Help With Older Zooms

newarts · 05-06-2012, 01:03 PM

Originally posted by PPPPPP42

Umm, ya we have way too many people that are way too smart on these boards and this happens every time, someone always breaks out the chalk board.

There is truth in mathematics. Frequency of repetition, volume, passion, "I read it somewhere", etc are no substitute for rational conclusions of mathematics, except reality of course which trumps all.

Without the chalk board my or anybody else's bold statements are likely to be taken with a grain of salt.

Quote:

Aside from all that I use mostly all MF A series lenses, and I started skimming when this thread got long winded so this mighta been said already, but if your lens has the A aperture setting leave it on that and set the aperture with the dials, if not go to the beginner Q&A forum here because I think how to use manual only aperture lenses is part of a sticky thread there.

I can buy the idea that being in a sticky is perhaps an indication of truth, but no guarantee.

Quote:

And I always set my SR to the highest a zoom goes to when I turn it on because that's where it needs the most shake reduction since every twitch is magnified. At the wide end movement isn't magnified so much so the SR doesn't really do much at all. Seems to produce good pics all across the range set this way.

I originally thought along those lines before closely analyzing the situation,

Assume that Shake reduction works ie shake error is completely compensated when the actual focal length equals the SR focal length.

If the actual focal length is longer than the SR focal length then too little correction will be applied. The error will be proportional to (actual-SR).

If the actual focal length is shorter than the SR focal length then too much correction will be applied. The error will be proportional to (actual-SR).

Note that the error is proportional to the deviation from the correct focal length, not from zero or the max focal length. That's what leads to the conclusion the mean is the best guess for SR if focal length will vary randomly within the max - min limit

m42man · 05-06-2012, 01:07 PM

Newarts, I know you mean well, but I have a feeling that you haven't read my first post on this thread. Please have a read, and follow the link, which points to a page on another thread which explains everything (if you look at my posts).

And reivax, don't worry, set the FL value to 40mm and you'll be OK. But you're right about the desirability of these vintage zooms - you won't get full SR, they're often inferior in terms of optical performance, and they're usually big and heavy. However, this means there are some bargains to be had, and some of the Pentax M-series zooms were very good indeed, also the Vivitar Series One zooms (I'm thinking the early 70-210s, though they're REALLY big) and some Tokinas too.

And PPPPPP42: don't do it, you'll get a lot of ruined shots!

newarts · 05-06-2012, 01:14 PM

Originally posted by m42man

Newarts, I know you mean well, but I have a feeling that you haven't read my first post on this thread. Please have a read, and follow the link, which points to a page on another thread which explains everything (if you look at my posts).

And reivax, don't worry, set the FL value to 40mm and you'll be OK. But you're right about the desirability of these vintage zooms - you won't get full SR, they're often inferior in terms of optical performance, and they're usually big and heavy. However, this means there are some bargains to be had, and some of the Pentax M-series zooms were very good indeed, also the Vivitar Series One zooms (I'm thinking the early 70-210s, though they're REALLY big) and some Tokinas too.

And PPPPPP42: don't do it, you'll get a lot of ruined shots!

Yes, I read the link - it said..

Originally posted by m42man

I figured out the formula a while ago, and the basis is illustrated here:...

In terms of derivation, it's fairly easy to come up with two simultaneous equations defining the amount of sensor movement error at max and min zoom settings, and you combine these to end up with a value which equalises the SR at min and max zoom. If I remember correctly, as follows:

1. Fully zoomed in: SR effectiveness = x/H
2. Fully zoomed out: SR effectiveness = (2*L - x)/L

(where x is the value you're looking for, H is max FL, L is min FL)

Combining (1) and (2), for equal SR effectiveness whether zoomed-in or out: x = 2*H*L/(H+L)

I believe those equations represent some kind of effectiveness. They are a kind of fractional error, not a measure of point's motion across the image plane. My approach directly minimizes the error in placement of the image on the sensor, based on the following:

Shake reduction is based on the rotation of a camera in modes like nodding one's head "yes" and "no". when the camera is rotated in that mode some small angle (call it alpha) an image point moves across the sensor a distance given by

Motion = alpha x actual.focal.length

when the camera's accelerometers measure motion that is going to result in alpha, the sensor is shifted a distance

Motion.correction = alpha x SR.focal.length (SR.focal.length is what the camera thinks the focal length is)

error = (Motion - motion correction) = alpha x (actual.focal.length - SR.focal.length)

This is only zero when SR.focal.length = actual.focal.length.

