[opendtv] Re: 720P = 540i ?
- From: Tom Barry <trbarry@xxxxxxxxxxx>
- To: opendtv@xxxxxxxxxxxxx
- Date: Tue, 03 May 2005 18:07:53 -0400
Jeroen Stessen wrote:
> A yet better solution, which we don't typically see for HDTV, is
> based on application of the "Generalized Sampling Theorem". This
> says that any two sets of samples, they don't have to be spaced
> evenly as long as they don't coincide exactly, can be used for
> reconstructing the original signal at the double Nyquist
bandwidth.
> With the use of motion vectors a "motion compensated"
de-interlacer
> can get the missing lines from previous frames, at least if the
> vertical component of the motion is not near a "critical speed".
>
Jeroen -
I don't suppose there exist any very very simple links on how that
process would work? I admit I'm skeptical.
Recently I've been becoming gradually convinced that in order to
completely remove aliasing artifacts from, say, 1080i you would
have to vertically filter to only 540 vertical pixels of
resolution. But if people were really doing that then I'm not
sure there would be ANY additional information available from
looking at adjacent fields, no matter how good or powerful your
motion compensation logic was. And that would even be true for
still scenes.
Hopefully I'm mistaken but, if not, maybe I'll go write a book on
helicopters or how deinterlacing doesn't work. ;-)
- Tom
> Hi,
>
> Bob Miller wrote:
> http://hardware.slashdot.org/hardware/05/05/02/120227.shtml?tid=129&tid=196&tid=137
>
>
>>This thread says that received 1080i on a native 720P will see the
>>conversion of a single 540i frame into 720P by 1980.
>
>
> I should hope not !
> Most if not all de-interlacers for HDTV are "motion adaptive".
> On still images, and on 24p or 30p film mode, they will weave two
> 540i fields into one 1080p frame. No problem at all. If you consider
> that much material is originated from 24p film, or from 24p video
> productions, then that is really 1080@24p transported as 1080@30i.
> Perfect reconstruction of the 1080p is then trivially easy, at the
> cost of one or two field memories in the receiver.
>
> It is only with moving images in (50i/60i) video mode that there is
> a fundamental problem for de-interlacers. There will no longer be
> two fields from the same motion phase, so weaving does not work
> anymore. Then intra-field interpolation, indeed from 540 lines, is
> the easiest (but incorrect) solution. Motion adaptive de-interlacers
> will necessarily switch to intra-field mode. The smarter MA designs
> make such decision (between still/film and video mode) for each
> individual pixel, preserving weaving e.g. for still backgrounds.
>
> A yet better solution, which we don't typically see for HDTV, is
> based on application of the "Generalized Sampling Theorem". This
> says that any two sets of samples, they don't have to be spaced
> evenly as long as they don't coincide exactly, can be used for
> reconstructing the original signal at the double Nyquist bandwidth.
> With the use of motion vectors a "motion compensated" de-interlacer
> can get the missing lines from previous frames, at least if the
> vertical component of the motion is not near a "critical speed".
>
> (A vertical motion of 2N+1 lines per field period reduces a 1080i
> grid to a real 540p grid, it makes the two sets of samples coincide
> and thus it makes de-interlacing and reconstruction of the full
> vertical resolution according to the GST fundamentally impossible.)
>
>
>>How does this impact what we you were discussing about
>>1080p>480P>720P?
>
>
> Not. We weren't discussing 1080i->1080p conversion, and it will
> only go wrong for moving images in video mode, which are can be
> largely avoided by choosing 720p for difficult (sports) programs.
> We should have some faith in today's decent de-interlacers.
>
>
>>How would that 720P compare to 720P converted in the former way
>>from 1080i? How would 640 x 480P compare to 1920 x 540i?
>
>
> Well... using only half the lines from a sharp 1080 line image
> implies a serious aliasing problem. Re-sampling is not going to
> make that aliasing go away. On the contrary, it can make it
> forever impossible to reduce the aliasing by proper de-interlacing.
> So it would be a bad idea to apply scaling to an improperly de-
> interlaced image, which is of course why all decent scaler ICs
> contain also a decent de-interlacer (at least MA) that can do a
> proper weaving of matching fields in film- or still image mode.
> The "DCDi" type of de-interlacer pioneered by Faroudja (who later
> seems to have denied its effectiveness, very strange) does already
> a good enough job for most applications. So don't worry, the
> statement on slashdot is a distortion of the truth.
>
> Regards,
> -- Jeroen
>
> +-------------------------------+------------------------------------------+
> | From: Jeroen H. Stessen | E-mail: Jeroen.Stessen@xxxxxxxxxxx |
> | Building: SFJ-5.22 Eindhoven | Deptmt.: Philips Applied Technologies |
> | Phone: ++31.40.2732739 | Visiting & mail address: Glaslaan 2 |
> | Mobile: ++31.6.44680021 | NL 5616 LW Eindhoven, the Netherlands |
> | Pager: ++31.6.65133818 | Website: http://www.apptech.philips.com/
> |
> +-------------------------------+------------------------------------------+
>
>
>
>
> ----------------------------------------------------------------------
> 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.
Other related posts:
