Feels good to me. Especially the fact that with normal resolution lenses, the reduction in sensor resolution caused by going from APS to FF doesnt have a big enuff effect on system resolution lpmm to swamp the huge advantage in recording area of FF vs. APS, 2.25 times as much area. Thats alot....FF beats APS even without increasing the number of sensor pixels.... ---------------------------------------------------------------------------- J.C. O'Connell mailto:[EMAIL PROTECTED] http://jcoconnell.com ----------------------------------------------------------------------------
-----Original Message----- From: keller.schaefer [mailto:[EMAIL PROTECTED] Sent: Monday, January 26, 2004 6:09 AM To: [EMAIL PROTECTED] Subject: RE: D: SMC 1.4/50mm FA / APS vs FF I used a simple test chart with lines - nothing serious - and I am not claiming the 40 lp/mm being more accurate than +/- 5. But still better than the 25 lp/mm that could be expected according to your table. I don't know better off hand, but your theory does not feel right to me. Sven Zitat von "J. C. O'Connell" <[EMAIL PROTECTED]>: > Because sensor resolution and lens resolution interact and > the result is NOT the minimum of the two, the result is > (sensor-res*lens-res)/(sensor-res+lens-res) from what I have > read on the subject. i.e. if sensor and lens are both 60 lpmm > then system total result is only 30 lpmm. In order to get a system total > resolution very close to the sensor resolution, the lens resolution > has to be very high like 10 to 100 times higher, not just a little higher. > > How are you measuring to get your results? Curious. > JCO > > -------------------------------------------------------------------------- -- > J.C. O'Connell mailto:[EMAIL PROTECTED] http://jcoconnell.com > -------------------------------------------------------------------------- -- > > -----Original Message----- > From: keller.schaefer [mailto:[EMAIL PROTECTED] > Sent: Monday, January 26, 2004 4:10 AM > To: [EMAIL PROTECTED] > Subject: RE: D: SMC 1.4/50mm FA / APS vs FF > > > JCO, > > I am puzzled. In my own humble attempts at testing *ist D performance I > achieved > a maximum resolution of around 40 lp/mm, which according to your figures > would > only be possible with a 110 lp/mm lens - a lens which I do not have (I > think). > > It is clear that lenses need more resolution as the image capturing area > gets > smaller. But I always assumed that as long as the lens resolution exceeded > the > (real life) sensor resolution, I would be in the green. This also > corresponds > to test shots taken with say a 1.4/50 where you can see the image improving, > as > you gradually stop down from f=1.4 to f=4. > Between f=4 and f=11 I am unable to see any difference in image quality (16 > and > 22 then get worse again). To me this says that between 4 and 11 the lens > resolution exceeds the sensor resolution. > > Why are you assuming that the effective resolution depends on the lens > resolution, even if it is way above the sensor resolution? > > Sven > > > > Zitat von "J. C. O'Connell" <[EMAIL PROTECTED]>: > > > Let me TRY to clarify. The "TOTAL RESOLUTION" or detail of an image is not > > determined > > by lpmm, it is determined by (lp/mm*Hdimension)*(lp/mm*Vdimension) > > This is essentially (TOTAL horizontal lines * TOTAL vertical lines) > > recorded. > > > > Now assuming you have two Better than Perfect lenses of say infinite lp/mm > > attached. > > The TOTAL RESOLUTION of a full frame or APS 6Mpixel sensor would be the > same > > assuming the correct focal lengh factor is used (1.5?) for the different > > sensor > > sizes. > > > > BUT, we dont have perfect lenses. I used Excel to calculate the > difference > > in TOTAL RESOLUTION which I call Megalines squared below. Since infinite > > resolution would crash the program, I used an imaginary lens with 10,000 > > lp/mm > > as a reference. As you can see by the data below, with a perfect lens, > > there > > would be nearly zero difference in TOTAL RESOLUTION using Full Frame vs. > APS > > sized 6Mpixel sensors. BUT with real lenses in the 30 to 100 lp/mm range, > > a FullFrame 6Mp sensor would yield an image with 40% to 25% more detail. > vs > > an > > APS 6Mp sensor. This is signifigant and not only that, each pixel would > > be 125% larger in area yielding less noise and/or more sensitivity. > > Bottom line is a full frame sensor will perform better than an APS sensor > > using real lenses if they both have the same Mpixels. Even using the best > > 100 lp/mm lenses, FF is going to be 25% sharper, and FF will be even more > > sharp as the lens resolution falls lower to more typical levels. Seems > > paradoxical > > but thats the geometry. Comments welcome, spreadsheet available. BTW > > is *istD effective sensor 16X24mm as I assumed in these calculations which > > I based on the 1.5 crop factor and same aspect ratio??? > > JCO > > > > FULL FRAME, 6 Mpixel sensor, 2000X3000 pixels, 24mmX36mm > > --------------------------------------------------------- > > lens sensortotal total resolution Mlines^2 > > lp/mm lp/mm lp/mm (lp/mm)*24mm*4*(lp/mm)*36mm*4 > > 10000 41.6 41.4 5.931 > > 200 41.6 34.4 4.099 > > 190 41.6 34.1 4.025 > > 180 41.6 33.8 3.946 > > 170 41.6 33.4 3.860 > > 160 41.6 33.0 3.767 > > 150 41.6 32.6 3.666 > > 140 41.6 32.1 3.555 > > 130 41.6 31.5 3.433 > > 120 41.6 30.9 3.298 > > 110 41.6 30.2 3.149 > > 100 41.6 29.4 2.983 > > 90 41.6 28.4 2.797 > > 80 41.6 27.4 2.589 > > 70 41.6 26.1 2.353 > > 60 41.6 24.6 2.086 > > 50 41.6 22.7 1.782 > > 40 41.6 20.4 1.437 > > 30 41.6 17.4 1.050 > > 20 41.6 13.5 0.630 > > 10 41.6 8.1 0.225 > > > > > > APS, 6 Mpixel sensor, 2000X3000 pixels, 16mmX24mm > > ---------------------------------------------------------------------- > > lens sensortotal total resolution Mlines^2 TOTAL Resolution Loss > > vs. > > lp/mm lp/mm lp/mm (lp/mm)*16mm*4*(lp/mm)*24mm*4 Full Frame (in %) > > 10000 62.5 62.1 5.926 -0.1 > > 200 62.5 47.6 3.483 -15.0 > > 190 62.5 47.0 3.397 -15.6 > > 180 62.5 46.4 3.306 -16.2 > > 170 62.5 45.7 3.208 -16.9 > > 160 62.5 44.9 3.103 -17.6 > > 150 62.5 44.1 2.990 -18.4 > > 140 62.5 43.2 2.868 -19.3 > > 130 62.5 42.2 2.736 -20.3 > > 120 62.5 41.1 2.594 -21.3 > > 110 62.5 39.9 2.440 -22.5 > > 100 62.5 38.5 2.272 -23.8 > > 90 62.5 36.9 2.090 -25.3 > > 80 62.5 35.1 1.891 -26.9 > > 70 62.5 33.0 1.675 -28.8 > > 60 62.5 30.6 1.439 -31.0 > > 50 62.5 27.8 1.185 -33.5 > > 40 62.5 24.4 0.914 -36.4 > > 30 62.5 20.3 0.631 -39.9 > > 20 62.5 15.2 0.353 -44.1 > > 10 62.5 8.6 0.114 -49.2 > > -------------------------------------------------------------------------- > -- > > J.C. O'Connell mailto:[EMAIL PROTECTED] http://jcoconnell.com > > -------------------------------------------------------------------------- > -- > > > >
