thanks daniel- now this begins to make sense: http://wiki.multimedia.cx/index.php?title=YCbCr_4:2:0 says:
"Since 6 bytes, or 48 bits represent 4 pixels, an average of 12 bits are required to represent a single pixel in this scheme." so the 48 bit depth in the set window command is for each 2x2 block. i guess they dont put all 4 Y samples together cause they are on different scanlines. if you look at the slope on the edge of the taillight you can see that each pair Y samples are switched, as you have stated. allan On Sat, Aug 30, 2008 at 5:54 PM, Daniel Gl?ckner <daniel-gl at gmx.net> wrote: > The picture is in YCbCr 4:2:0 format. > > 0. byte: Y(1,0) > 1. byte: Y(0,0) > 2. byte: Cr(0.5,0.5) > 3. byte: Cb(0.5,0.5) > > 4. byte: Y(3,0) > 5. byte: Y(2,0) > 6. byte: Cr(2.5,0.5) > 7. byte: Cb(2.5,0.5) > : > : > 4796. byte: Y(2399,0) > 4797. byte: Y(2398,0) > 4798. byte: Cr(2398.5,0.5) > 4799. byte: Cb(2398.5,0.5) > > 4800. byte: Y(1,1) > 4801. byte: Y(0,1) > > 4802. byte: Y(3,1) > 4803. byte: Y(2,1) > : > : > 7198. byte: Y(2399,1) > 7199. byte: Y(2398,1) > > repeat for the following 7200 byte blocks > > The chroma offset is guessed. > It would need some sharp colorful edges to verify that. > > Daniel > > -- > sane-devel mailing list: sane-devel at lists.alioth.debian.org > http://lists.alioth.debian.org/mailman/listinfo/sane-devel > Unsubscribe: Send mail with subject "unsubscribe your_password" > to sane-devel-request at lists.alioth.debian.org > -- "The truth is an offense, but not a sin"
