12-01-2010, 01:35 PM
[quote name='Klaus' timestamp='1291204714' post='4643']
*sigh*
I do well believe Falk's mathematical analysis that you can recover the maximum of details via sharpening but sharpening itself is lossy in terms of increased noise and sharpening artifacts - it will cost you probably something like one ISO step in terms of noise to do so just to point to the most obvious effect. Don't nail me on "one" - it could be less or (more likely) more - I haven't measured this.
The sharpening itself may not affect the validity of the MTF tests so I could leave the K5 as is - this is correct. However, it does diminish the field quality of the K5 -> for me<-. There's a use for the K5 beyond the sheer testing. And ... as discussed .... it is, of course, a matter of taste. :-) If you prefer to have a sharp output at ISO 400 with an equivalent ISO effect like at ISO 800 (after sharpening), well fine - just go on - but I prefer to have a similar sharp output at ISO 400 with an equivalent ISO effect like at ISO 200 (because I need to apply less sharpening). I reckon the trade-offs of either approach are well understood. As mentioned I would prefer to have a weak AA filter rather than none but this option is not available out there.
Regarding the tests - it is quite meaningless whether there's an AA filter in the camera or not. This will NOT CHANGE THE RANKING of the lenses nor the technical quality of a lens and THIS IS THE ULTIMATE ESSENCE of it all.
Readers do already put too much emphasis on the max. LW/PH values because they try to cross compare systems although we do always stress that this is invalid - here and elsewhere. In the very theory I could even use straight JPEGs from a factory K5 resulting in max. LW/PHs around 2000. It would be an obvious approach and also a technically valid one - finally this is "recommended by Pentax" (it wouldn't be the default otherwise, wouldn't it ?). I reckon there would be an outcry in the community if I did so though. So as always in life - the truth is always gray and neither black nor white.
[/quote]
You can avoid sharpening most noise (as noise, unless "filtered", appears as coloured specs), when you apply sharpening.
In Photoshop (a usual suspect for sharpening), go from RGB to Lab-mode. Select the lightness layer and apply sharpening only on that layer. The coloured specs will not be as affected by the sharpening as you would see in RGB mode.
*sigh*
I do well believe Falk's mathematical analysis that you can recover the maximum of details via sharpening but sharpening itself is lossy in terms of increased noise and sharpening artifacts - it will cost you probably something like one ISO step in terms of noise to do so just to point to the most obvious effect. Don't nail me on "one" - it could be less or (more likely) more - I haven't measured this.
The sharpening itself may not affect the validity of the MTF tests so I could leave the K5 as is - this is correct. However, it does diminish the field quality of the K5 -> for me<-. There's a use for the K5 beyond the sheer testing. And ... as discussed .... it is, of course, a matter of taste. :-) If you prefer to have a sharp output at ISO 400 with an equivalent ISO effect like at ISO 800 (after sharpening), well fine - just go on - but I prefer to have a similar sharp output at ISO 400 with an equivalent ISO effect like at ISO 200 (because I need to apply less sharpening). I reckon the trade-offs of either approach are well understood. As mentioned I would prefer to have a weak AA filter rather than none but this option is not available out there.
Regarding the tests - it is quite meaningless whether there's an AA filter in the camera or not. This will NOT CHANGE THE RANKING of the lenses nor the technical quality of a lens and THIS IS THE ULTIMATE ESSENCE of it all.
Readers do already put too much emphasis on the max. LW/PH values because they try to cross compare systems although we do always stress that this is invalid - here and elsewhere. In the very theory I could even use straight JPEGs from a factory K5 resulting in max. LW/PHs around 2000. It would be an obvious approach and also a technically valid one - finally this is "recommended by Pentax" (it wouldn't be the default otherwise, wouldn't it ?). I reckon there would be an outcry in the community if I did so though. So as always in life - the truth is always gray and neither black nor white.
[/quote]
You can avoid sharpening most noise (as noise, unless "filtered", appears as coloured specs), when you apply sharpening.
In Photoshop (a usual suspect for sharpening), go from RGB to Lab-mode. Select the lightness layer and apply sharpening only on that layer. The coloured specs will not be as affected by the sharpening as you would see in RGB mode.