Doug Kerr
Well-known member
Abstract and introduction
The Sony STF 135mm f/2.8 [T4.5] lens (originated by Minolta) has a number of design features intended to produce what is considered "especially nice bokeh", that being a term we use to mean the artistic properties of the blurring of out-of-focus foreground or background portions of an image.
These design features include two overt ingredients not found in "ordinary" lenses:
• An "apodization element" located near the aperture stops, which is a filter whose density gradually increases as we move out from the center of the lens.
• Two aperture stops, one electrically operated and one manually operated. The latter has a scrupulously-circular boundary.
In this note I will discuss these two special ingredients.
The apodization element
What does the name mean?
The use of the term apodization for the action of the radially-tapered density filter in this Minolta/Sony lens is curious - rather too cute
In fact, the conventional aperture stop is every bit as much an apodization filter as the tapered-density filter in this fancy lens.
In the realm where the term was first used (digital signal processing) an apodization filter is formally defined as one in which the response is zero outside some finite range. The term apodization (really too cute even there) etymologically means "removing the foot" (a= without, pod = foot). So it it fairly apt for a function that is zero below some point (perhaps a high-pass filter).
But the concept was stretched to also embrace filters whose function drops to zero above some point. (A more apt term there might be acapitation - removing the head.)
In any case, in digital signal processing, we often take the series of samples that describe a "waveform" and drop those outside a certain time range, by applying a windowing, or apodization, function. A simplistic one (fully meeting the definition of an apodization function) just suddenly drops the response to zero at the limits. But it is often advantageous to "taper" the response in some way as we approach the limits. (Doing so minimizes certain artifacts in the waveform implied by the train of samples.)
Somehow, in some quarters, it became the custom to speak only of an apodization filter with a non-trivial, "tapering" response as an apodization filter. Those with an abrupt drop in response at the limits were called "brick wall" windowing filters or "top-hat" windowing filters (a metaphor for the plot of their response, looking like the profile of a top hat). They are actually "brick wall" or "top-hat" apodization filters.
Now in a lens, we can consider the conventional aperture stop to be a windowing, or, equally-aptly, apodization, filter in that its response falls to zero at a certain (finite) distance from the center. It is a "top hat apodization filter". But of course we have no practice of calling it that.
In this Minolta/Sony lens, the transmission function of the new ingredient also drops to zero at a finite distance, but gradually. Thus, following the "overly cute" convention of some signal-processing authors, this particular kind of apodization filter becomes called an - apodization filter.
It's a wonderful language: "cute".
What does this filter do?
The nature of the bokeh in an image with significant out-of-focus elements is primarily influenced by these:
• The size of the blur figure formed from each point of an out-of-focus object.
• The shape of the "boundary" of the figure (and we realize that in generally, it really doesn't have a distinct boundary.
• The distribution of luminance across the figure.
And we must be concerned with the possible variation of these across the field (the blur figure for a point far from the center of the field may be quite different than for a point near the center).
If we considered an "ideal lens", with an aperture stop with a truly-circular boundary) we would find that its blur figure would be a circular disk with a sudden boundary and uniform luminance within the boundary.
Doe this give "really nice" bokeh? Well, since this is a very subjective matter at best, and since there are so many bokeh situations (are we speaking of blurred tall grass in the distance, or the bold geometric circles caused by gravely out-of-focus Christmas lights?), there is no easy answer.
In fact in real lenses, we may depart substantially from what I just described. The overall outline of the blur figure may be overtly polygonal, the luminance may increase as we move out from the center of the figure and then drop fairly quickly, and as we move from the center of the field, the overall shape of the figure may depart considerably from "circular".
In any case, many experts consider that the following attributes of the figure are desirable from a bokeh standpoint:
• All the following uniform for points at different locations in the field.
• Gross shape circular (not, for example elliptical or teardrop-shaped)
• Boundary exactly circular (not "polygonal")
• Luminance declining smoothly as we go from the center of the figure, perhaps following a "quasi-Gaussian" profile (at least to a certain radius)
It is the attainment of the last of these that is the job of the apodization element in the lenses of interest.
Without that element, the distribution of luminance might be approximately uniform.
But the apodization "filter" progressively attenuates the "hollow cones" of light passing through zones of the lens at progressively greater distances from the axis. In the creation of an out-of-focus bur figure, those successive hollow cones produce little circles (we hope) of light of successively-increasing diameter. Thus the luminance drops as we move from the center of the figure.
By properly tailoring the change in attenuation of the filter as we move from its center, we can tailor the distribution of luminance across the blur figure to be what has been decided (by Minolta) is optimal from an overall bokeh standpoint.
How is the filter made?
This figure shows the makeup of the lens:
The two-element group in the center is the apodization filter. (It is just behind the two aperture stop diaphragms, not shown in this figure.) The element shown shaded is made of "neutral tinted" glass. As we move out from the center, an increasing thickness of this is encountered, decreasing the lens transmission for rays passing there.
But the whole group is just a plane element, so it has no overall refractive effect.
The penalty
The penalty of course is that, especially for larger apertures (where the "darker" portions of the apodization filter are in play), the transmission of the lens becomes much less than in "ordinary" lenses. In fact, for the Sony STF 135mm f/2.8 [T4.5] lens, when set to an aperture of f/2.8, the effect on exposure would be as if a lens with 100% transmission were set to f/4.5. (That is the significance of the "T-stop" rating, T4.5.)
[To be continued. In Part 2: The dual aperture stop system.]
Time soon for breakfast.