To minimize the mean squared error for random focal lengths between min to max, one should set SR.focal.length = (min+max)/2 - this is a well established statistical relationship.

See what pentax says... Shake Reduction Setting for Manual-Focus Zoom Lenses | Pentax Support

**PPPPPP42** · 05-06-2012, 01:15 PM

I think it really only works for me because on the wide end there is little perceived movement of the image so the camera doesn't have much if anything to over correct for, wheras the opposite would be true on the top end. Though I would agree I am doing it wrong.

This was the sticky about M lenses on DSLR's. https://www.pentaxforums.com/forums/pentax-beginners-corner-q/110658-using-ma...x-dslrs-f.html

m42man · 05-06-2012, 02:17 PM

Newarts - yes, it's obvious you'll only get zero blur when the value you input is equal to the actual focal length thet you've set up on your zoom lens. But you then say:

"To minimize the mean squared error for random focal lengths between min to max, one should set SR.focal.length = (min+max)/2 - this is a well established statistical relationship."

This is a total non-sequitur (it's nothing to do with statistics, for a start). Please re-read what I've written on the subject.

PPPPPP42: yes, you certainly need more SR on the longer focal lengths (but just to confuse the issue, there's an analysis here: Falk Lumo: Pentax shake reduction revisited). But the problem is that most people are trying to get an a nice "compromise" value which allows the best SR for whatever zoom setting selected (and using my prescribed formula you get the same SR at both ends of the zoom range). If you set the value to "max", then whenever you set the zoom lens to a focal length less than max/2, you'll end up with more blur than you would have had if you'd turned SR off.

There's an argument which would suggest that you should have more SR at the longer FLs, so you might like to bias towards the long end - but you should remember that the rewards at the long end come with dramatic shortfalls at the short end.

reivax · 05-06-2012, 03:12 PM

As a non math guy, "You lost me at hello".

These words, I remember some.

jimr-pdx · 05-06-2012, 03:23 PM

Originally posted by reivax

What do you mean by one touch style?

Ah there I go, a photo-child of the '80s using old jargon
'One-touch' zooms are those without a dedicated zoom ring. You grab the focusing ring and push or pull to zoom, twist to focus. They were The Way to Go for quite some time, but one cannot be as precise at either zoom or focus; when focusing critically it's hard not to zoom, and vice versa. I don't know if any are still made, or when the last one came out.. but the two-ring design was the clear evolutionary winner on this.

Last edited by jimr-pdx; 05-06-2012 at 03:35 PM.

newarts · 05-06-2012, 05:11 PM

Originally posted by m42man

This is a total non-sequitur (it's nothing to do with statistics, for a start). Please re-read what I've written on the subject.

It is no non-sequitur. This problem has everything to do with statistics - the goal is to minimize the SR error for a random selection of focal lengths between some max and minimum.

In the math I gave earlier I stopped before deriving the whole concept of least squared error (or maximum likelyhood). It boils down to minimizing the average of the squares of the errors. Recall that the square of each error is (angle(actual.focal.length-SR.focal.length))^2.

The "mean square error" is the (integral of the squared errors for the range) divided by the range. You might look it up. In this case or in any case where a random error is proportional to the deviation from a value, the optimum estimate of that value is the mean of the range. Also see "Central limit theorem".

You say

Originally posted by m42man

(and using my prescribed formula you get the same SR at both ends of the zoom range). If you set the value to "max", then whenever you set the zoom lens to a focal length less than max/2, you'll end up with more blur than you would have had if you'd turned SR off.

Turning off SR is the same as setting the SR focal length to zero. If you set it to the middle of the range, the errors both above and below the SR chosen are proportional to (focal.length - SR).

Say you set the SR to zero like you said, the error you get for focal lengths less than max/2 are less than if you'd used max as the SR focal length. But that doesn't mean it is the best choice. The best choice is the one that minimizes the (focal.length-SR) error; if min is not zero then SR = (max+min)/2 gives a smaller error for max/2 than if SR=0

Case 1. SR=(max+min)/2, f.actual = max/2, error is max/2- (max + min)/2 = -min/2
Case 2. SR=0, f.actual = max/2, error is max/2

Since min < max the absolute error is less when SR=(max+min)/2

Go back and look at what your formula recommends. For a 28-250mm lens (a common extended zoom range), your recommendation is:

SR.focal.length = 2*28*250/(28+250) = 50mm !!

