Well, I tricked y4mscaler into doing what I want, but the results were a bit discouraging. To get a horizontal lowpass spatial filter with cutoff N/D<1 I just did a downsample by N/D followed by an upsample by N/D (using the sinc8lan kernel). What I found was that (to my eye) there was no visible difference between the original and a lowpass version with cutoff 1/2 (or even a little less). Furthermore this operation had essentially no effect on bitrate.
This suggested I should just downsample by 2 and encode at 352x480 (side question: should I clip 16 pixels and scale by 1/2, or not clip and scale by 22/45?). So I added these to my bitrate table: 720x480 yuvdenoise : 7700 kb/s 720x480 yuvdenoise, --reduce-hf : 4900 kb/s 352x480 yuvdenoise : 5100 kb/s 352x480 yuvdenoise, --reduce-hf : 2400 kb/s So, it appears that both visually and by bitrate that throwing away high-frequency info (that my source doesn't have anyway) either by downsampling or by using --reduce-hf has about the same effect. Using --reduce-hf at 352x480 lowers the bitrate a lot more but is quite visible, since the meaning of "high frequencies" has been scaled down by 2. Dan ------------------------------------------------------- This SF.net email is sponsored by:Crypto Challenge is now open! Get cracking and register here for some mind boggling fun and the chance of winning an Apple iPod: http://ads.sourceforge.net/cgi-bin/redirect.pl?thaw0031en _______________________________________________ Mjpeg-users mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/mjpeg-users
