Roger Cicala tests Lomography's Petzval lens

Gerald · Joined: 25 Mar 2014 Posts: 1196 Location: Brazil

Roger Cicala has tested Lomography's lens Petzval:

http://www.lensrentals.com/blog/2014/07/a-new-old-lens

As you probably know, Joseph Petzval was a Hungarian mathematician who is credited to have been the first person to design a lens using scientific methods. The Petzval lens is basically composed of two achromatic doublets separated by a certain distance. You can think of Petzal lens as a doublet followed by a focal reducer, which reduces the focal length but increases the relative aperture (F-number) of the compound lens. The idea behind the Petzal lens was to make a "fast" lens for portraits to reduce the long exposure time required with the extremely slow emulsions of the time.

The Petzval lens allows a reasonable definition in the center of the field, but the resolution is very bad at the edges due to the strong astigmatism and other off-axis aberrations. These characteristics are acceptable for portraiture and projection. Remember also that the Petzval lens is fairly inexpensive to be produced, which explains its popularity to this day.

The measurements made by Roger Cicala on the Lomography's Petzval lens are interesting any way, and as expected for a Petzval lens, it confirms that the resolution at the edges and corners is very bad. However, the resolution in the center is also not too brilliant..

To put the Lomography's Petzval lens in perspective, I compared the Cicala's measurements with the ones I've made on some lenses I have which could be used for portraits, too.

Measurements of MTF50 in the center of field:

Lomography's Petzval 85mm F2.2
@F2.2 MTF50 = 23 lines/mm

CZJ Biometar 80mm F2.8
@F2.8 MTF50 = 27 lines/mm
@F4 MTF50 = 39 lines/mm

Helios 44M-4 58mm F2
@F2 MTF50 = 30 lines/mm
@F2.8 MTF50 = 40 lines/mm

Pentacon 50mm F1.8
@F1.8 MTF50 = 28 lines/mm
@F2.8 MTF50 = 42 lines/mm

SMC Takumar 135mm F2.5
@F2.5 MTF50 = 34 lines/mm
@F4 MTF50 = 40 lines/mm

CZJ Sonnar 200mm F2.8
@F2.8 MTF50 = 32 lines/mm
@F4 MTF50 = 36 lines/mm

Quoting Cicala that quoted Rudolf Kingslake:
'As I�m sure you know, the indispensible Kingslake�s History of the Photographic Lens points out that the Petzval design �uses overcorrected astigmatism to flatten the tangential field . . . giving excellent definition in the center of the image deteriorating rapidly towards the edges.�'

"Excellent definition in the center of the image"? Hmmm�
Of all the lenses I own, only the Fujinon 54-270mm F4.5 zoom and the Meyer Telemegor 400mm F5.5 lenses have worse wide-open resolution than the Lomography's Petzval lens.

If you are interested, the Lomography's Petzval lens can be rented from Lens Rental at modest $35 for 5 days!

luisalegria · Expire: 2018-01-18

Interesting, thanks.

The main purpose of this thing, as said, is extreme curvature of field for special purposes, such as retro style portraits among others.

I have my own Petzval thats also f/2.2 roughly, and its certainly not perfectly sharp at the center either, but gives a more intense curvature than the Lomography lens seems to do.

danfromm · Joined: 04 Sep 2011 Posts: 576

Gerald · Joined: 25 Mar 2014 Posts: 1196 Location: Brazil

luisalegria · Expire: 2018-01-18

Interesting chart, thank you.

P1 seems like the classic type of 19th century portrait and projection lens, of which there were several variations even at the time, though they all seem to have at least retained the air gap between the last two elements. I think Dallmeyer reversed the convex and concave elements, etc.

The main purpose of a Petzval in our day it seems to me is to exploit its defects, not its virtues, as a special effects lens. Notably, we want plenty of field curvature and very bad corners. Sort of a radial Scheimpflug "toytown" effect.

iangreenhalgh1 · Expire: 2014-01-07

Measuring the resolution of a Petzval lens is totally missing the point of the thing. It is the unique rendering that is important, not the resolution.

ZoneV · Expire: 2011-12-02

Gerald, how did you measure your MTF values? Do you have a MTF bench, or is it measured with camera (with AA filter and unkown image processing) and testcharts? Waht illumination?
I suppose your results are line pairs per millimeter?

I think one could carefully make measurments and compare this measurments, but comparing own measurments with complete different made measurements is not very accurate.

