Poll: unquantifiable lens quality

sigala1 said:

mh2000 said:

sigala1 said:

amtberg said:

sigala1 said:

richarddd said:

Are there lens qualities that can't be objectively measured that make one lens better than another? MTF, distortion, chromatic aberration, etc. can all be measured, but are there additional factors (for example, "rendering") that are important to picture quality that can't be quantified.

This relates solely to image quality, not AF speed, build quality, etc.

Click to expand...

Set up an experiment in which some photos are taken with a lens that is supposed to "render" really well, and another lens that's merely "clinical", and see if expert photograph judges can tell the difference.

Click to expand...

I think that a description of "clinical" often refers to a lens that actually measures very well, whereas -- e.g., for video -- a lens described as "cinematic" may have rather poor measurables. "Artistic" bokeh is often far from what one would consider "good" bokeh. It may be swirly or busy looking.

This is why it would be difficult to devise a test for these qualities. Things like sharpness and vignetting can be measured, whereas bokeh and contrast are often a matter of taste. Good bokeh for one purpose may be terrible for another and v.v.

Click to expand...

If the qualities exist and make a difference, then expert photographic judges should be able to detect photos taken with the "better" lens. We need quantifiable research and not opinions.

Click to expand...

Who are these "expert photographic judges" that you keep talking about? Proffesional lens testers? Artists? Wedding photographers? Tastes are personal.

Click to expand...

If ANY of these groups could tell, to a statistically significant confidence level, whether photos were taken from a lens that had better "rendering," then that would scientifically prove that there's something real there.

Click to expand...

Doubt many people would buy this. Back when I was looking at the 17/2.8 and 20, I asked for samples. One kind person provided side by side samples and the expert panel reading the thread agreed that the "rendering" from the 17 was nicer. So by science, this was proven and I bought the lens. I'm happy enough with it. Then people who hate the 17 set out to prove that the rendering concusion was wrong and they put together other tests to prove that the 20 is just as nice or nicer using other sample, in these tests, the majority were swayed to the 20.

The best you can do is show that one lens renders one scene better than another. If the test happens to be the same scene that you plan on shooting, you're all set to pick a lens. Unfortunately, I don't shoot the test shots.

Anyway, best test for me is to buy a lens and shoot it a bunch and decide which I like better on average based on the photos that are meaningful to me. I don't care that much about other people's photos.

mh2000 said:

tt321 said:

Elemental Photography said:

The optical properties of a lens are pure physics. There is nothing unquantifiable with regards to how light transits the lens elements on the way to the sensor.

I will grant that not all of these properties are currently quantified in lens reviews. That doesn't make them unquantifiable, however, simply unknown quantities at the moment.

Click to expand...

Exactly!

Globally unquantifiability probably does not exist. But the existence of unquantified issues because of tech and effort limitations as well as not strong enough motivations has left room for mystique, and will always do so.

Click to expand...

No, something can be globally unquantifiable when finitely definable with infinitely fine gradiations!

Click to expand...

In the context of lenses, each lens, of the general size we are looking at and up to any finite size smaller than the entire Universe, must be built with a finite number of fundamental particles. It must have a finite number of elements with a finite number of surfaces. Each of these finite number of surfaces can only have a finite number of possible shapes simply because the surface consists of a finite number of atoms. The combinational 'rendering' effects the lens could possibly have, given the finite number of possible designs it has, must therefore be finite.

There is no 'infinite fine gradations' in this context, without even bringing in the fact that we are using a finite number of pixels each quantized into a finite number of light levels for a finite number of colours.

As I said, if a quality can be seen and different degrees from different lenses of this quality can be differentiated, it should be possible to mathematically model it and grade it numerically and, indeed, measured. Therefore it would be a quantifiable quality. Whether we want to deal with such a large number of numbers with every quality quantified is a different matter. With the limited qualities quantified as of today, testers such as DxO are already publishing 'overall scores' which are probably weighted sums of the quantified parameters in the interest of easy reading.

Dheorl said:

Anders W said:

Dheorl said:

Anders W said:

Dheorl said:

Anders W said:

tt321 said:

I think there is a slight misunderstanding between what can be measured and what are currently most popularly measured and presented with numbers. I am sure a mathematical representation could be found for anything that people might consider important. There is just not enough interest for people to set up precise measures of stuff like bokeh. 'Rendering' is too fuzzy a word to begin with, but if its effects are visible, then contributive factors should all be possible to model, measure and present in numerical terms. Whether one wants to deal with such a large number of numbers is a different matter.

Click to expand...

Agreed. Everything can be measured but we certainly don't have exhaustive measurements, they can't really be made exhaustive (there are infinitely many ways to slice the baloney), and people wouldn't be able to cope with the results if they could be made exhaustive.

That said, I think there is room for more/better measurements of some things, for example bokeh. I am not sure to which extent it would make sense to try to quantify it, but we do know the blur disc characteristics that most people associate with good and bad bokeh: shape of the blur disc (ideally round), light distribution of the blur disc (ideally brighter core than periphery and no "onion rings"), and discoloration (color fringes) of the blur disc (ideally none). I think it would help significantly if tests would let us judge these things, if only visually.

Click to expand...