Best regards,
Doug
The Sony STF 135mm f/2.8 [T4.5] lens (originated by Minolta) has a number of design features intended to produce what is considered "especially nice bokeh", that being a term we use to mean the artistic properties of the blurring of out-of-focus foreground or background portions of an image.
These design features include two overt ingredients not found in "ordinary" lenses:
• An "apodization element" located near the aperture stops, which is a filter whose density gradually increases as we move out from the center of the lens.
• Two aperture stops, one electrically operated and one manually operated. The latter has a scrupulously-circular boundary.
In this note I will discuss these two special ingredients.
The apodization element
What does the name mean?
The use of the term apodization for the action of the radially-tapered density filter in this Minolta/Sony lens is curious - rather too cute
In fact, the conventional aperture stop is every bit as much an apodization filter as the tapered-density filter in this fancy lens.
In the realm where the term was first used (digital signal processing) an apodization filter is formally defined as one in which the response is zero outside some finite range. The term apodization (really too cute even there) etymologically means "removing the foot" (a= without, pod = foot). So it it fairly apt for a function that is zero below some point (perhaps a high-pass filter).
But the concept was stretched to also embrace filters whose function drops to zero above some point. (A more apt term there might be acapitation - removing the head.)
In any case, in digital signal processing, we often take the series of samples that describe a "waveform" and drop those outside a certain time range, by applying a windowing, or apodization, function. A simplistic one (fully meeting the definition of an apodization function) just suddenly drops the response to zero at the limits. But it is often advantageous to "taper" the response in some way as we approach the limits. (Doing so minimizes certain artifacts in the waveform implied by the train of samples.)
Somehow, in some quarters, it became the custom to speak only of an apodization filter with a non-trivial, "tapering" response as an apodization filter. Those with an abrupt drop in response at the limits were called "brick wall" windowing filters or "top-hat" windowing filters (a metaphor for the plot of their response, looking like the profile of a top hat). They are actually "brick wall" or "top-hat" apodization filters.
Now in a lens, we can consider the conventional aperture stop to be a windowing, or, equally-aptly, apodization, filter in that its response falls to zero at a certain (finite) distance from the center. It is a "top hat apodization filter". But of course we have no practice of calling it that.
In this Minolta/Sony lens, the transmission function of the new ingredient also drops to zero at a finite distance, but gradually. Thus, following the "overly cute" convention of some signal-processing authors, this particular kind of apodization filter becomes called an - apodization filter.
It's a wonderful language: "cute".
What does this filter do?
The nature of the bokeh in an image with significant out-of-focus elements is primarily influenced by these:
• The size of the blur figure formed from each point of an out-of-focus object.
• The shape of the "boundary" of the figure (and we realize that in generally, it really doesn't have a distinct boundary.
• The distribution of luminance across the figure.
And we must be concerned with the possible variation of these across the field (the blur figure for a point far from the center of the field may be quite different than for a point near the center).
If we considered an "ideal lens", with an aperture stop with a truly-circular boundary) we would find that its blur figure would be a circular disk with a sudden boundary and uniform luminance within the boundary.
Doe this give "really nice" bokeh? Well, since this is a very subjective matter at best, and since there are so many bokeh situations (are we speaking of blurred tall grass in the distance, or the bold geometric circles caused by gravely out-of-focus Christmas lights?), there is no easy answer.
In fact in real lenses, we may depart substantially from what I just described. The overall outline of the blur figure may be overtly polygonal, the luminance may increase as we move out from the center of the figure and then drop fairly quickly, and as we move from the center of the field, the overall shape of the figure may depart considerably from "circular".
In any case, many experts consider that the following attributes of the figure are desirable from a bokeh standpoint:
• All the following uniform for points at different locations in the field.
• Gross shape circular (not, for example elliptical or teardrop-shaped)
• Boundary exactly circular (not "polygonal")
• Luminance declining smoothly as we go from the center of the figure, perhaps following a "quasi-Gaussian" profile (at least to a certain radius)
It is the attainment of the last of these that is the job of the apodization element in the lenses of interest.
Without that element, the distribution of luminance might be approximately uniform.
But the apodization "filter" progressively attenuates the "hollow cones" of light passing through zones of the lens at progressively greater distances from the axis. In the creation of an out-of-focus bur figure, those successive hollow cones produce little circles (we hope) of light of successively-increasing diameter. Thus the luminance drops as we move from the center of the figure.
By properly tailoring the change in attenuation of the filter as we move from its center, we can tailor the distribution of luminance across the blur figure to be what has been decided (by Minolta) is optimal from an overall bokeh standpoint.
How is the filter made?
This figure shows the makeup of the lens:
The two-element group in the center is the apodization filter. (It is just behind the two aperture stop diaphragms, not shown in this figure.) The element shown shaded is made of "neutral tinted" glass. As we move out from the center, an increasing thickness of this is encountered, decreasing the lens transmission for rays passing there.
But the whole group is just a plane element, so it has no overall refractive effect.
The penalty
The penalty of course is that, especially for larger apertures (where the "darker" portions of the apodization filter are in play), the transmission of the lens becomes much less than in "ordinary" lenses. In fact, for the Sony STF 135mm f/2.8 [T4.5] lens, when set to an aperture of f/2.8, the effect on exposure would be as if a lens with 100% transmission were set to f/4.5. (That is the significance of the "T-stop" rating, T4.5.)
Note that the typical conventional lens at an aperture of f/2.8 is not T2.8. If its transmission were 95%, it would be about T3.1
[To be continued. In Part 2: The dual aperture stop system.]
Time soon for breakfast.
Best regards,
Doug
penphotographyforums.com" in google, I find this: