nx_vcodec_color_codec.nx
buildroot/runtime/nx_vcodec_color_codec.nx
about
nx_vcodec_color_codec.nx -- COLOR on the REAL engine. Rather than ship my simpler WHT codec (H5/H6), this
runs the proven sophisticated nx_vcodec (directional intra prediction + sovereign H.264 integer DCT + the
self-describing packed bitstream) on EACH of the Y/Cb/Cr planes at 4:2:0, fed by the H4 RGB<->YCbCr
transform. DRY: the codec engine is the one already wasm-proven + live-deployed; this adds only the color
plane orchestration. Chroma planes are W/2 x H/2 (must be >=16 = one macroblock, so W,H are multiples of
32). Pure integer, sovereign. license_tier: ORIGINAL
dependencies 2 imports · 4 importers
imports: nx_vcodec.nxnx_vcodec_color_frame.nx
imported by: nx_vcodec_color_codec_test.nxnx_vcodec_color_gop_vm_gate.nxnx_vcodec_color_wasm_vm_gate.nxnx_vcodec_speed_bench.nx
structs
| none |
consts
| none |
functions
| 12 | func vc_color_encode_key(rgb: *u8, W: i64, H: i64, qp: i64, sad: i64, strY: *u8, strCb: *u8, strCr: *u8, bits_out: *i64) -> i64 |
| 28 | func vc_color_decode_key(W: i64, H: i64, qp: i64, strY: *u8, strCb: *u8, strCr: *u8, rgb_out: *u8) -> i64 |
| 47 | func vc_color_encode_p(rgb: *u8, W: i64, H: i64, qp: i64, kf: i64, sad: i64, prevY: *u8, prevCbs: *u8, prevCrs: *u8, strY: *u8, strCb: *u8, strCr: *u8, bits_out: *i64) -> i64 |
| 63 | func vc_color_decode_p(W: i64, H: i64, qp: i64, prevY: *u8, prevCbs: *u8, prevCrs: *u8, strY: *u8, strCb: *u8, strCr: *u8, rgb_out: *u8, decY: *u8, decCbs: *u8, decCrs: *u8) -> i64 |
| 76 | func main() -> i64 { return 0 } |