I'm not sure what is classed as good bokeh is so easy to determine. I think whereas for instance no-one would want a lens with more chromatic aberration (unless they're after some ye-olde effect I guess), and it could therefore be marked on a scale of one to ten. With bokeh on the other hand I know some people like the swirly bokeh effect you get with some lenses, others don't; some like foliage to be buttery, others prefer a bit of a choppy appearance. Personally it depends on what I want from the particular photo, but my point is one type of bokeh characteristic is the best a lot less often then it is with almost any other lens characteristic.

Click to expand...

I don't think any of that is really a problem. First, I doubt that the prevalence of odd tastes is any greater for bokeh than for other characteristics. For example, I have heard quite a few people say that they like the halation that comes with spherical aberration and that they want their portrait lens to be soft and "glowy" rather than sharp wide open. Second, as long as the blur disc characteristics I mention are there for us to see, everyone can arrive at their own judgment. If you like swirly bokeh, pick the lens with strongly deformed blur discs at the edges of the frame (swirly bokeh is a matter of vignetting and the deformed blur discs that come with it) rather than the one with round. If you like your OOF foliage choppy rather than buttery, pick the lens whose blur discs have strong outlining rather than the one having blur discs with a brighter core. And so on.

Click to expand...

Well yes, but what I'm saying is that because of this it's a characteristic that can't be ranked.

Click to expand...

Oh I think it can. As already pointed out, those in the minority are of course free to prefer lenses with lower rank to those with higher, just as they can with regard to sharpness, vignetting, distortion, CA and so on. But I certainly don't see such a quantification as particularly important and would like to see it supplemented by visual evidence anyway.

Click to expand...

I don't think it's as much of a minority as it is with any other lens characteristic. Also there's so many ways in which bokeh can be bad/good that a lower number would still be completely meaningless.

Click to expand...

I think suitable measures of the relatively few dimensions I mentioned would suffice. Why would they not?

Dheorl said:

Dheorl said:

Dheorl said:

I'm all for lens test including bokeh samples, but whereas sharpness or vignetting or something I would be happy to rely on a number for to make a judgement I would never be happy with a number to make a judgement on bokeh.

Click to expand...

As I indicated already in the reply to tt321 that you commented on, the important point for me in this case isn't quantification but suitable visual evidence that let us properly assess the blur disc characteristics in all relevant regards. Some test sites (Lenstip, Photozone) have already taken steps in this direction but neither of them do it as well as I would like.

Click to expand...

It's one thing (and possibly the only thing) which I actually think Ken Rockwell actually does a lot better than many sites. The initial thread was about unquantifiable characteristics though. Your method would not quantify the bokeh characteristic, therefore it is an unquantifiable characteristic that is important with regards to a lenses rendering.

Click to expand...

I only said that it was not essential to me to have quantified measures of bokeh. I certainly think that quantification of this characteristic is possible.

amtberg said:

Anders W said:

Dheorl said:

Anders W said:

Dheorl said:

Anders W said:

tt321 said:

I think there is a slight misunderstanding between what can be measured and what are currently most popularly measured and presented with numbers. I am sure a mathematical representation could be found for anything that people might consider important. There is just not enough interest for people to set up precise measures of stuff like bokeh. 'Rendering' is too fuzzy a word to begin with, but if its effects are visible, then contributive factors should all be possible to model, measure and present in numerical terms. Whether one wants to deal with such a large number of numbers is a different matter.

Click to expand...

Agreed. Everything can be measured but we certainly don't have exhaustive measurements, they can't really be made exhaustive (there are infinitely many ways to slice the baloney), and people wouldn't be able to cope with the results if they could be made exhaustive.

That said, I think there is room for more/better measurements of some things, for example bokeh. I am not sure to which extent it would make sense to try to quantify it, but we do know the blur disc characteristics that most people associate with good and bad bokeh: shape of the blur disc (ideally round), light distribution of the blur disc (ideally brighter core than periphery and no "onion rings"), and discoloration (color fringes) of the blur disc (ideally none). I think it would help significantly if tests would let us judge these things, if only visually.

Click to expand...

I'm not sure what is classed as good bokeh is so easy to determine. I think whereas for instance no-one would want a lens with more chromatic aberration (unless they're after some ye-olde effect I guess), and it could therefore be marked on a scale of one to ten. With bokeh on the other hand I know some people like the swirly bokeh effect you get with some lenses, others don't; some like foliage to be buttery, others prefer a bit of a choppy appearance. Personally it depends on what I want from the particular photo, but my point is one type of bokeh characteristic is the best a lot less often then it is with almost any other lens characteristic.

Click to expand...

I don't think any of that is really a problem. First, I doubt that the prevalence of odd tastes is any greater for bokeh than for other characteristics. For example, I have heard quite a few people say that they like the halation that comes with spherical aberration and that they want their portrait lens to be soft and "glowy" rather than sharp wide open. Second, as long as the blur disc characteristics I mention are there for us to see, everyone can arrive at their own judgment. If you like swirly bokeh, pick the lens with strongly deformed blur discs at the edges of the frame (swirly bokeh is a matter of vignetting and the deformed blur discs that come with it) rather than the one with round. If you like your OOF foliage choppy rather than buttery, pick the lens whose blur discs have strong outlining rather than the one having blur discs with a brighter core. And so on.

Click to expand...

Well yes, but what I'm saying is that because of this it's a characteristic that can't be ranked.

Click to expand...

Oh I think it can. As already pointed out, those in the minority are of course free to prefer lenses with lower rank to those with higher, just as they can with regard to sharpness, vignetting, distortion, CA and so on. But I certainly don't see such a quantification as particularly important and would like to see it supplemented by visual evidence anyway.

amtberg said:

I'm all for lens test including bokeh samples, but whereas sharpness or vignetting or something I would be happy to rely on a number for to make a judgement I would never be happy with a number to make a judgement on bokeh.

Click to expand...

As I indicated already in the reply to tt321 that you commented on, the important point for me in this case isn't quantification but suitable visual evidence that let us properly assess the blur disc characteristics in all relevant regards. Some test sites (Lenstip, Photozone) have already taken steps in this direction but neither of them do it as well as I would like.

Click to expand...

Can you post a link to the study or poll that determined what the majority thinks constitutes good bokeh?

Click to expand...

Can you post a link to the study or poll that determined what the majority thinks constitutes good rendering with regard to sharpness?

I don't think I misinterpreted the survey question, but putting that aside, perhaps the second point is a better one than the first. We may be able to find out why a lens produces the kind of result that it does.

But even if it were made "perfectly" is would still have a character that is not measurable, no matter how clever we are because by definition quality is not measurable. It can be statistically approximated (fallibly) through operationalization of some king, but is not reduceable to mathematics.

F.

Sorry for typos. I can't seem to edit my post.

FrankS009 said:

I don't think I misinterpreted the survey question, but putting that aside, perhaps the second point is a better one than the first. We may be able to find out why a lens produces the kind of result that it does.

But even if it were made "perfectly" is would still have a character that is not measurable, no matter how clever we are because by definition quality is not measurable. It can be statistically approximated (fallibly) through operationalization of some king, but is not reduceable to mathematics.

Click to expand...

I think any optical quality can be made measureable if we can agree on its definition. But ... a) there are infinitely many such qualities (which makes it somewhat difficult to measure them all ;-) ), b) we might not be able to agree on the definition, and c) our visual perception as well as evaluation of the same measureable quality might well differ.

Our conversation shifts to the notion of definition. But I think we are close enough in general that we don't need to beat the epistemological issues to death in this form. Appreciated your response to my thoughts.

F.

FrankS009 said:

Our conversation shifts to the notion of definition. But I think we are close enough in general that we don't need to beat the epistemological issues to death in this form. Appreciated your response to my thoughts.

Click to expand...

Yes, I think we are saying roughly the same thing although we express it differently.

One example I used to use when teaching scientific methods to my students was to hold up the textbook we were using in front of them and ask them to describe it. I had them provide examples for a while (its title, its general topic, the color of the cover, the number of pages, and so on) just to illustrate the fact that we could go on forever without still having provided an exhaustive description of the book. There is no such thing as an exhaustive description of anything simply because there is an infinite number of ways in which we can conceptualize (define) the properties of (and hence measure) anything (right down to the chemical composition of the paper in the upper right corner of page 189 of a particular copy of that textbook). Once we have conceptualized something, however, we can measure it (although obtaining that measure can sometimes be forbiddingly difficult).

tt321 said:

mh2000 said:

tt321 said:

Elemental Photography said:

The optical properties of a lens are pure physics. There is nothing unquantifiable with regards to how light transits the lens elements on the way to the sensor.

I will grant that not all of these properties are currently quantified in lens reviews. That doesn't make them unquantifiable, however, simply unknown quantities at the moment.

Click to expand...

Exactly!

Globally unquantifiability probably does not exist. But the existence of unquantified issues because of tech and effort limitations as well as not strong enough motivations has left room for mystique, and will always do so.

Click to expand...

No, something can be globally unquantifiable when finitely definable with infinitely fine gradiations!

Click to expand...

In the context of lenses, each lens, of the general size we are looking at and up to any finite size smaller than the entire Universe, must be built with a finite number of fundamental particles. It must have a finite number of elements with a finite number of surfaces. Each of these finite number of surfaces can only have a finite number of possible shapes simply because the surface consists of a finite number of atoms. The combinational 'rendering' effects the lens could possibly have, given the finite number of possible designs it has, must therefore be finite.

There is no 'infinite fine gradations' in this context, without even bringing in the fact that we are using a finite number of pixels each quantized into a finite number of light levels for a finite number of colours.

As I said, if a quality can be seen and different degrees from different lenses of this quality can be differentiated, it should be possible to mathematically model it and grade it numerically and, indeed, measured. Therefore it would be a quantifiable quality. Whether we want to deal with such a large number of numbers with every quality quantified is a different matter. With the limited qualities quantified as of today, testers such as DxO are already publishing 'overall scores' which are probably weighted sums of the quantified parameters in the interest of easy reading.

Click to expand...

And how many people on this forum don't agree with DxO's overall scores? These are even the techno-nerds not agreeing with their weighting factors, not the sensitive artist types!

Regarding the infinite, from the point of view of an alien, there would probably appear to be a very limited number of archetypes of human beings, but as a human being, we tend to think of ourselves as all being unique because we are very attuned to the subtle differences. As with lenses, we are all made from the same fundamental particals of the universe and subject to the same limitations in variation. The same can be said about pixelization limitations and our faces -- yes, a digital image has a finite number of possible captures, but when you run through the math, even as an argument, I think you'd have to agree that it very close to being a numeric infinite in magnitude, even for a 12MP sensor. IE. it would be very very very unlikely that any photo or human will be truly identical. There is no physical law that procludes the fact that all my atoms could spontaneously be disassembled and then reassembled on the face of the moon, but it is so unlikely that I don't worry about that being a realistic possibility.

Anders W said:

FrankS009 said:

I don't think I misinterpreted the survey question, but putting that aside, perhaps the second point is a better one than the first. We may be able to find out why a lens produces the kind of result that it does.

But even if it were made "perfectly" is would still have a character that is not measurable, no matter how clever we are because by definition quality is not measurable. It can be statistically approximated (fallibly) through operationalization of some king, but is not reduceable to mathematics.

Click to expand...

I think any optical quality can be made measureable if we can agree on its definition. But ... a) there are infinitely many such qualities (which makes it somewhat difficult to measure them all ;-) ), b) we might not be able to agree on the definition, and c) our visual perception as well as evaluation of the same measureable quality might well differ.

Click to expand...

Hahaha! Well Anders, we have not always agreed on lens assement, but I completely agree with this!

mh2000 said:

Anders W said:

FrankS009 said:

I don't think I misinterpreted the survey question, but putting that aside, perhaps the second point is a better one than the first. We may be able to find out why a lens produces the kind of result that it does.

But even if it were made "perfectly" is would still have a character that is not measurable, no matter how clever we are because by definition quality is not measurable. It can be statistically approximated (fallibly) through operationalization of some king, but is not reduceable to mathematics.

Click to expand...

I think any optical quality can be made measureable if we can agree on its definition. But ... a) there are infinitely many such qualities (which makes it somewhat difficult to measure them all ;-) ), b) we might not be able to agree on the definition, and c) our visual perception as well as evaluation of the same measureable quality might well differ.

Click to expand...

Hahaha! Well Anders, we have not always agreed on lens assement, but I completely agree with this!

Click to expand...

Good to see eye to eye with you for a change, although I am sure it won't last for long. ;-)

mh2000 said:

Elemental Photography said:

mh2000 said:

You can isolate all the parameters you want, but how you put them together into a complex overall parameter is subject to personal perception. How do you weight each? This is what people subjectively do in their head.

Click to expand...

Absolutely true. Beauty is subjective and in the eye of the beholder. The example of flare is an excellent one. I do not much care for flare, so lenses which exhibit a lot of it do not appeal to me. Others love flare and seek out old, scratched lenses to explore. The amount of flare can be quantified, with some work. Whether that flare amount is positive or negative is subjective.

So, yes, everything to do with light propagation through a lens assembly can be quantified. Which combination of quantities is desirable? That is subjective.

--
A photograph is a creative interpretation of reality.

Click to expand...

How exactly you would quantify the fullness of flare would be very hard. Of three lenses that know how they flare, the Oly 17/2.8, the P20 and the S30, the 17 always makes a soft diffuse glow when it flares around bright lights while the P20 always produces very sharp starbursts.

Click to expand...

Nope. The P20 doesn't always produce very sharp starburts, although, like every other lens, it will do so if you stop it down sufficiently.

mh2000 said:

The 17 more preferable to me because it appears more "cinimatic" to my eye. The Sigma 30 produces a weird combination of the other two:

You certaingly can't capture the property with a single value. To my eye, I've accepted the behavior of the Sigma 30 even if I prefer the look of the 17 under these conditions... but I prefer the 30mm FL.

--
mark hahn
http://markhahnphotography.wordpress.com/

Click to expand...

mh2000 said:

tt321 said:

mh2000 said:

tt321 said:

Elemental Photography said:

The optical properties of a lens are pure physics. There is nothing unquantifiable with regards to how light transits the lens elements on the way to the sensor.

I will grant that not all of these properties are currently quantified in lens reviews. That doesn't make them unquantifiable, however, simply unknown quantities at the moment.

Click to expand...

Exactly!

Globally unquantifiability probably does not exist. But the existence of unquantified issues because of tech and effort limitations as well as not strong enough motivations has left room for mystique, and will always do so.

Click to expand...

No, something can be globally unquantifiable when finitely definable with infinitely fine gradiations!

Click to expand...

In the context of lenses, each lens, of the general size we are looking at and up to any finite size smaller than the entire Universe, must be built with a finite number of fundamental particles. It must have a finite number of elements with a finite number of surfaces. Each of these finite number of surfaces can only have a finite number of possible shapes simply because the surface consists of a finite number of atoms. The combinational 'rendering' effects the lens could possibly have, given the finite number of possible designs it has, must therefore be finite.

There is no 'infinite fine gradations' in this context, without even bringing in the fact that we are using a finite number of pixels each quantized into a finite number of light levels for a finite number of colours.

As I said, if a quality can be seen and different degrees from different lenses of this quality can be differentiated, it should be possible to mathematically model it and grade it numerically and, indeed, measured. Therefore it would be a quantifiable quality. Whether we want to deal with such a large number of numbers with every quality quantified is a different matter. With the limited qualities quantified as of today, testers such as DxO are already publishing 'overall scores' which are probably weighted sums of the quantified parameters in the interest of easy reading.

Click to expand...

And how many people on this forum don't agree with DxO's overall scores? These are even the techno-nerds not agreeing with their weighting factors, not the sensitive artist types!

