Log in

View Full Version : question about yv12 and capture from ntsc


fisix
16th February 2003, 13:58
i'm asking this here because the experts in yv12 seem to reside here.

i have a wintv xxxx card and it has the option of capturing in '4:2:2 packed' and 'YUV12 planar'. am i right when i assume that YUV12 is YV12?

i've read through the faq and see that the effective bpp for yuy2 is 16 and for yv12 is 12. i'm sure you've answered this before, but is this a quality difference you might see in the end product of captured avi->divx?

captured processing => avisynth 2.07
vob processing => avisynth 2.5

or would anyone like to argue that the bpp difference isn't noticeable OR that the speed increase (especially since it seems i can capture directly to yv12) makes up for it? heck, for anime, it must be worth it, i wonder about live action though.

-fisix

trbarry
16th February 2003, 16:17
The actual data in a transmitted analog TV signal is more like YUY2 than YV12. That is, chroma is transmitted on every line.

So unless you gain some needed performance improvements you are probably better off leaving it in YUY2, at least until you deinterlace it. The process of making YV12 out of YUY2 interlaced source is ugly and lossy.

The push for YV12 Avisynth support came partially for speed considerations but also because much of the data used in Avisynth was from DVD's or HDTV where it is already effectively in YV12 form.

But you can try it and if you can't see the difference then YV12 would probably make smaller faster files.

- Tom

sh0dan
16th February 2003, 17:08
It highly depends on the hardware capturing the signal. If it captures a field at the time, YV12 should very well be good enough.

This is however theoretical until you do some actual tests.

We can talk when you have some screenshots or test sequences! :)

McQuaid
17th February 2003, 01:10
I also have YUV12 as an option with my capture card. Is that the same thing as YV12? I use picvideo, and from what I understand is only YUY2. But reading the thread on huffy, -h was considering adding a near lossless method while he's already working on YV12 support. From what I understand this would benefit being who capture as well, and not just dvd sources. Or maybe I've got this wrong.