That doesn't make sense; if you pick up a 28-250 focal length lens and randomly pick a focal length the most probable choice of focal length is (250+28)/2 = 139mm, far from 50mm. Surely you are best off using a SR focal length close to the most probable focal length used. You would not recommend that 50mm be used as the SR parameter for a 139mm fixed focal length lens, so why tell someone to use a SR parameter of 50mm for a lens that will most probably be used at 139mm?

I will not let this go; people are at best being confused, at worst, mislead.

m42man · 05-07-2012, 01:42 AM

Reivax, sorry about this, your thread seems to be getting somewhat hijacked!

Newarts, you don't seem to be able to grasp what my analysis was about: the idea is to come up with a focal length value which ends up with the same amount of SR for the zoom lens at max and min. Then if you were to operate the zoom from one end of the range to the other, you would obviously find the SR effectiveness increase, until the actual FL matches the value that you entered into the camera. Beyond that the SR effectiveness will gradually fall, until you get to the opposite end of the zoom range. At no point will the SR mechanism fail to provide some degree of effectiveness.

If you follow the link in my first post in this thread, this leads to a forum page, at the bottom of which you'll find my post (number 10) which explains my workings (and you'll need to follow the link provided). If you don't understand or disagree with the analysis, say so, and tell me which bit you have any problems with.

Perhaps the key aspect of this is that, if the value you set is more than 2x the actual FL, then the camera will apply sufficient sensor movement that the blur is magnified (but in the opposite direction to the blur you'd get with SR off - not that this fact matters). If you don't understand this, then you'll always be barking up the wrong tree. And do note that your formula will run into problems of this nature (at least for zoom ranges greater than 3x).

And no more talk about statistics - this is all about simple geometry and algebra. As I say, if you don't follow any aspect of my reasoning, just tell me.

newarts · 05-07-2012, 07:48 AM

Originally posted by m42man

....
Newarts, you don't seem to be able to grasp what my analysis was about: the idea is to come up with a focal length value which ends up with the same amount of SR for the zoom lens at max and min.

There is only one SR setting which ends up with the same amount of SR for the zoom lens at max and min. Thats when:

(SR-max) = -(SR-min) which is the same as SR=(max+min)/2 in complete agreement with your above requirement.

Further this choice of SR parameter is the one that minimizes the average error for random focal length settings between max & min.

Quote:

If you follow the link in my first post in this thread, this leads to a forum page, at the bottom of which you'll find my post (number 10) which explains my workings (and you'll need to follow the link provided). If you don't understand or disagree with the analysis, say so, and tell me which bit you have any problems with.

I did this originally, then again when I posted a copy of your analysis above. In particular, while the quantities you gave are fractional measures of SR errors they are not the only such quantities, just finding an x which satisfies both equations does not mean that is the value that minimizes the error on the image. If one uses absolute errors, one gets the result shown above.

It is the distance a dot of light moves on the image we care about, not the error as a fraction of distance from endpoints (which is what your parameters seem to be measuring.) Why should a error at the low end be given a weight 1/min, an the same error at the high end be given a weight 1/max? This is what your formulas do in effect; much more importance is given to errors at the low end than at the high end in the ratio of max/min. But why? Your original statement above requires no such relative weighting of errors.

Quote:

Perhaps the key aspect of this is that, if the value you set is more than 2x the actual FL, then the camera will apply sufficient sensor movement that the blur is magnified (but in the opposite direction to the blur you'd get with SR off - not that this fact matters). If you don't understand this, then you'll always be barking up the wrong tree.[I do understand this] And do note that your formula will run into problems of this nature (at least for zoom ranges greater than 3x).

Yes, anytime min is less than ( (max+min)/2 - min) or 3min < max, SR off is a better choice for focal.length = min. But that doesn't mean the that the mean is not the best choice for random actual focal lengths between max and min.

Quote:

And no more talk about statistics - this is all about simple geometry and algebra. As I say, if you don't follow any aspect of my reasoning, just tell me.

Statistics and calculus are necessary for showing a solution is optimal.

I don't follow your logic in dividing (1-upper.error) by max (H), ie. x/H and (1-lower.error) by min (L),ie (2L-x)/L. If these weights are not used you and I get the same result.

