin framework for media extractors — standalone plugin executables answer what is at a URL and what can be downloaded from it.
go-xrkit5sitedocsXR glasses: head orientation, stereo frame packing, flat, equirectangular and fisheye projection, a 360° screen ribbon, a graded catalogue of what each headset can show — plus a player, a virtual desktop on the ribbon, and Android capture.
go-viture2——The VITURE headsets go-xrkit drives: the Beast and the Luma Ultra, spoken to directly with no vendor library.
go-xreal1——The XREAL Air family’s USB packet protocol, the other half of what go-xrkit reads head orientation from.
go-videodsp1——Planar YUV to RGBA, the conversion every video decoder in the map ends with, kept in one place rather than once per codec.

Counts are public repositories that hold code; brand, docs and landing repositories are excluded.