Regarding the infinite, from the point of view of an alien, there would probably appear to be a very limited number of archetypes of human beings, but as a human being, we tend to think of ourselves as all being unique because we are very attuned to the subtle differences. As with lenses, we are all made from the same fundamental particals of the universe and subject to the same limitations in variation. The same can be said about pixelization limitations and our faces -- yes, a digital image has a finite number of possible captures, but when you run through the math, even as an argument, I think you'd have to agree that it very close to being a numeric infinite in magnitude, even for a 12MP sensor. IE. it would be very very very unlikely that any photo or human will be truly identical. There is no physical law that procludes the fact that all my atoms could spontaneously be disassembled and then reassembled on the face of the moon, but it is so unlikely that I don't worry about that being a realistic possibility.

Click to expand...

This is arguing that there is practical unquantifiability, which I suggested many posts ago. I was arguing that there is very likely no absolute unquantifiability.

With human beings, for instance, last time I checked we have not achieved infinity yet. Using infinity to replace very large numbers is just a practical convenience. Even with infinity, we could still compare bad, good, better, etc. in a numerical scale. For instance, even if we assume that temperature could rise to infinity, two different temperatures can still be decided as hotter or cooler than each other by comparing them numerically.

The OP did not clarify what unquantifiable meant. Hence the request for clarification about unquantifiable and unquantified at present.

Anders W said:

mh2000 said:

Anders W said:

FrankS009 said:

I don't think I misinterpreted the survey question, but putting that aside, perhaps the second point is a better one than the first. We may be able to find out why a lens produces the kind of result that it does.

But even if it were made "perfectly" is would still have a character that is not measurable, no matter how clever we are because by definition quality is not measurable. It can be statistically approximated (fallibly) through operationalization of some king, but is not reduceable to mathematics.

Click to expand...

I think any optical quality can be made measureable if we can agree on its definition. But ... a) there are infinitely many such qualities (which makes it somewhat difficult to measure them all ;-) ), b) we might not be able to agree on the definition, and c) our visual perception as well as evaluation of the same measureable quality might well differ.

Click to expand...

Hahaha! Well Anders, we have not always agreed on lens assement, but I completely agree with this!

Click to expand...

Good to see eye to eye with you for a change, although I am sure it won't last for long. ;-)

Click to expand...

I'm kind of drifting back into toy lens territory lately, so quibling about the minor differences in otherwise good lenses is not at the top of my priority list.



Just picked up a set of my prints from a show that were taken with the m43 Holga lens -- hahaha, no, they didn't sell, but I really really like them. Sometimes you really need to study prints to make an overall assessment of your work and equipment.

tt321 said:

mh2000 said:

tt321 said:

mh2000 said:

tt321 said:

Elemental Photography said:

The optical properties of a lens are pure physics. There is nothing unquantifiable with regards to how light transits the lens elements on the way to the sensor.

I will grant that not all of these properties are currently quantified in lens reviews. That doesn't make them unquantifiable, however, simply unknown quantities at the moment.

Click to expand...

Exactly!

Globally unquantifiability probably does not exist. But the existence of unquantified issues because of tech and effort limitations as well as not strong enough motivations has left room for mystique, and will always do so.

Click to expand...

No, something can be globally unquantifiable when finitely definable with infinitely fine gradiations!

Click to expand...

In the context of lenses, each lens, of the general size we are looking at and up to any finite size smaller than the entire Universe, must be built with a finite number of fundamental particles. It must have a finite number of elements with a finite number of surfaces. Each of these finite number of surfaces can only have a finite number of possible shapes simply because the surface consists of a finite number of atoms. The combinational 'rendering' effects the lens could possibly have, given the finite number of possible designs it has, must therefore be finite.

There is no 'infinite fine gradations' in this context, without even bringing in the fact that we are using a finite number of pixels each quantized into a finite number of light levels for a finite number of colours.

As I said, if a quality can be seen and different degrees from different lenses of this quality can be differentiated, it should be possible to mathematically model it and grade it numerically and, indeed, measured. Therefore it would be a quantifiable quality. Whether we want to deal with such a large number of numbers with every quality quantified is a different matter. With the limited qualities quantified as of today, testers such as DxO are already publishing 'overall scores' which are probably weighted sums of the quantified parameters in the interest of easy reading.

Click to expand...

And how many people on this forum don't agree with DxO's overall scores? These are even the techno-nerds not agreeing with their weighting factors, not the sensitive artist types!

Regarding the infinite, from the point of view of an alien, there would probably appear to be a very limited number of archetypes of human beings, but as a human being, we tend to think of ourselves as all being unique because we are very attuned to the subtle differences. As with lenses, we are all made from the same fundamental particals of the universe and subject to the same limitations in variation. The same can be said about pixelization limitations and our faces -- yes, a digital image has a finite number of possible captures, but when you run through the math, even as an argument, I think you'd have to agree that it very close to being a numeric infinite in magnitude, even for a 12MP sensor. IE. it would be very very very unlikely that any photo or human will be truly identical. There is no physical law that procludes the fact that all my atoms could spontaneously be disassembled and then reassembled on the face of the moon, but it is so unlikely that I don't worry about that being a realistic possibility.

Click to expand...

This is arguing that there is practical unquantifiability, which I suggested many posts ago. I was arguing that there is very likely no absolute unquantifiability.

With human beings, for instance, last time I checked we have not achieved infinity yet. Using infinity to replace very large numbers is just a practical convenience. Even with infinity, we could still compare bad, good, better, etc. in a numerical scale. For instance, even if we assume that temperature could rise to infinity, two different temperatures can still be decided as hotter or cooler than each other by comparing them numerically.