Dave

PS perhaps you could tell me where the fault lies in my analysis?

reivax · 05-07-2012, 06:42 PM

I've got to admit, even though I understand none of this, it's still pretty entertaining to pretend I do.

RioRico · 05-07-2012, 08:14 PM

This is one of those unresolved no-consensus issues.

There are various ways of calculating a focal length to tell the SR'bot. I am unhappy with all such calculations -- they work OK near the calculated value, an not-so-good elsewhere. This is especially true with zooms with longer focal lengths and zoom ranges. Whatever midpoint I use with a 70-210 (3x) or 60-300 (5x) MF zoom, the extremes suck. Whatever midpoint I use with a 19-35 (2x) zoom is pretty acceptable.

So if I want SHARP images, I turn off SR and shoot in bright light using the 1/(5*FL) rule. For acceptable sharpness, I use the 1/FL rule with SR off.

Are MF zooms worth using? That depends on the lens, and on our preferences. Some have absolutely gorgeous optics or other features (*) and are worth whatever labor is entailed. Most are completely outclassed by modern AF zooms, which is why of my current 235 lenses, only a dozen are MF zooms. And I'll probably off some of those soon.
___________________________________

(*) My favortie MF zooms:

* Vivitar Series 1 70-210: V1/3.5 (Komine) and V3/2.8-4 (Kiron)
* Vivitar (Cosina) Series 1 19-35/3.5-4.5 (also in AF)
* Schneider Betavaron 50-125/4-5.6 enlarger zoom
* Sears (Tokina) 55-135/3.5, wizard for portraits

Some are tolerable and I'll probably keep them because they're A-type:

* Sears (Samyang) 80-200 and 70-200, both f/4
* A35-80/4-5.6, but ONLY reversed for macros
* Promaster (Tamron) 60-300/4-5.6
* Tokina SD 35-200/4-5.6

My other non-A-types just aren't worth bothering with IMHO. Your mileage may vary.

newarts · 05-07-2012, 09:04 PM

Originally posted by RioRico

This is one of those unresolved no-consensus issues.

There are various ways of calculating a focal length to tell the SR'bot. I am unhappy with all such calculations -- they work OK near the calculated value, an not-so-good elsewhere. This is especially true with zooms with longer focal lengths and zoom ranges. Whatever midpoint I use with a 70-210 (3x) or 60-300 (5x) MF zoom, the extremes suck. Whatever midpoint I use with a 19-35 (2x) zoom is pretty acceptable.

There is a theoretically (and practically) best SR setting; the one in the middle of the range you'll be using - if you will be using less than the full range of a zoom, input the middle of the range you expect to be using. The unfortunate part is that not-to-far away from that setting, things get bad pretty fast.

Simple theory says SR error increases in proportion to the difference between actual focal length and SR setting.

As you say, the extremes suck and the farther the extremes are from the SR parameter, the more they suck.

Quote:

So if I want SHARP images, I turn off SR and shoot in bright light using the 1/(5*FL) rule. For acceptable sharpness, I use the 1/FL rule with SR off.

Wow, I wish I were that steady... I find SR helps me a lot for low light & long lens hand-held photos.... also, if I'm concerned about shake, I'll cycle the camera off-on so I can input the middle of the focal length range I'm going to use.

Dave

Dave

reivax · 05-07-2012, 11:20 PM

Originally posted by RioRico

* Vivitar Series 1 70-210: V3/2.8-4 (Kiron)

I keep hearing great things about this lens. If you don't mind me asking, what's your favorite use for this lens? Good for portraits? Video?

RioRico · 05-07-2012, 11:54 PM

Originally posted by reivax

Originally posted by RioRico:

* Vivitar Series 1 70-210: V3/2.8-4 (Kiron)

I keep hearing great things about this lens. If you don't mind me asking, what's your favorite use for this lens? Good for portraits? Video?

The Series 1 70-210 was built in at least 4 different versions -- read about them here: http://www.robertstech.com/vivitar.htm.

The V1 is a legendary NASA design; mine is M42 and lives on an advanced M42 camera, a perfect longer zoom from portraiture to action to near-macro closeness.

My V3 in PK-A mount arrived just a few weeks ago whilst I was on the road and hasn't yet been exercised much. Due to technical constraints (my travel computer kept crashing), I haven't even examined its shots. I need to do some shootouts vs my other 70-210 and 80-200 MF zooms. I can't really report on it yet; sorry.

