RE: [Deinterlace-discuss] Is a non-linear stretch mode planned?
Brought to you by:
adcockj,
dschmelzer
Menu
▾
▴
RE: [Deinterlace-discuss] Is a non-linear stretch mode planned?
|
From: Paul S. <psi...@in...> - 2001-01-10 22:00:36
|
Hi there, Mark's just outlined what I was thinking, but in much clearer and technically-correct terminology. The key is getting the CPU usage low enough to not have to require a 1.2 GHz Athlon (or two) to get the job done. Will probably have to be very MMX heavy... > -----Original Message----- > From: dei...@li... > [mailto:dei...@li...]On Behalf Of > Mark Rejhon > Sent: Wednesday, January 10, 2001 1:49 PM > To: dei...@li... > Subject: Re: [Deinterlace-discuss] Is a non-linear stretch mode planned? > > > Hi, > > Basically, it would be some kind of an algorithm that does the > pre-distort on a scanline basis (panorama) before sending to the > video overlay hardware to do the scaling. > > The video overlay hardware will actually continue to do the > scaling, of course. > > Should be doable. Might need a faster CPU, but fast-enough CPU exists > nowadays to do stuff like this, I believe. It would be a huge challenge > to get the CPU requirement low enough though. > > I give my vote for this panorama feature!! > > > > > Michael Eskin wrote: > > > > I don't think this is going to fit in with how dTV does its > scaling, since > > it uses the overlay scaling capability of the graphics controller to > > accomplish its display, and as far as I know, all graphics > adapters pretty > > much just implement linear scaling on their overlay surfaces. > > > > Of course, I'd love to be proven wrong here... I can't see the host > > processor doing the scaling, maybe on a P4? > > > > Cheers, > > > > Michael Eskin > > Conexant Systems > > > > ----- Original Message ----- > > From: "Paul Simoneau" <psi...@in...> > > To: <dei...@li...> > > Sent: Wednesday, January 10, 2001 6:45 AM > > Subject: RE: [Deinterlace-discuss] Is a non-linear stretch mode planned? > > > > I'll agree that there are cases when the APAC feature isn't used. > > It's disabled when using high-bandwidth inputs, so if we could get > > dTV to implement something to fill out the entire raster, I think > > that would be a welcome feature. > > > > As far as the Panorama mode, I believe it's semi-hyperbolic at the > > edges, but remains pretty linear in the middle section of the screen. > > If you're watching a TV channel that's got a stock ticker at the bottom > > of the screen, you should be able to see large amounts of stretching > > going on at the edges, but once you get to the middle, very little > > is done to the image. > > > > I think some sort of balancing act between two main > requirements would be : > > 1) leave as much of the original image unaltered as possible > > 2) that which you do have to touch, tread lightly > > > > Sure, it'd be great to fill the whole screen, but I'd rather avoid > > having a distracting "circus mirror" (due to too much stretching) > > or "banding" (poor image interpolation) effects on the image. The > > Philips does a pretty good job, and as I said before, I watch all > > my 4:3 sourced material in that mode. > > > > Once again, how much processing this would take is a major > concern, also. > > > > We currently render a 480 pixel image height for the image. To > > form a 16:9 version of this, we're in the neighborhood of 853x480. > > So, given our 720-pixel width, we need to "create" ~130 pixels more > > than the original 720 pixels to fill out that 16:9 frame. > > > > > -----Original Message----- > > > From: dei...@li... > > > [mailto:dei...@li...]On Behalf Of > > > Eric Schmidt > > > Sent: Tuesday, January 09, 2001 6:37 PM > > > To: dei...@li... > > > Subject: RE: [Deinterlace-discuss] Is a non-linear stretch > mode planned? > > > > > > > > > This panaramic stretch feature has actually been on my private list of > > > features I'd like to see in the future. I am unfamiliar with > any of the > > > real "guts" of the dTV program so I don't know what size job > it would be. > > > > > > I would guess that it's a hyperbolic transformation in x where as > > > the normal > > > anomorphic stretch is linear. I don't have any evidence to > support this, > > > just a feeling. If I were to take on the challenge, I'd be sure > > > to do some > > > measurements with a grid input signal first. But I do not > believe it's > > > simply a matter of stretching the edges and leaving the > center unchanged. > > > > > > Also, the APAC feature on the 64PH9905 only kicks in when watching 4:3 > > > signal through either the set's Composite or COAX video inputs. > > > Not through > > > the YPrPb or VGA inputs which I would guess those of us who > have this set > > > are using with our HTPCs. Just mentioning this to support > the need for > > > orbiting in dTV. > > > > > > -Eric > > > > > > | -----Original Message----- > > > | From: dei...@li... > > > | [mailto:dei...@li...]On Behalf Of > > > | Paul Simoneau > > > | Sent: Monday, January 08, 2001 6:39 PM > > > | To: dei...