The OP did not clarify what unquantifiable meant. Hence the request for clarification about unquantifiable and unquantified at present.

Click to expand...

A precise definition of undefinable is hard to articulate (sounds like a set theory paradox ).

The motivation for this thread was the frequent statement that a lens has some superior quality that is not quantified (or stronger, can not be quantified or weaker, can not be described with much precision). An early reply in this thread is a good example, "There is also something about the Canon 135L, and the Leica Summicron 50mm from the 1970s that cannot be explained with technical data but they are magic lenses." I hadn't really envisioned discussions of epistemology.

What would you regard as a good way to phrase a poll to elicit people's attitude's towards these statements about rendering or magic or the like?

Anders W said:

mh2000 said:

Elemental Photography said:

mh2000 said:

You can isolate all the parameters you want, but how you put them together into a complex overall parameter is subject to personal perception. How do you weight each? This is what people subjectively do in their head.

Click to expand...

Absolutely true. Beauty is subjective and in the eye of the beholder. The example of flare is an excellent one. I do not much care for flare, so lenses which exhibit a lot of it do not appeal to me. Others love flare and seek out old, scratched lenses to explore. The amount of flare can be quantified, with some work. Whether that flare amount is positive or negative is subjective.

So, yes, everything to do with light propagation through a lens assembly can be quantified. Which combination of quantities is desirable? That is subjective.

--
A photograph is a creative interpretation of reality.

Click to expand...

How exactly you would quantify the fullness of flare would be very hard. Of three lenses that know how they flare, the Oly 17/2.8, the P20 and the S30, the 17 always makes a soft diffuse glow when it flares around bright lights while the P20 always produces very sharp starbursts.

Click to expand...

Nope. The P20 doesn't always produce very sharp starburts, although, like every other lens, it will do so if you stop it down sufficiently.

Anders W said:

The 17 more preferable to me because it appears more "cinimatic" to my eye. The Sigma 30 produces a weird combination of the other two:

You certaingly can't capture the property with a single value. To my eye, I've accepted the behavior of the Sigma 30 even if I prefer the look of the 17 under these conditions... but I prefer the 30mm FL.

--
mark hahn
http://markhahnphotography.wordpress.com/

Click to expand...

Click to expand...

From my experience, in a shot like the one I posted, the Oly 17/2.8 would tend to render the street lights as a diffuse glow, the 20, in this case whould probably produce sharp stars and the S30 as above. Since there are essentially infinite shots to be take, there will be variations, but these are the tendancy of these lenses at f5.6. This is just my expererince with scenes like above. If someone never shot something like this the test doesn't matter to them, just like others posting a single example and trying to prove something about two lenses... all tests and all photos are different. Lenses have tendencies, tests can be designed to capture these tendencies, but not prove that a lens is "always better" etc. than another. For instance, as I pointed out, one person "proved" through test that the Oly 17/2.8 had better bokeh then the P20 and then you later "proved" that it didn't. Both tests could be right and both results can be right too since they were different tests. Most things with aethetic properties are not an absolute.

--
mark hahn

mh2000 said:

Anders W said:

mh2000 said:

Elemental Photography said:

mh2000 said:

You can isolate all the parameters you want, but how you put them together into a complex overall parameter is subject to personal perception. How do you weight each? This is what people subjectively do in their head.

Click to expand...

Absolutely true. Beauty is subjective and in the eye of the beholder. The example of flare is an excellent one. I do not much care for flare, so lenses which exhibit a lot of it do not appeal to me. Others love flare and seek out old, scratched lenses to explore. The amount of flare can be quantified, with some work. Whether that flare amount is positive or negative is subjective.

So, yes, everything to do with light propagation through a lens assembly can be quantified. Which combination of quantities is desirable? That is subjective.

--
A photograph is a creative interpretation of reality.

Click to expand...

How exactly you would quantify the fullness of flare would be very hard. Of three lenses that know how they flare, the Oly 17/2.8, the P20 and the S30, the 17 always makes a soft diffuse glow when it flares around bright lights while the P20 always produces very sharp starbursts.

Click to expand...

Nope. The P20 doesn't always produce very sharp starburts, although, like every other lens, it will do so if you stop it down sufficiently.

mh2000 said:

The 17 more preferable to me because it appears more "cinimatic" to my eye. The Sigma 30 produces a weird combination of the other two:

You certaingly can't capture the property with a single value. To my eye, I've accepted the behavior of the Sigma 30 even if I prefer the look of the 17 under these conditions... but I prefer the 30mm FL.

--
mark hahn
http://markhahnphotography.wordpress.com/

Click to expand...

Click to expand...

From my experience, in a shot like the one I posted, the Oly 17/2.8 would tend to render the street lights as a diffuse glow, the 20, in this case whould probably produce sharp stars and the S30 as above. Since there are essentially infinite shots to be take, there will be variations, but these are the tendancy of these lenses at f5.6. This is just my expererince with scenes like above. If someone never shot something like this the test doesn't matter to them, just like others posting a single example and trying to prove something about two lenses... all tests and all photos are different. Lenses have tendencies, tests can be designed to capture these tendencies, but not prove that a lens is "always better" etc. than another.

Click to expand...

