Real scanning details

[Les Sarile]

I was immersed in manual focus cameras and picked up a few lenses off local CL for cheap. SO cheap I wasn't really sure if they were even worthwhile. That and the fact that I was seeing a lot of these "old lenses are not even good enough for the new sensors" posts. One of those lenses I got was the SMC-Pentax-M-50mm-F4-Macro. After a few rolls of film, I started to notice that I was getting very good results with it - not lacking for detail and edge sharpness handheld such as this one on Fuji RVP (Velvia).

Full res -> Fuji RVP (Velvia)

So I got curious to find out just how good it was.
I did a quick set of tests using my Pentax K20D and it didn't appear to standout compared to my other Pentax lenses.

Full res -> Pentax lenses
I also noticed the highest they could achieve was all about the same. Turns out they were achieving the max resolution that the Pentax 14.6MP sensor could achieve.

So I setup a 4 X 4 arrangement of the 12233 rescharts and shot some Kodak Techpan @ ISO25 and processed it in Kodak Technidol since that is the highest resolving film I have available. I then used the center area to compare results with a 14.6MP Pentax K20D+autobellows, 36.3MP Nikon D800+autobellows and scans from a Coolscan at 4000dpi shown below.

Full res -> Kodak Techpan scan compare
Clearly the 4000dpi scans from this frame of film far exceed the K20Dsensor and is almost the equivalent with the D800 although the D800 applies more pixels.

I then used my K20D+autobellows+50mm macro to optically enlarge the central target area (far right) and it is plain to see that there is far more detail captured on the film then can be resolved with these tools.

Of course this means that my cheaply bought SMC-Pentax-M-50mm-F4-Macro will not be the reason I am unable to achieve considerable detail and likely the scanning method, the film or sensor used or how the shot was taken will be the limiting factors. Now this has me wondering if an original factory new version would be even better?

 

[mdarnton]

Essentially, it is BS to make the blanket statement that old lenses are not good enough for digital. I would mark the people who said that off my list of possible consultants.

 

[Les Sarile]

Essentially, it is BS to make the blanket statement that old lenses are not good enough for digital. I would mark the people who said that off my list of possible consultants.

Yup, BS = blanket statements . . .

 

[Mick Fagan]

Not being that into electronic capture, it took me a while to understand what you were saying. From what I understand, your Pentax system is using lenses designed for film imagery with the capture field being 24mm x 36mm and the light hitting the film at ever widening angles the further away film is from the centre of the lens.

Your electronic capture is smaller, therefore you are using the more straight part of the lenses capabilities; thereby ameliorating what could be perceived as a problem. In this case, there is not a problem, perhaps the reverse; interesting experiment.

Do you think you would you have the same results (or similar) if your sensor size was at the edge of the image spread; that is, 24mm x 36mm?

As I understand it, lenses designed for electronic capture are designed with as straight a line of the imagery to the sensor as possible, which, as a side effect, could be a slight advantage for film capture. By that I mean, edge fall off should be less. An extreme example of edge fall off is when I use a 65mm lens on a 4x5" film camera without a centre filter.

I do like your chart, quite interesting.

I like the capability of the f/1.7 and f/2.0 lenses; nice.

Mick.

 

[Les Sarile]

I tested my Nikon lenses on the D800 with the chart and they are achieving the max res the sensor is capable of. I am assuming that since the detail captured on Techpan far far exceeds what the D800 can achieve I can correlate that with my Pentax lenses on a FF body. It is however an assumption that I intend to put to the test now that Pentax does have a FF but alas it is is only similar to the D800.

I got the files for the 12233 at http://www.graphics.cornell.edu/~westin/misc/res-chart.html if you're interested.

I have tested my other MF prime lenses in like manner and I've only found the mirror lens to be poor performing in this respect.

 

[jim10219]

It’s mainly the old zoom lenses that don’t hold up so well in this day and age (with some exceptions). The technology to design them has advanced considerable with the use of powerful computers. With prime lenses, those assumptions do not apply. Prime lenses are much easier to calculate because you’re only correcting for aberrations at one focal length. This doesn’t take a modern super computer to do in a reasonable time scale. On a zoom, you’re having to correct multiple aberrations across an infinite number of possible focal lengths between two limits. And whatever decisions you make to correct one aberration at one focal length will cause compromises at every other. So the goal isn’t perfection, but balance. For those type of calculations, you’re going to need some serious computing power to get it done in your lifetime.

In fact, it can be argued that some prime lenses from the film era out perform modern primes. This is often due to them being able to use materials in the glass, like thorium, lanthanum (both radioactive), and lead. In today’s world, most designers avoid such additives due to the difficulty in manufacturing (safely) and waste disposal. Thus they have to figure out ways to make do with sometimes inferior materials. Of course, there are still many top quality primes being made today, and quite a few of them can out perform most of the older manual focus primes from the film era. But they make a lot of junk these days, just like they made a lot of junk back then.

In any event, lens technologies and their marks in performance tests don’t mean much to most serious professional photographers. That’s the realm of amateurs and scientists (who rely on precision, not artistry). One of my favorite lenses that I use often is a no-name lens from the 1880’s. I use about 5 other lenses that are over 100 years old and frequently use about 20 more that are over 50 years old. Even a some of my newest lenses are just autofocus, auto aperture, metering, multi-coated versions of designs that date back more than 50 years. The glass is the same.

 

[mdarnton]

Mick, I suspect the problem of tangential approach to the sensor is more of an issue with RF wide angles that aren't retrofocus, progressively less so with SLR teles. I do know that the widest Voigtlander RF lenses have been reformulated for digital because of this problem. . . . but not the longer ones.

I'm understanding you don't know exactly why this is? I think I have this part right: It's because digital sensors are thick, in layers, with a tiny lens at the top of each pixel. So if the light comes in at an angle, it's not taking the easy path straight down the chute. There's a cutaway drawing of a pixel here: https://www.ephotozine.com/article/digital-camera-image-sensor-technology-guide-16808

Side issue, in the vein of light coming in straight to the sensor, vs from the side, just thought of this: Am I the only one who's blown away by the idea that no matter how complex a lens is, that one single ray of light coming right through the center ideally travels through only parallel layers of glass, and doesn't get bent a bit all the way from the subject to the film? Cool.

 

[Les Sarile]

That’s the realm of . . . scientists (who rely on precision, not artistry).

I look at it from the perspective that the designers made the cameras and lenses accurate so that there would be no technical impediment to one's art. That way the artists - I use the term very loosely, can feel free to achieve their artistic expression.

 

[Mick Fagan]

M. Darnton, thank you for that link, quite interesting and reasonably informative. I am coming from little knowledge of electronic capture, learning bit by bit.

As for light passing directly through glass, I'm not so sure. As I understand it, 100% of light presented to glass comes out the other side with less than 100% of what went in.

Mick.

 

[Mick Fagan]

Les, thanks for the link to that chart.

My 600mm mirror lens is also a poor performer, but, and it is a big but, it allows me to get results I otherwise would not be able to get.

Mick.

 

[Les Sarile]

Les, thanks for the link to that chart.

My 600mm mirror lens is also a poor performer, but, and it is a big but, it allows me to get results I otherwise would not be able to get.

Mick.

Yeah who wouldn't want a Canon 1200mm f5.6 except for the weight the price!

 

[WilmarcoImaging]

Les, well done. This is the most concise and data-supported article I have seen on this topic, or similar topics.

There are a large number of people using film-era Minolta/Sony A-mount lenses on modern digital cameras with highly satisfactory results.

 

[hk1]

I then used my K20D+autobellows+50mm macro to optically enlarge the central target area (far right) and it is plain to see that there is far more detail captured on the film then can be resolved with these tools.

So are you saying Coolscan/Nikon D800 don't resolve what's on film? Interesting.
And when I look at the film you resolve as much as the K20D with the lens(es) in question, don't you?

 

[Les Sarile]

So are you saying Coolscan/Nikon D800 don't resolve what's on film? Interesting.
And when I look at the film you resolve as much as the K20D with the lens(es) in question, don't you?

I am specifically showing that 1:1, none of these method I used can fully resolve Techan film @Iso 25 processed in Technidol. You can see from the optical enlargement of that area on the film the details it failed to reach.

 

[gorbas]

Well done Les!
I'm thinking about doing similar test myself for 35mm and 6x6 negatives to compare NikonScan 8000 scans and DSLR scanning files.
I have a few questions for you:
Did you buy test targets or you printed them? On laser printer? How big did you print targets?
How did you find right distance to photograph targets from, regarding of the focal length of the lens?
Most likely I will have more questions but those 2 will be good for start.
Thank you in advance
Goran

 

[Les Sarile]

Well done Les!
I'm thinking about doing similar test myself for 35mm and 6x6 negatives to compare NikonScan 8000 scans and DSLR scanning files.
I have a few questions for you:
Did you buy test targets or you printed them? On laser printer? How big did you print targets?
How did you find right distance to photograph targets from, regarding of the focal length of the lens?
Most likely I will have more questions but those 2 will be good for start.
Thank you in advance
Goran

I got the files for the 12233 at http://www.graphics.cornell.edu/~westin/misc/res-chart.html which I printed on tabloid paper using the highest res on a laser printer. Distance will depend on focal length to fill the vertical dimensions of the film. You might have to compensare since most viewfinders don't offer 100% coverage. Of course I provided ample lighting to get the fastest shutter speed I could reach while changing aperture. Tripod, mirror lockup, remote/timer and viewfinder magnification helps.

 

[gorbas]

Thank you Les! I have seen link you provided for the target. I will use just 50mm lens on 35mm camera. My goal is not to test the lens but to create test negative for the scanner and DSLR scanning rig. Lighting is no issue for me, I have full strobe kit + continuous light sources.
Versalab parallel laser will be used to square up the camera and wall with targets. Don't have TechPan in 35 but have APX-25, Fuji Neopan F, Panatomic X and Pan F. Have some Techpan in 120.
My only issue is what size to print targets and how far camera need to be away?
Will keep you posted

 

[Les Sarile]

I also conducted this testing using Fuji RVP - original Velvia 50, as well as Kodak TMAX 100 but have yet to review the results. I wanted to correlate/compare the results with available film.

Good luck!

 

[hk1]

I am specifically showing that 1:1, none of these method I used can fully resolve Techan film @Iso 25 processed in Technidol. You can see from the optical enlargement of that area on the film the details it failed to reach.

But if I interpret your 1:1 macro slightly differently then the Techpan film gives the same resolution as the K20D with one of those Pentax lenses attached (your second figure). At least the red arrows are at a similar position. I find this quite interesting - you need an extremely sharp film to get close to the resolution of a rather average digital camera with its 14MP.

 

[Les Sarile]

But if I interpret your 1:1 macro slightly differently then the Techpan film gives the same resolution as the K20D with one of those Pentax lenses attached (your second figure). At least the red arrows are at a similar position. I find this quite interesting - you need an extremely sharp film to get close to the resolution of a rather average digital camera with its 14MP.

The second picture I posted - my Pentax lenses, was done with a 2 X 2 targets as shown on the bottom left of it while the Techpan was 4 X 4 arrangement as shown on the bottom left. I hope that clears that up.

Techpan is not only super sharp but very high resolving but so are Fuji Velvia and TMAX 100 which I will post when I get a chance. Since you also have Coolscan, you will know this yourself.

BTW, my initial evaluation of Velvia and TMAX indicates they far exceed my 14.6MP K20D also. Of course this is not surprising to me since I've owned the Coolscan 5000 since it was first released.