Very high resolution lens?
I've made two tests using one of my recently acquired cameras (Fujica ST801 with EBC Fujinon 1.8/55) on the resolving power of the lens. For the first test I used a moderate resolution film (Fomapan 200) and the results showed actually the resolution of the film (110 lp/mm) and not lens.
Today I made new tests using Delta 100 developed in Rodinal 1:25. For the illumination of the test chart I used a 1000W halogen lamp, so contrast was high.
These are my results (lp/mm)
Aperture: centre, very close to edge, edge
f1.8: 85, 57, under 50
f2.8: 156, 69, under 50
f4: 161, 89, under 50
f5.6: 147, 101, 55
f8: 115, 101, 69
f11: 69, 55, 69
f16: 69, 69, 69
Do you think I was doing something wrong during this test or it is actually possible to achieve such a high resolution with this setup? The lens is a 'Planar' type design.The prints from this camera do look very sharp (to me), even heavily cropped ones taken at f1.8. I'm somewhat surprised. :rolleyes:
It also is somewhat interesting for me (I used a Tessar lens for a long time), that diffraction already takes its toll at f5.6-f8, most notably at f11.
Don't know, I never measured resolution that high, but I don't use acutance developers either. I find the value for f/11 peculiar. It does not fit well between the values of f/8 and f/16.
Why do you think a 1000W lamps creates a high contrast? Higher than what?
I haven't done the math and so it's always dangerous to prejudge, but hey, I'll go ahead and do so anyway. I doubt the falloff between f/4 and f5.6 is due to diffraction. I would guess it's more likely to be something like a focus shift on stopping down or some other similar conditions due to lens correction across the aperture range. Between f/11 and f/16 I could believe diffraction could be more significant.
The figures look to be about twice what one would expect. How are you counting lp/mm?
For the test I was using the ISO 12233 test chart as it is easier for me to work with it than the USAF 1951 chart.
On the test chart the width of the lines simply are constantly getting lower and I take the value (1-2-3-4 in my case as the distance between the camera and the chart was fairly high (2.5-2.8m) to minimize the my printer's resolution limit (1200 dpi) and that of the film causing trouble) where I still can count all of the lines (5 black and 4 white), then I multiply this number with 100 and devide it by the height of the test chart on my negative. No need for camera-test chart distance measurments and this way I think it is also more accurate.
I put the negative in my enlarger (I don't have my microscope with me right now) and used my grain focuser for focusing and measurements.
The englarger was an acceptably aligned Durst M605 with a Schneider Componon 50mm lens (wide open). The grain was fairly crisp, so I believe there were no big issues regarding the sharpness of the lens.
The whole projected negative was exactly 32cm high using the heighest column height (13.3x magnification) and in this position was the negative measured. Assuming the negative was 24mm high, the height for the chart was calculated to be 2.18mm.
I checked it a few times just to be sure I wasn't too optimistic and my eyes didn't try to see lines where it would be only one blurry thick line and I arrived at practically the same numbers both of the times.
Could it be the common confusion of lines/mm and line pairs/mm?
I think strong lights create higher contrast than just one candle, for example. If we think about 100% black (hypothetical) and a silvery surface, the black will never radiate back any light, no matter how strong it is, on the other hand the silvery surface will reflect more and more as we use consequently increase light. Hence the contrast will get higher. And if I understand it well, resolution also depends on contrast. Maybe I'm wrong at this point, but still, it was much easier to focus on the test target this way, so it had at least one benefit.
Some one also mentioned in an early post, that under practical circumstances one should hope for approximately half of the resolution values measured in a lab. If this is true, many factors can contribute to this, improper focusing or camera shake, but I think also contrast, since the contrast between two points close to each other in a real scene is often lower than the contrast present on a test chart.
Sponsored Ad. (Subscribers to APUG have the option to remove this ad.)
One more thing, just if some one might be interested: the same test with an uncoated filter attached to the lens resulted in a loss of resolution of 20-25% magnitude. Some filters could be that bad.
At f5.6 on a 35mm lens the size of the Airy disc should be about 7.5 microns which if I'm correct would lead to 133 line/mm or 66 lp/mm. This means the 147 figure Aron listed seems high to me any way you figure it.
