Log in

View Full Version : Codec that supports fields: what does it mean?


PaulJBis
8th May 2005, 16:23
Hi all:

I've read in several threads, both here and in other boards, about how the MJPEG codec "supports fields", and I have also seen the option in the codec about "2 fields if more than XXX lines". What I don't understand is: what does that "field support" thing mean in practice? I have tried capturing to MJPEG with both that option on and off, and in both cases the end result appeared to be interlaced (you could see those typical "lines" in the image). I have also tried capturing in realtime to DIVX 5, just as an experiment (since I understand that DIVX doesn't support fields), and the image was also interlaced.

So... what does it mean to say that a codec "supports frames"? And since we are on it, does HuffyUV support them?

jggimi
8th May 2005, 16:38
It means that the codec will take interlacing into account for compression.Each field is compressed as if it is a separate video.Without this, encoding interlaced frames would be improperly compressed.

Use the feature only when encoding interlaced content for display on an interlaced device, such as a CRT television. Otherwise, best practice is to deinterlace and encode progressively for a progressive device, such as a PC monitor.

PaulJBis
8th May 2005, 18:06
I see. Thanks for your reply. So, which codecs (aside of MJPEG) do and don't support fields?

Boulder
8th May 2005, 18:21
Any decent codec for capturing should support interlacing as you should always capture both fields. And since you asked, HuffYUV has the option.

zilog jones
8th May 2005, 19:37
Even XviD has interlaced video support. So it's pretty much any *good* encoder.

sprit
10th May 2005, 22:33
Originally posted by jggimi
Without this, encoding interlaced frames would be improperly compressed.

Wouldn't the codec (possibly) compress the video stream less, rather than compress it improperly? I would be more concerned with improper colorspace conversions (f.ex. YUY2<->YV12).

zilog jones
10th May 2005, 22:42
I'm pretty sure it will encode it improperly, as it won't know that it has to keep each field intact therefore potentially destroying the "interlacedness" of the video.

jggimi
11th May 2005, 02:56
For an excellent description of MPEG encoding technology for lay people -- macroblocks, motion vectors, quantizers, etc. --- see http://www.divx.com/divx/divxpro/guides/. The section on video compression is very well written, gives graphical, easily understood examples, and is applicable to all sorts of MPEG codecs, not just DivX.

bb
17th May 2005, 20:26
I'd like to add that there are different ways of interlace awareness for encoders. For instance interlaced sources can be MPEG-2 encoded on a frame basis by using an alternate scan, or on a field basis, i.e. each field gets encoded separately.

bb