I'm curious about this "higher quality recordings" phrase. I've seen it a few times,
yet I'm not one to automatically believe that a card does better just because it does
mpeg 2 compression. (I'm assuming a 150 would have the same quality if one got
that working.)
It's not the MPEG2 compression that inherently gives it "higher quality recordings." I know some people who have CPU to burn like capturing with a software card so they can do direct to MPEG4 with denoising, etc. I looked into this awhile back, and came up with some interesting conclusions.
The BT878/9 uses a simple notch filter like older NTSC TV's to separate the Y/C from composite signals. That limits the luma resolution to less than 3 MHz (as well as above 4.5 MHz, but broadcast won't have any luma up there). That limits luma "lines of resolution" to about 240... or max 4:3 capture at about 320x480. Thus, anything over 352x480 for a BT878/9 card is silly, since they chip itself can't capture more than that. I used a resolution test pattern to confirm this with my bt878-based card.
The hauppauge [23]50 cards use a Philips SAA7115 capture chip. That has a 2D comb filter, which in perfect circumstances captures *all* luma information up to the maximum 6 MHz channel width. Broadcast limits luma to 4.2 MHz (IIRC), but even still, that's 4.2*80=336 "analog lines of resolution." That means a maximum 4:3 capture of 448x480.... so 480x480 is about all that broadcast can do. My testing of my 250 with the same resolution test pattern confirmed that anything over about 540x480 is silly for a pvr-250.
The hauppauge [15]50 uses the cx88 chip with 10-bit A/D's for capture. It has a bit more advanced comb filter than the Philips chip, but that will mainly help some color artifacts. A software-based capture card based on one of these has the potential to have higher resolution.
Has anyone ever done a direct comparison between a 250 and other ordinary bt87x capture cards? One would think it was the ultimate quality of the A/D circuitry that was the key...
As you can probably tell, I've wasted^H^H^H^H^H^H spent a fair bit of time looking at this. I was really only looking at the "softness" of the image, which is a function of the luma resolution. Lots of other things go into the mess as well (noise susceptibility, color decoder accuracy, hue/sat, A/D density, scaler filtering, etc) that affect "quality." I'd be willing to argue that a software-based capture card with a newer video decoder chip (such as the cx88) could have equal/better quality than the hauppauge cards, but it'll burn up a helluvallotta CPU to do so.
-Cory
************************************************************************* * Cory Papenfuss * * Electrical Engineering candidate Ph.D. graduate student * * Virginia Polytechnic Institute and State University * *************************************************************************
_______________________________________________ mythtv-users mailing list mythtv-users@mythtv.org http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