@li... > > > | Subject: RE: [Deinterlace-discuss] Is a non-linear stretch > mode planned? > > > | > > > | > > > | > > > | To add to the point : > > > | > > > | I have a very similar HDTV model, the Philips 55PP9701, > with the only > > > | differences being the size and the lack of an integrated ATSC tuner. > > > | So, I can expand a little bit on what Steve proposes. > > > | > > > | The Philips models offer a few of the "standard" widescreen > "stretch" > > > | modes, and like many other manufacturer's sets, the only > operate in the > > > | interlaced mode. Once a progressive or HDTV input is > selected, you're > > > | stuck in HDTV/progressive-squeeze mode. No aspect ratio > manipulation. > > > | So, this isn't a problem unique to Philips, but to most HDTV sets. > > > | > > > | 1. Standard 4:3 mode : 4:3 frame with black bars on the sides > > > of the image > > > | 2. 16x9 mode : do the anamorphic squeeze > > > | 3. Theater-1 : zoom in on a 4:3 image which contains a > 1.85:1 letterbox > > > | 4. Theater-2 : zoom in MORE on a 4:3 image which contains a > > > | 2.35:1 letterbox > > > | 5. Panoramic : approximately the middle 1/2 of the 4:3 image is left > > > | intact, while the outer 1/4's of the image > on either side > > > | are "streeeetched" to fill the rest of the raster and > > > | prevent the need for any bars. The further > towards the > > > | edge you go, the more stretch is introduced. > > > | > > > | I view all 4:3 source material in the Panorama mode. The stretching > > > | on the sides isn't too objectionable, and you don't lose any of the > > > | original source frame (as you do in the Theater-X modes). > > > | > > > | (BTW, the Philips already incorporate an "orbit" feature. > > > | Automatic Phosphor Aging Compensation - APAC.) > > > | > > > | > > > | I think this would be a big win for all widescreen users of dTV. > > > | > > > | However, given that there would have to be a field-by-field > > > | manipulation in order to do some kind of stretch function on > > > | the source data, how much additional CPU horsepower would this > > > | require ? Quite a bit, I might imagine. > > > | > > > | Would be nice, though... > > > | > > > | > > > | Paul Simoneau > > > | Integral Access, Inc. > > > | psi...@in... > > > | > > > | > > > | > > > | > > > | > -----Original Message----- > > > | > From: dei...@li... > > > | > > [mailto:dei...@li...]On Behalf Of > > > | > Daniel Schmelzer > > > | > Sent: Monday, January 08, 2001 10:34 PM > > > | > To: dei...@li... > > > | > Subject: [Deinterlace-discuss] Is a non-linear stretch > mode planned? > > > | > > > > | > > > > | > Hi All-- > > > | > > > > | > Interesting requests from SteveT on AVS Forum at: > > > | > > > > | > http://www.avsforum.com/ubb/Forum12/HTML/006836.html > > > | > > > > | > "First, let me say thanks to all the dTV developers for making > > > | > this awesome > > > | > piece of software available. This work of art the rivals (and in > > > | > many ways > > > | > is better than) the Genesis de-interlacer in my Philips > 64PH9905 HD > > > | > RPTV--particularly with dss and video game sources. > > > | > > > > | > I have a question however, regarding the "stretch" mode. Since > > > | > the majority > > > | > of dss programming is in 4:3 AR, I have gotten quite used > to using the > > > | > 9905's "Panoramic" video mode. This mode stretches the edges of > > > | the image > > > | > but leaves the center untouched. The result of this non-linear > > > | > stretch is a > > > | > good alternative to viewing 4:3 material on a 16:9 screen. > > > And while not > > > | > perfect, it is far less distracting than a linear stretch where > > > | > the entire > > > | > image is stretched horizontally. I believe the type of > viewing mode is > > > | > standard on most, if not all HD and HD Ready sets. IMHO, > it would be a > > > | > wonderful addition to dTV's feature set. > > > | > > > > | > On a similar topic, are there any plans to incorporate any > > > form of image > > > | > "orbiting" (a la yXy) to reduce image burn-in when watching > > > | > letterboxed or > > > | > side-paneled material? > > > | > > > > | > Thanks again for the wonderful work! > > > | > > > > | > Regards -- Steve" > > > | > _________________________________________________________________ > > > | > Get your FREE download of MSN Explorer at http://explorer.msn.com > > > | > > > > | > > > > | > _______________________________________________ > > > | > Deinterlace-discuss mailing list > > > | > Dei...@li... > > > | > http://lists.sourceforge.net/mailman/listinfo/deinterlace-discuss > > > | > > > > | > > > | > > > | > > > > > > > > > _______________________________________________ > > > Deinterlace-discuss mailing list > > > Dei...@li... > > > http://lists.sourceforge.net/mailman/listinfo/deinterlace-discuss > > > > > > > _______________________________________________ > > Deinterlace-discuss mailing list > > Dei...@li... > > http://lists.sourceforge.net/mailman/listinfo/deinterlace-discuss > > > > _______________________________________________ > > Deinterlace-discuss mailing list > > Dei...@li... > > http://lists.sourceforge.net/mailman/listinfo/deinterlace-discuss > > -- > _______________________________________________________________________ > ____ > Mark D. Rejhon Win2k.Linux.Win98 \ / mailto:ma...@ot... > http://www.marky.com/ C.C++.VB.Shell \/ AlphaWorld Home 10S 15W > > "A friend is someone who will be there without asking anything of you. > A friend is someone you know that knows you, and accepts you." > _______________________________________________________________________ > > _______________________________________________ > Deinterlace-discuss mailing list > Dei...@li... > http://lists.sourceforge.net/mailman/listinfo/deinterlace-discuss > |