05-06-2012, 01:14 PM	#18
newarts Inactive Account Join Date: Dec 2008 Location: Ames, Iowa, USA Photos: Albums Posts: 2,965	Originally posted by m42man Newarts, I know you mean well, but I have a feeling that you haven't read my first post on this thread. Please have a read, and follow the link, which points to a page on another thread which explains everything (if you look at my posts). And reivax, don't worry, set the FL value to 40mm and you'll be OK. But you're right about the desirability of these vintage zooms - you won't get full SR, they're often inferior in terms of optical performance, and they're usually big and heavy. However, this means there are some bargains to be had, and some of the Pentax M-series zooms were very good indeed, also the Vivitar Series One zooms (I'm thinking the early 70-210s, though they're REALLY big) and some Tokinas too. And PPPPPP42: don't do it, you'll get a lot of ruined shots! Yes, I read the link - it said.. Originally posted by m42man I figured out the formula a while ago, and the basis is illustrated here:... In terms of derivation, it's fairly easy to come up with two simultaneous equations defining the amount of sensor movement error at max and min zoom settings, and you combine these to end up with a value which equalises the SR at min and max zoom. If I remember correctly, as follows: 1. Fully zoomed in: SR effectiveness = x/H 2. Fully zoomed out: SR effectiveness = (2L - x)/L (where x is the value you're looking for, H is max FL, L is min FL) Combining (1) and (2), for equal SR effectiveness whether zoomed-in or out: x = 2HL/(H+L) I believe those equations represent some kind of effectiveness. They are a kind of fractional error, not a measure of point's motion across the image plane. My approach directly minimizes the error in placement of the image on the sensor, based on the following: Shake reduction is based on the rotation of a camera in modes like nodding one's head "yes" and "no". when the camera is rotated in that mode some small angle (call it alpha) an image point moves across the sensor a distance given by Motion = alpha x actual.focal.length when the camera's accelerometers measure motion that is going to result in alpha, the sensor is shifted a distance Motion.correction = alpha x SR.focal.length (SR.focal.length is what the camera thinks the focal length is)* error = (Motion - motion correction) = alpha x (actual.focal.length - SR.focal.length) This is only zero when SR.focal.length = actual.focal.length. To minimize the mean squared error for random focal lengths between min to max, one should set SR.focal.length = (min+max)/2 - this is a well established statistical relationship. See what pentax says... Shake Reduction Setting for Manual-Focus Zoom Lenses \| Pentax Support Last edited by newarts; 05-06-2012 at 01:48 PM.

05-07-2012, 07:48 AM	#25
newarts Inactive Account Join Date: Dec 2008 Location: Ames, Iowa, USA Photos: Albums Posts: 2,965	Originally posted by m42man .... Newarts, you don't seem to be able to grasp what my analysis was about: the idea is to come up with a focal length value which ends up with the same amount of SR for the zoom lens at max and min. There is only one SR setting which ends up with the same amount of SR for the zoom lens at max and min. Thats when: (SR-max) = -(SR-min) which is the same as SR=(max+min)/2 in complete agreement with your above requirement. Further this choice of SR parameter is the one that minimizes the average error for random focal length settings between max & min. Quote: If you follow the link in my first post in this thread, this leads to a forum page, at the bottom of which you'll find my post (number 10) which explains my workings (and you'll need to follow the link provided). If you don't understand or disagree with the analysis, say so, and tell me which bit you have any problems with. I did this originally, then again when I posted a copy of your analysis above. In particular, while the quantities you gave are fractional measures of SR errors they are not the only such quantities, just finding an x which satisfies both equations does not mean that is the value that minimizes the error on the image. If one uses absolute errors, one gets the result shown above. It is the distance a dot of light moves on the image we care about, not the error as a fraction of distance from endpoints (which is what your parameters seem to be measuring.) Why should a error at the low end be given a weight 1/min, an the same error at the high end be given a weight 1/max? This is what your formulas do in effect; much more importance is given to errors at the low end than at the high end in the ratio of max/min. But why? Your original statement above requires no such relative weighting of errors. Quote: Perhaps the key aspect of this is that, if the value you set is more than 2x the actual FL, then the camera will apply sufficient sensor movement that the blur is magnified (but in the opposite direction to the blur you'd get with SR off - not that this fact matters). If you don't understand this, then you'll always be barking up the wrong tree.[I do understand this] And do note that your formula will run into problems of this nature (at least for zoom ranges greater than 3x). Yes, anytime min is less than ( (max+min)/2 - min) or 3min < max, SR off is a better choice for focal.length = min. But that doesn't mean the that the mean is not the best choice for random actual focal lengths between max and min. Quote: And no more talk about statistics - this is all about simple geometry and algebra. As I say, if you don't follow any aspect of my reasoning, just tell me. Statistics and calculus are necessary for showing a solution is optimal. I don't follow your logic in dividing (1-upper.error) by max (H), ie. x/H and (1-lower.error) by min (L),ie (2L-x)/L. If these weights are not used you and I get the same result. Dave PS perhaps you could tell me where the fault lies in my analysis? Last edited by newarts; 05-07-2012 at 08:43 AM.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
zooms	Dudley	Troubleshooting and Beginner Help	13	07-15-2011 07:38 AM
DA15mm vs 16-XX zooms?	Mr_Canuck	Pentax SLR Lens Discussion	2	03-15-2011 03:57 PM
Travel Zooms	NRM7000	Pentax SLR Lens Discussion	7	09-19-2010 10:05 AM
shake reduction and older zooms	NaClH2O	Pentax DSLR Discussion	15	12-04-2007 11:33 PM