Gerald · Joined: 25 Mar 2014 Posts: 1196 Location: Brazil

I have measured MTF50 with the MTF Mapper program, which is based on the slanted edge method. The test chart used is one meter wide. By the way, I'm thinking of writing a post with instructions for how to use efficiently the MTF Mapper, which is a free but very interesting program.

The images were captured with a Sony A99 camera that has a pixel pitch equal to 6 microns. The RAW images were converted by Adobe Camera RAW with sharpening set to zero. The illumination used was white light with a temperature of approximately 3000 �K.

The Nyquist frequency for the Sony A99 is 83.33 lp/mm (line pairs per millimeter). The ideal AA filter for this camera should therefore have a unit response from 0 to 83.33 lp/mm, and zero thereafter. Sony does not publish the response of the A99's AA filter, but it is reasonable to assume that the deviation from the ideal filter is small and can be ignored for frequencies in the range 20 to 40 lp/mm.

Attila · Expire: 2025-11-18

Gerald · Joined: 25 Mar 2014 Posts: 1196 Location: Brazil

I'm just wondering how it is possible to design and evaluate a lens without measuring it. Joseph Petzval was a brilliant Hungarian mathematician and physicist. No doubt he knew the value of measurements.

You can be sure that the Petzval lens is only being able to be revived by Lomography because it was thoroughly measured, as shown by the MTF curves measured by the designers themselves on the first working prototype:

https://www.kickstarter.com/projects/lomography/the-lomography-petzval-portrait-lens

Speaking of brilliant minds, one of the greatest English scientists, Lord Kelvin (Sir William Thomson), became famous, among other things, for his comments on the importance of measurements:

"To measure is to know."

"If you cannot measure it, you cannot improve it."

"I often say that when you can measure what you are speaking about, and express it in numbers, you know something about it, but when you cannot measure it, when you cannot express it in numbers, your knowledge is of a meager and unsatisfactory kind; it may be the beginning of knowledge, but you have scarcely in your thoughts advanced to the state of Science, whatever the matter may be."

luisalegria · Expire: 2018-01-18

In my professional career I have done all I can to make sure that we do have valid measures.
Its amazing how hard it is to convince people of that though.

iangreenhalgh1 · Expire: 2014-01-07

Scientists and geeks measure things, photographers are just thankful the thing exists and then go and put it to good use.

There is a simple truism : if it's sharp enough, it's sharp enough

What is sharp enough is very much a matter of taste when it comes to portraits, shooting portraits is an art form and measurements don't matter at all when one is creating art.

philslizzy · Joined: 07 Aug 2012 Posts: 4748 Location: Cheshire, England

so where are the photos? charts say nothing

dan_ · Expire: 2016-12-19

The excessive use of MTF charts, as induced by the lens makers nowadays, is only in their commercial advantage. It creates a false demand for sharper and sharper (and therefore more expensive) lenses no matter of their intended use.

Measurements are useful for the user only if they are focused on those characteristics important for the intended use of an object (lens in our case).
Clinical sharpness is rarely the most important characteristic to look for when you make a portrait. Therefor measuring it tells very little about the required optical qualitys of a portrait lens.

Nevertheless, I admit it could be fun Smile

.
Has anybody draw MTF charts for soft focus lenses, too?

Gerald · Joined: 25 Mar 2014 Posts: 1196 Location: Brazil

RSalles · Posted: Sat Jul 19, 2014 5:21 pm Post subject:

Gerald,

I agree with what you've wrote, the main problem is when people begin to "discard" the lens based only on MTF chart datasets...
Example: what picture has more potential to please the the 40's female public portrait, a Leica Noctilux,r a Petzval-type or an Anastigmat/Rapid Rectilinear one? And theese lens for Architecture will be - probably - a disaster compared with the latest MC Rodenstock APO Gandragon.
My understanding is that the Petzval lens is a tool basically for portraiture, where perfect definition is a character of the lens that is desirable in some extent but is not the more important factor,

Cheers,

Renato

Gerald · Joined: 25 Mar 2014 Posts: 1196 Location: Brazil

RSalles · Posted: Sun Jul 20, 2014 1:10 am Post subject:

Gerald wrote:

Another conclusion we can draw is that the "theory" that a good portrait requires a low-resolution lens is questionable. But if a photographer really wants to do a portrait with soft focus, he just uses a high-resolution lens and softens the photo with post-processing.

Yes, but the charming of these lens is that it comes with "pos-processing" built-in! I saw some portraits made with V�ightlander Heliar with its creamy bokeh sharp in the center and soft in the borders, so beautiful that makes me want to "eat" the picture, Never found a portrait like this with another small-format MTF steroided lens...

