README
This commit is contained in:
parent
a6d204dc42
commit
14a95f0c18
1 changed files with 14 additions and 0 deletions
14
README.md
Normal file
14
README.md
Normal file
|
@ -0,0 +1,14 @@
|
||||||
|
# img-stream-enc
|
||||||
|
|
||||||
|
continuous jpg encoder for piping raw YUV420P images to. Works well enough for 0.5 qual. For some reason anything higher tends to corrupt the output file.
|
||||||
|
|
||||||
|
Zig std lib only, Uses SIMD Vector optimization + pre-computing coefficients for quantization. Bottleneck is on rle, huffman code and output.
|
||||||
|
|
||||||
|
**Not intended for actual use**. Again higher qual tends to corrupt files, Corruption is progressive, and pixels start getting wonky well before, decoders complain, image is still usable though
|
||||||
|
|
||||||
|
`zig build` to a bin. Usage is `img-stream-enc <width> <height> <quality> <n_jobs>`
|
||||||
|
|
||||||
|
* `width`: pixel width of input image (must be multiple of 16)
|
||||||
|
* `height`: pixel height of input (multiple of 16)
|
||||||
|
* `quality`: quality factor, float between 0 and 1
|
||||||
|
* `n_jobs`: number of threads to spawn for quantization
|
Loading…
Reference in a new issue