05-06-2012, 01:03 PM	#16
newarts Inactive Account Join Date: Dec 2008 Location: Ames, Iowa, USA Photos: Albums Posts: 2,965	Originally posted by PPPPPP42 Umm, ya we have way too many people that are way too smart on these boards and this happens every time, someone always breaks out the chalk board. There is truth in mathematics. Frequency of repetition, volume, passion, "I read it somewhere", etc are no substitute for rational conclusions of mathematics, except reality of course which trumps all. Without the chalk board my or anybody else's bold statements are likely to be taken with a grain of salt. Quote: Aside from all that I use mostly all MF A series lenses, and I started skimming when this thread got long winded so this mighta been said already, but if your lens has the A aperture setting leave it on that and set the aperture with the dials, if not go to the beginner Q&A forum here because I think how to use manual only aperture lenses is part of a sticky thread there. I can buy the idea that being in a sticky is perhaps an indication of truth, but no guarantee. Quote: And I always set my SR to the highest a zoom goes to when I turn it on because that's where it needs the most shake reduction since every twitch is magnified. At the wide end movement isn't magnified so much so the SR doesn't really do much at all. Seems to produce good pics all across the range set this way. I originally thought along those lines before closely analyzing the situation, Assume that Shake reduction works ie shake error is completely compensated when the actual focal length equals the SR focal length. If the actual focal length is longer than the SR focal length then too little correction will be applied. The error will be proportional to (actual-SR). If the actual focal length is shorter than the SR focal length then too much correction will be applied. The error will be proportional to (actual-SR). Note that the error is proportional to the deviation from the correct focal length, not from zero or the max focal length. That's what leads to the conclusion the mean is the best guess for SR if focal length will vary randomly within the max - min limit

05-06-2012, 01:07 PM	#17
m42man Veteran Member Join Date: Jan 2010 Location: Southern England Posts: 623	Newarts, I know you mean well, but I have a feeling that you haven't read my first post on this thread. Please have a read, and follow the link, which points to a page on another thread which explains everything (if you look at my posts). And reivax, don't worry, set the FL value to 40mm and you'll be OK. But you're right about the desirability of these vintage zooms - you won't get full SR, they're often inferior in terms of optical performance, and they're usually big and heavy. However, this means there are some bargains to be had, and some of the Pentax M-series zooms were very good indeed, also the Vivitar Series One zooms (I'm thinking the early 70-210s, though they're REALLY big) and some Tokinas too. And PPPPPP42: don't do it, you'll get a lot of ruined shots!

