Forgot Password
Pentax Camera Forums Home
 

Reply
Show Printable Version Search this Thread
07-20-2012, 01:11 PM   #1
Site Supporter
Site Supporter




Join Date: Aug 2009
Posts: 385
lens resolution

I just performed some testing of my lenses:

Lens M 28 f2.8 M 28 f3.5 M 35 f2.8 M 50 f4 M 50 f1.7 M 100 f2.8 M 135 f3.5
film lpm 53 84 75 74 74 59 55

I used a tripod and shutter speeds between 1/250 and 1/500, and a proper test chart, and then inspected under an enlarger at 8x enlargement, with a 25x grain focusing scope. The difference between the 28mm 2.8 and the 3.5 was dramatic. The other surprise was that the 50/4 was no different to the 50/1.7. All shot at f8, with hoods, and at 3m for the lenses to 50mm, and at 6m for the 100/2.8 and 135/3.5. The film was tmax 400 developed in xtol 1:1.

Will repeat the same test again soon, including a tmrn sp 90 2.5.


Last edited by whojammyflip; 07-20-2012 at 02:39 PM. Reason: stupid advert links being added into my post :(
07-20-2012, 02:43 PM   #2
Pentaxian
johnyates's Avatar

Join Date: Aug 2011
Location: Saskatoon, SK
Photos: Albums
Posts: 1,257
QuoteOriginally posted by whojammyflip Quote
The difference between the 28mm 2.8 and the 3.5 was dramatic.
So, which has the better resolution?
07-20-2012, 03:19 PM   #3
Pentaxian




Join Date: May 2011
Location: Santa Cruz, CA
Photos: Albums
Posts: 2,348
Is the 50/4 the M 50mm F4 Macro?

I did the test with my 50mm lenses and got these results.

Link to larger version -> http://www.fototime.com/AFA3C6EDB6A663D/orig.jpg
07-20-2012, 10:38 PM   #4
Otis Memorial Pentaxian
Loyal Site Supporter
stevebrot's Avatar

Join Date: Mar 2007
Location: Vancouver (USA)
Photos: Gallery | Albums
Posts: 31,502
Those numbers look about right, though I am surprised that the 28/3.5 ace'd the field so definitively. At f/8, I would have expected the honors to go to either of the 50s. It might have done even better if you had normalized the shooting distance for all lenses to 20x the focal length (sort of a standard for lens testing) to provide equivalent target magnification.


Steve

07-23-2012, 12:02 PM   #5
Site Supporter
Site Supporter




Join Date: Aug 2009
Posts: 385
Original Poster
The 28/3.5 had the best resolution. I had just sent it off to have it cleaned, as it came through from ebay with a load of oil on the inside. However, it was in perfect external condition and had been kept in a case. The thing is, in order to maximise the resolution, you would have to be half the distance to the subject to get the same size on the film, as with a 50, so in practice, the 50 shots are going to have more detail in individual items, unless you get close up, if you see what I mean. And portraits of people with a 28 look funny.

It was difficult to classify which grouping the 28/3.5 fell into, so I popped it in the lower of the two, so really, its half way between 84lpm and 94lpm. The other factor to remember here is that its dependent on everything, which means tripod, film flatness, film etc etc. Plane of focus could also be a limiting factor. It was like night and day comparing the two 28's.
__________________
QuoteOriginally posted by LesDMess Quote
Is the 50/4 the M 50mm F4 Macro?

I did the test with my 50mm lenses and got these results.

Link to larger version -> http://www.fototime.com/AFA3C6EDB6A663D/orig.jpg

Yep, 50/4 is the macro. Maybe 100x focal length is no longer playing to the strength of this lens.

Are your target charts drum scanned? Hobby film scanners are limited to 30lpm. Digital bodies are limited to about 100lpm, if everything is lined up perfectly, due to Nyquist sampling and even then you will get aliasing. You see this in step like patterns forming in the image.

The only way to test the resolution of the lenses is to use film which can comfortably exceed the lens, as once the resolution starts knocking on 100lpm, you start to get limited by the sensor. Even 160lpm film will degrade the image from a lens capable of 100lpm, due to convolution. This is why I posted this in the film section.

Looking directly at film through an enlarger, is looking at the aerial resolution of the combination of film, taking lens, and enlarging lens. This way you can achieve something like a magnification of 200x on your retina, so 100lpm on film becomes 0.5lpm at the back of your eye.

I will repeat the test with some TMAX 100 and maybe some Rodinal to see whether the agency affects help me out. I reckon I was still getting 50% contrast in the images I was looking at, so these lpm's are not cut offs for the maximum contrast of the lens, but the last acceptable level for me. I will also use 100x, as Steve pointed out.

Last edited by whojammyflip; 07-23-2012 at 01:07 PM.
07-23-2012, 04:43 PM   #6
Pentaxian




Join Date: May 2011
Location: Santa Cruz, CA
Photos: Albums
Posts: 2,348
In that compilation of res tests it is identified as conducted with the K20D and not film at all. The test target was only 2X height & width so it is about 12X2=24. This is correlated to DPREVIEW's own test of the K20D.

Since we are in the film processing forum, I should have posted the film version scan of at least the 50mm f4 macro as shown below with the corresponding K20D for reference - but this time at 4X height & width. As you can see in this comparison, it is the K20D's 14.6MP sensor that was limiting all the lenses tested to about 6X4=24. In this test, I used Kodak Techpan shot at ISO25, processed in Technidol and scanned with the Coolscan 9000 and achieving almost 10X4=40 vertical and almost 8X4=32 horizontal. I also used the K20D to take a 4.5X enlargement of the center portion and it is evident the Coolscan's 4000dpi is now the limiting factor because this is showing almost 13X4=52 vertical and horizontal. This probably requires higher then the Imacon's 8000dpi to fully resolve. Maybe the Heidelberg Tango's 11,000dpi will do it . . .

Link to larger version -> http://www.fototime.com/A08A371F28ED137/orig.jpg

This is of course not surprising as I have previously observed that my sample of this macro lens has consistently produced particularly sharp results as in this example from Kodak Gold 100.

Link to full res -> http://www.fototime.com/AA49E1A00A2B5A7/orig.jpg
07-23-2012, 04:55 PM   #7
Pentaxian




Join Date: May 2011
Location: Santa Cruz, CA
Photos: Albums
Posts: 2,348
I hope you are using the ISO12233 test chart so we can correlate results. Of course if you have digis, you can correlate them to DPREVIEW as a sanity check. In my case, I downloaded it from http://www.graphics.cornell.edu/~westin/misc/res-chart.html and made 11 X 17 prints on my laser to maximize that printer's 1200dpi print. I hope you can appreciate the 13X4=52 vertical and horizontal results because I am certain there are not many sensors/films in a 35mm format that are capable of matching that . . .
07-23-2012, 09:42 PM   #8
Site Supporter
Site Supporter




Join Date: Aug 2009
Posts: 385
Original Poster
Edmund Scientific with the USAF categorisation of lpm. This makes the reading granular, which meant that for instance I rated the 28/3.5 as 84lpm, when it could potentially have scraped in at 94lpm. However, it also means I can give a specific resolution easily. The surface of the chart is important as it determines the modulation of the target (difference between black and white).

Any testing of lens resolution is going to be subject to the method used. The important thing is to note what the method was, at least!

[I dont want a debate on digital vs film. I did post in the film section to avoid any digital. If you can see 90lpm, then that is the system resolution. What system you use is entirely up to you!]

07-24-2012, 04:26 AM   #9
Pentaxian




Join Date: May 2011
Location: Santa Cruz, CA
Photos: Albums
Posts: 2,348
To be sure, there is no need for debate when actual results that can be correlated are available. I am presenting system wide resolution. Also, I am showing you the results of my 50mm f4 macro and as you can see, it is pretty good. I also have the results for the other lenses tested with Techpan & Technidol and I think I also have the same test conducted with TMAX 100 and XTOL.
07-24-2012, 11:48 AM   #10
Site Supporter
Site Supporter




Join Date: Aug 2009
Posts: 385
Original Poster
Just to run through the principal of convolution: its the combination of the individual component mtf's.

This in turn means the system resolution is equal to the inverse of the sqrt of the sum of the squared errors. In numbers, with a 100lpm lens, 160lpm film and 200lpm enlarging lens, you would have a combined 78lpm.

Alternatively, you can deconvolve, which means given a system 90lpm, you actually have an aerial resolution from the taking lens of 130lpm.

Given a scanner can see only 60lpm at best, you will only see a system resolution of 54lpm, even though the lens might have an aerial resolution of 130lpm. The problem becomes that the system is insensitive to the taking lens. Bumping the taking lens to 200lpm still only produces a system resolution of 57lpm.

Anyway, the main point of my post was to illustrate that for my lenses, the 28/3.5 lived up to its reputation of being substantially sharper than the 28/2.8, and the macro M 50 was not any better than the M 50/1.7 (albeit its not optimised for 60x focal length probably).

Last edited by whojammyflip; 07-24-2012 at 12:09 PM.
07-24-2012, 02:58 PM   #11
Pentaxian




Join Date: May 2011
Location: Santa Cruz, CA
Photos: Albums
Posts: 2,348
Photozone used to publish the MTF particularly of these older lenses. Is there a source for this now?
07-24-2012, 09:01 PM   #12
Site Supporter
Site Supporter




Join Date: Aug 2009
Posts: 385
Original Poster
Just had a look at the photozone site which linked through to the Imatest software.

Dont know whether you are refering to it, but it is possible to use a slanted edge, with software, on a lower spatial resolution sensor. The slanted edge effectively resamples repeatedly the image, shifting the sensor along fractions on the interpixel distance, as you move down the slanted edge. These resamples are added together and build a picture of how the contrast at the edge changes, over fractions of the sensors interpixel distance....effectively, the interpixel distance has been reduced.

The CCD can be regarded as a comb, with each tooth sampling the image. By sliding the comb (fractions of the distance between the teeth of the comb) along the image and overlapping the information from successive moves, a finer sampling frequency can be achieved. This is the principal to a slanted edge MTF reading.

This information is then used for calculating the edge spread function and line spread function. By Fourier transforms you get the MTF. The important thing is to have this gentle wedge type rotation between the sensor and the edge being imaged. As you can imagine, the process introduces noise, so some smoothing is necessary when forming the MTF curve.

This is all in the manual of photography, version 9, with a very intuitive explanation on page 411. Its not something I dreamt up. Here is another link too:
http://dougkerr.net/Pumpkin/articles/MTF_Slant_Edge.pdf

Basically, in order to use a lower resolution sensor, you must digitally process the information, and cannot look at it directly, in order to draw up a MTF curve from an edge. In turn, this means its invalid to use spatial resolution charts with a digital sensor, due to Nyquist sampling limits, aliasing etc.

But this is the film section...and you can look directly at film under a microscope, and dodge the dirac comb...

Last edited by whojammyflip; 07-24-2012 at 09:40 PM.
07-25-2012, 04:39 AM   #13
Pentaxian




Join Date: May 2011
Location: Santa Cruz, CA
Photos: Albums
Posts: 2,348
Similar to your question -> Lens rating guidelines - are there any? (https://www.pentaxforums.com/forums/pentax-slr-lens-discussion/109929-lens-ra...there-any.html) I am trying to correlate my resolution testing to the performance of the lenses I am testing since the reviews posted only provide opinions. These old manual focus lenses are bought used and I wanted to know if they are performing as intended more precisely.
07-25-2012, 02:00 PM   #14
Site Supporter
Site Supporter




Join Date: Aug 2009
Posts: 385
Original Poster
This thread is about my lens resolution.... the stark difference between 28/3.5 and 28/2.8.

Your resolution measurements will not help with your ranking due to poor resolution in digital scanner or digital sensor vs film. Using digital to view old school res charts is like looking through a net curtain to try to judge another net curtain.

You need to use the slanted edge procedure. Basically, your res charts just let you know the limits of your scanner/digital sensor, and let you know nothing about your lenses.

Here is futher information on the slanted edge procedure for digital capture:
Quick MTF: About resolution

Last edited by whojammyflip; 07-25-2012 at 02:06 PM.
07-25-2012, 08:46 PM   #15
Pentaxian




Join Date: May 2011
Location: Santa Cruz, CA
Photos: Albums
Posts: 2,348
Let me clarify what I have presented here. I printed on 11 copies of the ISO12233 test charts on 11"X17" paper using a 1200dpi laser. I then arranged them at 4 X 4 and cutout the middle section (highlighted in red) and put it on the middle.



Using this as the test target, I shot it with my 14.6MP Pentax K20D and with my LX using Kodak Techpan at ISO25 developed in Technidol with the various lenses and scanned with my Coolscan. The image below shows the 100% crops of the red highlighted area. Used perfect shooting conditions - tripods, all apertures, numerous times, etc.

Link to full size -> http://www.fototime.com/A08A371F28ED137/orig.jpg

I then used the K20D on bellows to optically magnify the red highlighted portion - applying the full resolution on just that portion. This is about 4.5X optical magnification. Like a microscope view blown-up. Perhaps this next representations is better as it shows 100% crops of the K20D capture, the LX+Techpan capture and then the K20D on bellows magnification of the red highlighted area.

Linke to full res -> http://www.fototime.com/1F7747A69031070/orig.jpg

As I said, the main purpose is to verify just how good my lenses are relative to each other since these are used old lenses. I have older magazines that show the MTF's of these lenses when they were brand new that I just need to compile. This way, when I find with these tests that my lens A is outresolving my lens B I will know that it is to be expected based on their original specification.

A residual benefit happens to be that I can correlate my results with DPREVIEW's results from their digital testing as well as others who use this test target.

My K20D+Bellows+Slide Copier assembly.
Reply

Bookmarks
  • Submit Thread to Facebook Facebook
  • Submit Thread to Twitter Twitter
  • Submit Thread to Digg Digg
Tags - Make this thread easier to find by adding keywords to it!
f2.8, f3.5, film, lens, lenses, photography, test
Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
Max resolution from a lens Vantage-Point Pentax SLR Lens Discussion 15 02-29-2012 05:28 PM
Pixel pitch and lens resolution surfotog Pentax Medium Format 4 06-09-2011 12:01 PM
Lens resolution Tim_R Pentax SLR Lens Discussion 11 08-25-2010 02:44 PM
Lens Resolution and the K-7 kickboxingpenguin Pentax SLR Lens Discussion 5 04-16-2010 04:30 AM
What wide...ish (10 - 40 mm) lens with best resolution? emr Pentax SLR Lens Discussion 6 11-12-2009 11:53 AM



All times are GMT -7. The time now is 12:15 AM. | See also: NikonForums.com, CanonForums.com part of our network of photo forums!
  • Red (Default)
  • Green
  • Gray
  • Dark
  • Dark Yellow
  • Dark Blue
  • Old Red
  • Old Green
  • Old Gray
  • Dial-Up Style
Hello! It's great to see you back on the forum! Have you considered joining the community?
register
Creating a FREE ACCOUNT takes under a minute, removes ads, and lets you post! [Dismiss]
Top