All I am saying is that you need to take the aperture into account when discussing this issue. With the 20/1.7 wide open, as in the image I posted, there will be no "starbursts". If you stop it down sufficiently, there will.

The precise shape of the "starbursts" depends on the number of aperture blades. Lenses with an even number of blades, will produce as many rays as the number of blades. Lenses with an on odd number of blades, will produce twice as many rays as the number of blades. See for example here:

http://www.photocrati.com/fun-with-sunstars/

mh2000 said:

For instance, as I pointed out, one person "proved" through test that the Oly 17/2.8 had better bokeh then the P20 and then you later "proved" that it didn't. Both tests could be right and both results can be right too since they were different tests. Most things with aethetic properties are not an absolute.

Click to expand...

Who proved where that the 17/2.8 has better bokeh than the P20? Got a link?

Anders W said:

mh2000 said:

Anders W said:

mh2000 said:

Elemental Photography said:

mh2000 said:

You can isolate all the parameters you want, but how you put them together into a complex overall parameter is subject to personal perception. How do you weight each? This is what people subjectively do in their head.

Click to expand...

Absolutely true. Beauty is subjective and in the eye of the beholder. The example of flare is an excellent one. I do not much care for flare, so lenses which exhibit a lot of it do not appeal to me. Others love flare and seek out old, scratched lenses to explore. The amount of flare can be quantified, with some work. Whether that flare amount is positive or negative is subjective.

So, yes, everything to do with light propagation through a lens assembly can be quantified. Which combination of quantities is desirable? That is subjective.

--
A photograph is a creative interpretation of reality.

Click to expand...

How exactly you would quantify the fullness of flare would be very hard. Of three lenses that know how they flare, the Oly 17/2.8, the P20 and the S30, the 17 always makes a soft diffuse glow when it flares around bright lights while the P20 always produces very sharp starbursts.

Click to expand...

Nope. The P20 doesn't always produce very sharp starburts, although, like every other lens, it will do so if you stop it down sufficiently.

Anders W said:

The 17 more preferable to me because it appears more "cinimatic" to my eye. The Sigma 30 produces a weird combination of the other two:

You certaingly can't capture the property with a single value. To my eye, I've accepted the behavior of the Sigma 30 even if I prefer the look of the 17 under these conditions... but I prefer the 30mm FL.

--
mark hahn
http://markhahnphotography.wordpress.com/

Click to expand...

Click to expand...

From my experience, in a shot like the one I posted, the Oly 17/2.8 would tend to render the street lights as a diffuse glow, the 20, in this case whould probably produce sharp stars and the S30 as above. Since there are essentially infinite shots to be take, there will be variations, but these are the tendancy of these lenses at f5.6. This is just my expererince with scenes like above. If someone never shot something like this the test doesn't matter to them, just like others posting a single example and trying to prove something about two lenses... all tests and all photos are different. Lenses have tendencies, tests can be designed to capture these tendencies, but not prove that a lens is "always better" etc. than another.

Click to expand...

All I am saying is that you need to take the aperture into account when discussing this issue. With the 20/1.7 wide open, as in the image I posted, there will be no "starbursts". If you stop it down sufficiently, there will.

The precise shape of the "starbursts" depends on the number of aperture blades. Lenses with an even number of blades, will produce as many rays as the number of blades. Lenses with an on odd number of blades, will produce twice as many rays as the number of blades. See for example here:

http://www.photocrati.com/fun-with-sunstars/

Anders W said:

For instance, as I pointed out, one person "proved" through test that the Oly 17/2.8 had better bokeh then the P20 and then you later "proved" that it didn't. Both tests could be right and both results can be right too since they were different tests. Most things with aethetic properties are not an absolute.

Click to expand...

Who proved where that the 17/2.8 has better bokeh than the P20? Got a link?

Click to expand...

Right, that is why I stated that I was commenting on the all lenses set to f5.6... pretty much my standard aperture setting for m43 lenses.

The asterics around "proved" were meant to suggest that it was not a diffinitive usage of the word. This was a long time ago and the last time I looked the OP had removed the images, but you might be able to find the thread. More "expert" DPR threadsters agreed that in this particular instance the 17/2.8 was less harsh and more pleasing. At the time, I was new to m43 and used the example to sway me toward the cheaper 17/2.8 over the 20... I didn't have any bias toward one lens or the other. I know you have posted test shots later that don't agree with those initial comparisons I saw. But really, nothing to agrue about here. My girlfriend shoots a 20 almost exclusively and I have made extensive comparisons between the 17/2.8 and the 20, both pixel peeping and at identical print sizes, and while recognizing the 20 is in general the "better lens," still find some properties that I prefer coming from the 17 than from the 20, but the differences are really all in the noise in final analysis IMO. Stopped down, both lenses produce very good results IMO.

If you are suggesting that no test could show an instance of where the 17/2.8 produces more pleasing bokeh than the 20, then I would suggest that you do not understand bokeh and the subjectivity in assessing it in individual images.

Back when I shot a lot more wide aperture photos and was much more concerned with bokeh, I found that more telling than one gear head's attempt at a controlled bokeh test was a spin through 100's of flickr images from a lens to try and assess the average bokeh properties that emerge. All lenses can produce good and bad bokeh and it is very sensitive to many things.

