[theora] [?] About YUV formats...

Ralph Giles giles at xiph.org
Tue Nov 7 15:14:31 PST 2006

On Tue, Nov 07, 2006 at 12:04:42PM +0100, Andrea Fontana wrote:

> How theora manage them internally? I mean: if i encode the same video 
> twice, first time encoding it with YUV444 packets and then with YUV420, 
> will second video be smaller than first (due to subsampling...)? Or theora 
> compress them in the same way?

The theora format allows three different chroma sampling formats for 
the compressed data, and two specific colorspace definitions (to match 
PAL and NTSC oriented content, roughly).

So if you compress 4:4:4 video and 4:2:0 video, the 4:2:0 version will 
generally be smaller and poorer quality, but depending on the source
maybe not by much. Unfortunately, there's isn't actually an encoder that 
supports anything but 4:2:0 at the moment, so I can't give you any 
specific numbers.

The example_encoder client in theora-exp has routines to convert between 
the various chroma sampling schemes to the three theora supports, but 
the api in both libraries is designed to just take and encode frames in 
whatever format you configure it to expect.


More information about the theora mailing list