CVS: main/libmpcodecs vf_scale.c,1.24,1.25
Update of /cvsroot/mplayer/main/libmpcodecs In directory mail:/var/tmp.root/cvs-serv19337 Modified Files: vf_scale.c Log Message: out formats reordered, quality order (we should prefer 422P to YV12 if available...) Index: vf_scale.c =================================================================== RCS file: /cvsroot/mplayer/main/libmpcodecs/vf_scale.c,v retrieving revision 1.24 retrieving revision 1.25 diff -u -r1.24 -r1.25 --- vf_scale.c 10 Sep 2002 22:18:32 -0000 1.24 +++ vf_scale.c 15 Sep 2002 13:36:37 -0000 1.25 @@ -27,30 +27,32 @@ //===========================================================================// static unsigned int outfmt_list[]={ +// RGB: IMGFMT_BGR32, - IMGFMT_BGR24, - IMGFMT_BGR16, - IMGFMT_BGR15, - IMGFMT_BGR8, - IMGFMT_BGR4, - IMGFMT_BGR1, IMGFMT_RGB32, + IMGFMT_BGR24, IMGFMT_RGB24, + IMGFMT_BGR16, IMGFMT_RGB16, + IMGFMT_BGR15, IMGFMT_RGB15, + IMGFMT_BGR8, IMGFMT_RGB8, + IMGFMT_BGR4, IMGFMT_RGB4, + IMGFMT_BGR1, IMGFMT_RGB1, +// YUV: + IMGFMT_444P, + IMGFMT_422P, IMGFMT_YV12, IMGFMT_I420, IMGFMT_IYUV, - IMGFMT_Y800, - IMGFMT_Y8, IMGFMT_YVU9, IMGFMT_IF09, - IMGFMT_444P, - IMGFMT_422P, IMGFMT_411P, + IMGFMT_Y800, + IMGFMT_Y8, 0 };
On Sun, Sep 15, 2002 at 03:36:40PM +0200, Arpi of Ize wrote:
Update of /cvsroot/mplayer/main/libmpcodecs In directory mail:/var/tmp.root/cvs-serv19337
Modified Files: vf_scale.c Log Message: out formats reordered, quality order (we should prefer 422P to YV12 if available...)
Shouldn't we prefer 422p only if scaling to yv12 would output a chroma plane of lower vertical size than the input? Or at least something like that...? Otherwise it could slow things down a good bit when 422p is available but not needed, e.g. if a source image in yv12 format is only being scaled horizontally, or being scaled up. Rich
Hi,
Modified Files: vf_scale.c Log Message: out formats reordered, quality order (we should prefer 422P to YV12 if available...)
Shouldn't we prefer 422p only if scaling to yv12 would output a chroma plane of lower vertical size than the input? Or at least something like that...? Otherwise it could slow things down a good bit when 422p is available but not needed, e.g. if a source image in yv12 format is only being scaled horizontally, or being scaled up.
dunno actually we should decide if speed or quality is more important 422p is better for quality and worse for speed, compared to yv12 but if someone wants realy fast vo, he should force yvu9 :) A'rpi / Astral & ESP-team -- Developer of MPlayer, the Movie Player for Linux - http://www.MPlayerHQ.hu
On Sun, Sep 15, 2002 at 07:23:29PM +0200, Arpi wrote:
Hi,
Modified Files: vf_scale.c Log Message: out formats reordered, quality order (we should prefer 422P to YV12 if available...)
Shouldn't we prefer 422p only if scaling to yv12 would output a chroma plane of lower vertical size than the input? Or at least something like that...? Otherwise it could slow things down a good bit when 422p is available but not needed, e.g. if a source image in yv12 format is only being scaled horizontally, or being scaled up.
dunno actually we should decide if speed or quality is more important 422p is better for quality and worse for speed, compared to yv12 but if someone wants realy fast vo, he should force yvu9 :)
I was just saying there are certain cases where 422p is not significantly better (scaling up) or not better at all (leaving height alone and just scaling width, e.g. for fixing aspect). At least in the latter case, it's a total waste of cpu time to use 422p. Rich
Hi,
actually we should decide if speed or quality is more important 422p is better for quality and worse for speed, compared to yv12 but if someone wants realy fast vo, he should force yvu9 :)
I was just saying there are certain cases where 422p is not significantly better (scaling up) or not better at all (leaving height alone and just scaling width, e.g. for fixing aspect). At least in the latter case, it's a total waste of cpu time to use 422p.
hmm. did you ever tried it? in vf_scale: if(fmt!=best) flags&=~VFCAP_CSP_SUPPORTED_BY_HW; this will avoid doing nonsense yv12->422p. it will only prefer 422p if colorspace conversion is made _and_ the output device supports 422p. A'rpi / Astral & ESP-team -- Developer of MPlayer, the Movie Player for Linux - http://www.MPlayerHQ.hu
On Sun, Sep 15, 2002 at 09:52:19PM +0200, Arpi wrote:
Hi,
actually we should decide if speed or quality is more important 422p is better for quality and worse for speed, compared to yv12 but if someone wants realy fast vo, he should force yvu9 :)
I was just saying there are certain cases where 422p is not significantly better (scaling up) or not better at all (leaving height alone and just scaling width, e.g. for fixing aspect). At least in the latter case, it's a total waste of cpu time to use 422p.
hmm. did you ever tried it?
in vf_scale: if(fmt!=best) flags&=~VFCAP_CSP_SUPPORTED_BY_HW;
this will avoid doing nonsense yv12->422p. it will only prefer 422p if colorspace conversion is made _and_ the output device supports 422p.
Oh, 10l. :(( Rich
participants (3)
-
Arpi -
Arpi of Ize -
D Richard Felker III