Hi, On Thu, Mar 8, 2012 at 4:07 PM, Paul B Mahol <[email protected]> wrote: > On 3/8/12, Paul B Mahol <[email protected]> wrote: >> On 3/8/12, Paul B Mahol <[email protected]> wrote: >>> On 3/8/12, Justin Ruggles <[email protected]> wrote: >>>> On 03/08/2012 11:08 AM, Paul B Mahol wrote: >>>> >>>>> >>>>> Signed-off-by: Paul B Mahol <[email protected]> >>>>> --- >>>>> Changelog | 1 + >>>>> doc/general.texi | 4 ++- >>>>> libavcodec/Makefile | 1 + >>>>> libavcodec/allcodecs.c | 1 + >>>>> libavcodec/avcodec.h | 1 + >>>>> libavcodec/version.h | 2 +- >>>>> libavcodec/xbmenc.c | 93 >>>>> ++++++++++++++++++++++++++++++++++++++++++++++++ >>>>> libavformat/img2.c | 1 + >>>>> libavformat/img2enc.c | 2 +- >>>>> 9 files changed, 103 insertions(+), 3 deletions(-) >>>>> create mode 100644 libavcodec/xbmenc.c >>>> >>>> [...] >>>> >>>>> +static av_cold int xbm_encode_init(AVCodecContext *avctx) >>>>> +{ >>>>> + if (avctx->width & 7) { >>>>> + av_log(avctx, AV_LOG_ERROR, "width is not divisible by >>>>> eight.\n"); >>>>> + return AVERROR(EINVAL); >>>>> + } >>>> >>>> >>>> Wikipedia says this is allowed: >>>> "If the image width does not match a multiple of 8, the display >>>> mechanism ignores and discards the extra bits in the last byte of each >>>> row." >>>> >>>> The input frame should also be padded in such cases, so it should be >>>> trivial to support. >>> >>> Wrong, mono is bitstream format, implementing support for width not >>> multiple >>> of 8 is going to be a little bit ugly. >>> >>> Perhaps byte variant should be added to sws? >>> >>> FYI both ImageMagick and Gimp (and Libav) appears to encode/decode >>> HHHHxWWW3 pbm wrong. >>> >> Actually not, ImageMagick & Gimp decodes ImageMagick output just fine. >> >> On other hand our pbm decoder displays black rows on right side. >> Our pbm encoder writes random crap on right side too. >> > > As Justin just pointed out on irc this is swscale bug.
Please file yet another bugzilla bug. ;-). Ronald _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