Very Happy

Cheers,

Renato

scsambrook · Expire: 2011-11-18

This is another interesting thread, to which I'm coming late.
I can't really agree with Gerald who wrote "And the best way to tell if a lens is functioning as it should, is measuring the resolution with a chart. Then someone could contest: "Hey, I can test a lens simply taking photographs." Yes, you can, but this is a dumb way to test a lens! Erwin Puts, a well known author of books on Leica cameras and lenses, created a rule for lens testing, which he called "rule of 500" which is in his own words: "To get a good idea of how the lens performs we need to take at least 500 pictures!" What a dumb rule!" (My added emphasis)

Whilst not demeaning the value of scientific testing in a carefully controlled environment, I'm not sure why it's 'dumb' to test a lens by taking plenty of photographs - at least not for hobby photographers as opposed to the professional journalists or self-publishing bloggers who pontificate on how 'good' a lens is. Sophisticated test equipment in the hands of thoughtful and experienced people may indeed tell us a great deal about the technical performance of a lens, and whether it's properly put together - "functioning as it should" - but it won't necessarily tell us about how 'pleasing' or �disappointing� the lens will be to its user. After all, do we not have lenses to take photographs, to make aesthetically satisfying images that satisfy us according to our own perceptions?

Erwin Puts's 'rule of 500' is neither new nor foolish. In my late teens I was impressed by the common-sense writing of Theo Scheerer who, amongst many other things, wrote that the best way to get to know and understand a lens was to use it on its own for a time. He suggested putting away everything else, using it extensively and then carefully looking at what it and the user turned out between them. That, surely, is sensible. Over half a century after first reading Scheerer I'm more convinced than ever of the wisdom of such a philosophy. There's more to using a lens than we learn from from MTF curves or other laboratory outputs. They certainly have their place but they don't tell the whole story, do they?

ZoneV · Expire: 2011-12-02

MTF curves are not valuable without knowing how exact are they generated. Are they baed only on the optical design theory, or real measurements. Are they made with broad band illumination and not weighted, or only smaller spectral bandwith.
At which distance or magnification. How are the graphs processed (smoothing).

I do get sometimes MTF curves for lenses I test. But this is not much help - I measure all the lenses myself, to get comparable results.

dan_ · Expire: 2016-12-19

To come back to the Petzval portrait lens, have a look at this Petzval-made portrait by Jim Galli:
http://tonopahpictures.0catch.com/JesusAt100.jpg
The fine gradation of tones, the mild transition from sharp to unsharp, the ponderate contrast - all are pointing to what's important in the portrait and have a very strong power to express both the inner life of the old man and the photographer atitude on the subject. These, the swirly bokeh and a mild vignetting are the optical qualities one should expect from a Petzval lens.
Used well, as Mr. Galli did, they could be very expresive tools.

Could that be obtained with a modern clinically sharp and contrasty lens and post processing? I very much doubt of it.
Could all this optical qualityes of the lens be intuited by reading its MTF chart? No. If you'd read the MTF of this Petzval your conclusion should be that it's a crappy lens.
Not saying that MTFs are useless. If done properly they succeed well in describing sharpness, contrast, etc. but they can't properly describe all the optical qualities of a lens. And therefore they are quite an improper way to see, for instance, if a Petzval lens is better or worse than other Petzval lens.

In this case (and a lot of other cases) the '500 photos rule' of Erwin Puts seems much more appropriate.
I don't have to make all those photos by myself - I just surf the internet for as many samples as I can find before buying a lens.
They'll reveal the potential of the lens. If what I see is what I want then I'll buy it and use it extensively for a while to get used to it.
If I feel comfortable with it I'll keep it, if not I'll re-sell it (well, that's theoreticaly - in fact, in most cases, my greedy nature makes me keep it anyway Smile

).

And, as always, there is a shortcut : just ask Attila (or other trustfull and experienced members of this forum) and if he says it's a good lens, buy it. Wink

kds315* · Expire: 2021-03-09

Excellent examples Dan, Jim is a master and knows his lenses very well.

Nordentro · Expire: 2015-01-29

MTF charts is for the pixelpeepers who has interest in modern lenses without character Laughing

I care about different lens designs and how different lens design renders and try to use this as a benefit when shooting. Helios 40 is probably not a sharp knife either, but it is loved by many because of it`s rendering. Most lenses are sharp enough IMO! Wink

iangreenhalgh1 · Expire: 2014-01-07

Tedat · Joined: 08 Nov 2011 Posts: 800 Location: Berlin/Germany