[FFmpeg-devel] [RFC] What should pixdesc.h:write_line() do with PAL formats?
Tue Mar 31 22:56:06 CEST 2009
read_line() currently reads from the image the components
corresponding to c and puts them in dst:
(index, c) -> comp(index, c)
The inverse operations cannot be done, so maybe write_line() should
simply take in input a list of indexes and write them down to the
destination image, but this will not be simmetric with read_line().
To remove the asymmetry we could treat PAL formats like non-PAL ones,
so having read_line() read the indexes and writing them to dst, while
write_line() would write the indexes back to the image.
Then we could define a new function:
read_line_pal(const uint16_t *src, uint8_t *data, const int linesize,
const AVPixFmtDescriptor *desc, int x, int y, int c, int w);
with the current behavior as read_line().
Would this fit well?
FFmpeg = Fierce Fast Mournful Prodigious Eretic Goblin
More information about the ffmpeg-devel