Archived posting to the Leica Users Group, 2014/05/05
[Author Prev] [Author Next] [Thread Prev] [Thread Next] [Author Index] [Topic Index] [Home] [Search]You're assuming scaleability of the lenses without scaling the aberrations. That doesn't happen, and that's why lenses that cover larger formats don't resolve as well as smaller lenses. There are lenses that, especially at optimum apertures, can resolve much finer detail than the sensors used with them can make use of, but even with the 36mp of the A7R or D800 we're starting to push the boundaries. At 350mp (3.1x the resolution) almost all lenses out there now will be inadequate. Our best chance at super resolution is for stitched arrays, say something like an iPad with 32x45 lenses on the back, and then have a computer assemble them automatically and make the perspective corrections. That is, if we really need 350mp. Henning On 2014-05-04, at 9:26 PM, Howard Ritter <hlritter at bex.net> wrote: > OK, too much time now that I?m fully retired as of 10 days ago. As I > previously posted here and in the Gallery, I was impressed by the > improvement in resolution with the NEX7?s 24 Mpx APS-C sensor over the > D800?s 36MPx FF sensor (and much moreso the M typ 240?s FF sensor) when > all were used with 35mm lenses (actual zoom FL, not "35mm equivalent?, in > the case of the NEX7). The use of a true 35mm FL with the NEX7 was the > equivalent, in image scale on the sensor, of testing a FF sensor of 54 > Mpx. Clearly the limits of the inherent resolution of camera lenses are > far beyond the resolving power of sensors up to 36 Mpx. How far, I > wondered. > > So I shot the same scene with a Sony RX100 (20 MPx in a sensor that would > be about 1/8th the surface area of FF if it were cut down from 4:3 to 3:2 > and from 20 to 18 Mpx) and a Panasonic TS3 (12 Mpx, and about 1/32nd of FF > if cut down to 3:2 and 10.7 Mpx). These are therefore like small sections > of FF sensors of roughly 144 Mpx and 350Mpx, respectively. I shot > comparison images as well as with a D800. I made images with the RX100 > with the zoom set to actual FLs of 24mm and 35mm, and with the TS3 set to > 23mm (its maximum actual FL), as confirmed in the EXIF. The D800 was used > with 24 and 35mm primes. The results are posted at: > > http://gallery.leica-users.org/v/hlritter/Res+Test+Crops/ > > I was not surprised by the improvement from the NEX7?s 54 Mpx equivalent > to the RX100?s 144 Mpx equivalent, but to see a further increase in detail > in going to the TS3?s ~350 Mpx equivalent was really eye-opening. There?s > not as much more detail in this last step as the one before, but the true > case is masked by a lot of what looks like heavy JPEG artifact (no raw > option) in the TS3?s image. This is puzzling, as the degree of compression > here is quite mild (12 Mpx down to 5) and actually less than with the > RX100. There may be more noise as well, inevitable with smaller pixels. > There are numerous trade-offs with tiny pixels, but I would expect > technological improvements to alleviate some of these. When one of the > advantages of medium and larger formats is higher sensor resolution and > rendition of detail, it seems to me that when the FF sensor pixel count > gets into the range above 100 Mpx (assuming that will happen), we?ll have > resolution and detail that surpass MF. Noise and dynamic range will suffer > as the pixels shrink, but for many users I suspect that stellar resolution > will trump the disadvantages, at least if technology keeps these to about > the level they have now. > > Regardless, I think this is a revealing demonstration of how the > resolution of even the highest-Mpx sensor still falls far short of lens > resolution. What will be done with/about this state of affairs will be > interesting to see. > > ?howard > > _______________________________________________ > Leica Users Group. > See http://leica-users.org/mailman/listinfo/lug for more information > Henning Wulff henningw at archiphoto.com