The filters were designed before pixellated displays were common. There was always the assumption that the image would have the "defocusing" effect of the crt spot. I believe that this is why compression artefacts look worse on pixellated displays, they have no post-filter. A simple layer of light diffuser would fix it. The big JVC projectors effectively do that, and the pictures look better. ----- Original Message ----- From: "Tom Barry" <trbarry@xxxxxxxxxxx> To: <opendtv@xxxxxxxxxxxxx> Sent: Saturday, December 11, 2004 5:39 PM Subject: [opendtv] Re: Why Europe should choose 720P for HDTV > Alan - > > But does that 42% amount account for the fact that fixed pixel displays > may be effectively tiny little squares with sharply defined edges? Or > is it intend for some CRT spot beam with approx Gaussian distribution? > > Or does it matter? > > BTW, It's been pointed out here on the list that my concerns about this > possible non-problem could easily be fixed anyway by simple blurring the > fixed pixels a bit as displayed. This could even more easily be done > optically in projectors like DLP's and I suspect many digital movie > theaters would look slightly better when very slightly out of focus if > the owners were not so proud of their ability to display non-Nyquist > graphics images (or concerned they hadn't waited for higher rez). > > Comments? > > - Tom > > > Alan Roberts wrote: > > > The Nyquist limit is exactly 50% of the sampling frequency, you have to > > filter to lower that this to get smooth motion. Typically, digital systems > > have analogue filters to prevent frequencies above 0.45 of sampling > > frequencies getting into the system. The ITU recommended filters for 601 and > > 709 sampling are both flat to 42%, then dive steeply to -12dB at 50%, > > allowing some aliases to creep in. Conventional wisdom is that this is a > > good idea, I don't go along with it fully because those aliases cause havoc > > in compressors, even at low levels. > > > > So, you need 20 pixels to show 18 lines or 9 cycles. The filtering refuses > > to allow such frequencies to start/stop suddenly, so you get confusion at > > the ends of a block of such frequencies, but it all looks fine once the edge > > effects have passed. It's interesting to work out the actual pixel pattern > > for such frequencies, and to see what happens when it all moves, I've done > > so and the results are fascinating, it explains why aliases are such a bad > > idea. > > > > ----- Original Message ----- > > From: "Tom Barry" <trbarry@xxxxxxxxxxx> > > To: <opendtv@xxxxxxxxxxxxx> > > Sent: Saturday, December 11, 2004 4:32 PM > > Subject: [opendtv] Re: Why Europe should choose 720P for HDTV > > > > > > > >>Mark - > >> > >>How much do you have to filter so that an image can move smoothly across > >>or down a fixed pixel display 1/2 pixel at a time without aperture > >>effect artifacts. Is it just to the Nyquist limit, or more? > >> > > > > > > > > > > > > ---------------------------------------------------------------------- > > You can UNSUBSCRIBE from the OpenDTV list in two ways: > > > > - Using the UNSUBSCRIBE command in your user configuration settings at FreeLists.org > > > > - By sending a message to: opendtv-request@xxxxxxxxxxxxx with the word unsubscribe in the subject line. > > > > > > > ---------------------------------------------------------------------- > You can UNSUBSCRIBE from the OpenDTV list in two ways: > > - Using the UNSUBSCRIBE command in your user configuration settings at FreeLists.org > > - By sending a message to: opendtv-request@xxxxxxxxxxxxx with the word unsubscribe in the subject line. > ---------------------------------------------------------------------- You can UNSUBSCRIBE from the OpenDTV list in two ways: - Using the UNSUBSCRIBE command in your user configuration settings at FreeLists.org - By sending a message to: opendtv-request@xxxxxxxxxxxxx with the word unsubscribe in the subject line.