Originally Posted by Nicholas Lindan
Correction: To be more accurate, the "diameter" of the Airy disc would be 7.5 microns, which would make the radius half of that or 3.75 microns. Furthermore, lens resolution calculations typically assumes that you can resolve two points separated by the Airy radius and not the diameter as I calculated above. This would lead to a resolution of closer to 266 lp/mm at f/5.6 in light at 550 nm.
Last edited by Denis K; 09-10-2009 at 12:59 PM. Click to view previous post history.
Reason: Stupid mistake
Just a stray comment, since I don't know the specifics of this test you did: in my experience the Fuji EBC lenses are veeery contrasty. I don't mean that as a criticism- for landscape and such they never disappointed me.
thank you for your helpful comment. It is not very easy to clearly see where the resolution limit is for a given aperture (without a device contstructed for the task, I can't get rid of the human factor at evaluation) on this chart, especially that despite the kind of film I used, grain still proved to be a major problem. I tried to be pessimistic and I got a number of 130. This is below the theoretical top resolving power at the mentioned f5.6.
However, for the other apertures I examined again (1.8-2.8-4) the earlier results seemed to be correct.
May I conclude that the simple mistake was made of taking the results as lp/mm and not as lines/mm? If yes, are these results still very good?
I have nearly 20 years experience in such tests, so may be I can help you.
That is not automatically the case. It depends on the surface and the reflection characteristics of the surface of the test chart.
Originally Posted by Aron
If it is matte and little reflecting, or brillant and strong reflecting. In most cases not the power of the light, but the direction of the light has influence. In this case, when your halogen lamp is working in a specific angle, it may happen that even the black lines are a reflecting and turning to grey, the object contrast is significantly reduced.
A proper, even and indirect lighting with no reflection spots of the test chart is necessary for a reliable test.
In general it is possible, yes. With an excellent lens, modern films, exact focusing and avoiding of camera shake extremely high resolution values are possible.
Originally Posted by Aron
You may have a look at the scientific tests made at Carl Zeiss in Germany.
They are published in the Zeiss camera lens news 17, 19, 20, 24 and 30 (look on their home page www.zeiss.de).
For example they achieved 180 linepairs per millimeter with Kodak TMX, 170 linepairs per millimeter with Fuji Velvia, and 160 lp/mm with Acros 100. With Agfa Ortho 25 250 lp/mm. At an object contrast of 1:100.
In an MTF test Dr. Hubert Nasse, chief optic designer of Zeiss, got 180 lp/mm with TMX.
Or look here, Delta 100 test:
My own test results:
I am working with a test chart with quite low contrast, only 1:16, four stops.
Because in nearly all scenes you have at least such a contrast, often higher.
I want to know how good my lenses and films are in low / moderate contrast situations.
With a Zeiss ZF 2/50 at aperture 4 and 5,6 in the centre of the negative I achieved 130 - 150 linepairs per millimeter with Ilford Delta 100.
130 lp/mm are clearly seperated, with 150 lp/mm you can still see a contrast difference. Delta 100 is one of my favourite films since its introduction 1992.
For these tests a developer test was integrated (Spur HRX-3, Spur SD2525, Rollei RLS / CG512, Rodinal 1:50; highest resolution values were achieved with HRX-3 and RLS).
With Kodak TMX and Rollei Retro 80S I got nearly the same resolution, the differences between these three films concerning resolution are so small that they are not really worth to be mentioned. Resolution of Acros 100 is a bit lower, but you can see the difference only under a microscope.
The difference in grain is a bit bigger: Retro 80S has the finest grain, then Acros 100, TMX and Delta are following.
With Rollei Ortho 25 / Adox Ortho 25 (a film made by Filmotec) I got resolution values of about 150 - 170 linepairs per millimeter, with Rollei ATP and Spur DSX ( = Agfa Copex Rapid) 170 - 200 lp/mm.
And with Spur Orthopan UR / Adox CMS 20 (older version of Agfa Copex HDP microfilm) I reached the limits of my microscope: 270 Lp/mm. Probably the resolution is even a bit higher, but my microscope delivers not enough magnification to further analyse it .
These high resolution films are incredible.