--
mark hahn

mh2000 said:

Anders W said:

mh2000 said:

Anders W said:

mh2000 said:

Elemental Photography said:

mh2000 said:

You can isolate all the parameters you want, but how you put them together into a complex overall parameter is subject to personal perception. How do you weight each? This is what people subjectively do in their head.

Click to expand...

Absolutely true. Beauty is subjective and in the eye of the beholder. The example of flare is an excellent one. I do not much care for flare, so lenses which exhibit a lot of it do not appeal to me. Others love flare and seek out old, scratched lenses to explore. The amount of flare can be quantified, with some work. Whether that flare amount is positive or negative is subjective.

So, yes, everything to do with light propagation through a lens assembly can be quantified. Which combination of quantities is desirable? That is subjective.

--
A photograph is a creative interpretation of reality.

Click to expand...

How exactly you would quantify the fullness of flare would be very hard. Of three lenses that know how they flare, the Oly 17/2.8, the P20 and the S30, the 17 always makes a soft diffuse glow when it flares around bright lights while the P20 always produces very sharp starbursts.

Click to expand...

Nope. The P20 doesn't always produce very sharp starburts, although, like every other lens, it will do so if you stop it down sufficiently.

mh2000 said:

The 17 more preferable to me because it appears more "cinimatic" to my eye. The Sigma 30 produces a weird combination of the other two:

You certaingly can't capture the property with a single value. To my eye, I've accepted the behavior of the Sigma 30 even if I prefer the look of the 17 under these conditions... but I prefer the 30mm FL.

--
mark hahn
http://markhahnphotography.wordpress.com/

Click to expand...

Click to expand...

From my experience, in a shot like the one I posted, the Oly 17/2.8 would tend to render the street lights as a diffuse glow, the 20, in this case whould probably produce sharp stars and the S30 as above. Since there are essentially infinite shots to be take, there will be variations, but these are the tendancy of these lenses at f5.6. This is just my expererince with scenes like above. If someone never shot something like this the test doesn't matter to them, just like others posting a single example and trying to prove something about two lenses... all tests and all photos are different. Lenses have tendencies, tests can be designed to capture these tendencies, but not prove that a lens is "always better" etc. than another.

Click to expand...

All I am saying is that you need to take the aperture into account when discussing this issue. With the 20/1.7 wide open, as in the image I posted, there will be no "starbursts". If you stop it down sufficiently, there will.

The precise shape of the "starbursts" depends on the number of aperture blades. Lenses with an even number of blades, will produce as many rays as the number of blades. Lenses with an on odd number of blades, will produce twice as many rays as the number of blades. See for example here:

http://www.photocrati.com/fun-with-sunstars/

mh2000 said:

For instance, as I pointed out, one person "proved" through test that the Oly 17/2.8 had better bokeh then the P20 and then you later "proved" that it didn't. Both tests could be right and both results can be right too since they were different tests. Most things with aethetic properties are not an absolute.

Click to expand...

Who proved where that the 17/2.8 has better bokeh than the P20? Got a link?

Click to expand...

Right, that is why I stated that I was commenting on the all lenses set to f5.6... pretty much my standard aperture setting for m43 lenses.

The asterics around "proved" were meant to suggest that it was not a diffinitive usage of the word. This was a long time ago and the last time I looked the OP had removed the images, but you might be able to find the thread. More "expert" DPR threadsters agreed that in this particular instance the 17/2.8 was less harsh and more pleasing. At the time, I was new to m43 and used the example to sway me toward the cheaper 17/2.8 over the 20... I didn't have any bias toward one lens or the other.

Click to expand...

The only thread I am aware of that discusses the issue (bokeh of the 17/2.8 versus the 20) at any length is this one:

http://www.dpreview.com/forums/thread/3262251

The problem is that there are no directly comparable images in that thread. That's one reason why I performed my test of the bokeh of MFT WAs, including these two lenses but also others. Those interested find it here:

http://www.dpreview.com/forums/post/50023507

mh2000 said:

I know you have posted test shots later that don't agree with those initial comparisons I saw. But really, nothing to agrue about here. My girlfriend shoots a 20 almost exclusively and I have made extensive comparisons between the 17/2.8 and the 20, both pixel peeping and at identical print sizes, and while recognizing the 20 is in general the "better lens," still find some properties that I prefer coming from the 17 than from the 20, but the differences are really all in the noise in final analysis IMO. Stopped down, both lenses produce very good results IMO.

If you are suggesting that no test could show an instance of where the 17/2.8 produces more pleasing bokeh than the 20, then I would suggest that you do not understand bokeh and the subjectivity in assessing it in individual images.

Click to expand...

The bokeh (the character as opposed to quantity of OOF, especially background, blur) is an objective property of the lens. Consequently, no matter how you vary image content, the bokeh will be the same. The perception as well as evaluation of that bokeh can of course vary between individuals but the bokeh will be the same (at a given f-stop and amount of blur).

mh2000 said:

Back when I shot a lot more wide aperture photos and was much more concerned with bokeh, I found that more telling than one gear head's attempt at a controlled bokeh test was a spin through 100's of flickr images from a lens to try and assess the average bokeh properties that emerge. All lenses can produce good and bad bokeh and it is very sensitive to many things.

Click to expand...

On the contrary, a controlled test will be far superior, if it is done correctly. The bokeh is defined by the properties of the blur disc and a correctly done test will give you a far better idea of how the blur disc properties of one lens compares to that of another than looking at hundreds of incomparable images on flickr.