Am I oversharpening
in [Digital Cameras]
|
From: JPS on 27 Jun 2005 17:54 In message <Hh5iEtFkK6vCFwp0(a)objectech.co.uk>, Ken Tough <ken(a)objectech.co.uk> wrote: >The objective tests at dpreview show that at high ISO the SD10 loses >saturation and hue accuracy. I suppose this shows up as performance >in the shadows too. One ISO's highlights are another ISO's shadows. They're just the same analog sensor signals digitized into different RAW number ranges (and precisions). -- <>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<> John P Sheehy <JPS(a)no.komm> ><<> <>>< <>>< ><<> <>>< ><<> ><<> <>><
From: jennifer.wilson2 on 2 Jul 2005 10:29 Bill Funk wrote: > Are you trying to say that each sensor site in an X3 sensor is a > "pixel"? > Or, what is it you're trying to say? It should be obvious that it takes three sensors, one red, one green, one blue, to provide the information for a full color output pixel. So Foveon's full color photosites do correspond to pixels, but Bayer's don't, until each color exposure is interpolated up to the size of the interpolated recorded image. Bayer cameras record upsclaled images, as George likes to say.
From: JPS on 2 Jul 2005 10:54 In message <1120314584.614337.230530(a)z14g2000cwz.googlegroups.com>, jennifer.wilson2(a)lycos.com wrote: >Bill Funk wrote: >> Are you trying to say that each sensor site in an X3 sensor is a >> "pixel"? >> Or, what is it you're trying to say? >It should be obvious that it takes three sensors, one red, one green, >one blue, to provide the information for a full color output pixel. No, that is what is needed for a three-channel *input* pixel, and it is no guarantee of accurate color. >So >Foveon's full color photosites do correspond to pixels, but Bayer's >don't, until each color exposure is interpolated up to the size of the >interpolated recorded image. Bayer cameras record upsclaled images, as >George likes to say. No, they do not, as they have unique spatial information for each of the advertised pixels. -- <>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<> John P Sheehy <JPS(a)no.komm> ><<> <>>< <>>< ><<> <>>< ><<> ><<> <>><
From: mark on 7 Jul 2005 07:52 JPS(a)no.komm schreef: > In message <1118723676.491758.184380(a)o13g2000cwo.googlegroups.com>, > george_preddy(a)yahoo.com wrote: > > >13.72MP isn't that common. The Sigmas output 4536 x 3024 pixels images. > > The Sigma software does; the camera images are only 3.43MP. The same way debayer software creates 6MP images off the rebel sensor... > -- > > <>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<> > John P Sheehy <JPS(a)no.komm> > ><<> <>>< <>>< ><<> <>>< ><<> ><<> <>><
From: JPS on 7 Jul 2005 18:02
In message <1120737139.692894.77290(a)g43g2000cwa.googlegroups.com>, "mark" <mspan(a)mad.scientist.com> wrote: >JPS(a)no.komm schreef: >> In message <1118723676.491758.184380(a)o13g2000cwo.googlegroups.com>, >> george_preddy(a)yahoo.com wrote: >> >> >13.72MP isn't that common. The Sigmas output 4536 x 3024 pixels images. >> >> The Sigma software does; the camera images are only 3.43MP. > >The same way debayer software creates 6MP images off the rebel >sensor... Not at all; unlike the 13.7MP output from the Sigma software, 6MP Rebels actually take a measurement at 6M locations on the sensor. The 13.7MP Sigma images are merely aliased 3.43MP images softened in upsampling; the stair-steps get a plush carpet. -- <>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<> John P Sheehy <JPS(a)no.komm> ><<> <>>< <>>< ><<> <>>< ><<> ><<> <>>< |