05-06-2012, 01:15 PM	#19
PPPPPP42 Site Supporter Join Date: Jan 2012 Location: Wisconsin Photos: Albums Posts: 947	I think it really only works for me because on the wide end there is little perceived movement of the image so the camera doesn't have much if anything to over correct for, wheras the opposite would be true on the top end. Though I would agree I am doing it wrong. This was the sticky about M lenses on DSLR's. https://www.pentaxforums.com/forums/pentax-beginners-corner-q/110658-using-ma...x-dslrs-f.html

05-06-2012, 02:17 PM	#20
m42man Veteran Member Join Date: Jan 2010 Location: Southern England Posts: 623	Newarts - yes, it's obvious you'll only get zero blur when the value you input is equal to the actual focal length thet you've set up on your zoom lens. But you then say: "To minimize the mean squared error for random focal lengths between min to max, one should set SR.focal.length = (min+max)/2 - this is a well established statistical relationship." This is a total non-sequitur (it's nothing to do with statistics, for a start). Please re-read what I've written on the subject. PPPPPP42: yes, you certainly need more SR on the longer focal lengths (but just to confuse the issue, there's an analysis here: Falk Lumo: Pentax shake reduction revisited). But the problem is that most people are trying to get an a nice "compromise" value which allows the best SR for whatever zoom setting selected (and using my prescribed formula you get the same SR at both ends of the zoom range). If you set the value to "max", then whenever you set the zoom lens to a focal length less than max/2, you'll end up with more blur than you would have had if you'd turned SR off. There's an argument which would suggest that you should have more SR at the longer FLs, so you might like to bias towards the long end - but you should remember that the rewards at the long end come with dramatic shortfalls at the short end.

05-06-2012, 03:12 PM	#21
reivax Veteran Member Join Date: Jun 2011 Location: California Posts: 826 Original Poster	As a non math guy, "You lost me at hello". These words, I remember some.

05-06-2012, 03:23 PM	#22
jimr-pdx Pentaxian Join Date: May 2010 Location: now 1 hour north of PDX Photos: Gallery \| Albums Posts: 3,897	Originally posted by reivax What do you mean by one touch style? Ah there I go, a photo-child of the '80s using old jargon 'One-touch' zooms are those without a dedicated zoom ring. You grab the focusing ring and push or pull to zoom, twist to focus. They were The Way to Go for quite some time, but one cannot be as precise at either zoom or focus; when focusing critically it's hard not to zoom, and vice versa. I don't know if any are still made, or when the last one came out.. but the two-ring design was the clear evolutionary winner on this. Last edited by jimr-pdx; 05-06-2012 at 03:35 PM.

05-06-2012, 05:11 PM	#23
newarts Inactive Account Join Date: Dec 2008 Location: Ames, Iowa, USA Photos: Albums Posts: 2,965	Originally posted by m42man This is a total non-sequitur (it's nothing to do with statistics, for a start). Please re-read what I've written on the subject. It is no non-sequitur. This problem has everything to do with statistics - the goal is to minimize the SR error for a random selection of focal lengths between some max and minimum. In the math I gave earlier I stopped before deriving the whole concept of least squared error (or maximum likelyhood). It boils down to minimizing the average of the squares of the errors. Recall that the square of each error is (angle(actual.focal.length-SR.focal.length))^2. The "mean square error" is the (integral of the squared errors for the range) divided by the range. You might look it up. In this case or in any case where a random error is proportional to the deviation from a value, the optimum estimate of that value is the mean of the range. Also see "Central limit theorem". You say Originally posted by m42man (and using my prescribed formula you get the same SR at both ends of the zoom range). If you set the value to "max", then whenever you set the zoom lens to a focal length less than max/2, you'll end up with more blur than you would have had if you'd turned SR off. Turning off SR is the same as setting the SR focal length to zero. If you set it to the middle of the range, the errors both above and below the SR chosen are proportional to (focal.length - SR). Say you set the SR to zero like you said, the error you get for focal lengths less than max/2 are less than if you'd used max as the SR focal length. But that doesn't mean it is the best choice. The best choice is the one that minimizes the (focal.length-SR) error; if min is not zero then SR = (max+min)/2 gives a smaller error for max/2 than if SR=0 Case 1. SR=(max+min)/2, f.actual = max/2, error is max/2- (max + min)/2 = -min/2 Case 2. SR=0, f.actual = max/2, error is max/2 Since min < max the absolute error is less when SR=(max+min)/2 Go back and look at what your formula recommends. For a 28-250mm lens (a common extended zoom range), your recommendation is: SR.focal.length = 228250/(28+250) = 50mm !! That doesn't make sense; if you pick up a 28-250 focal length lens and randomly pick a focal length the most probable choice of focal length is (250+28)/2 = 139mm, far from 50mm. Surely you are best off using a SR focal length close to the most probable focal length used. You would not recommend that 50mm be used as the SR parameter for a 139mm fixed focal length lens, so why tell someone to use a SR parameter of 50mm for a lens that will most probably be used at 139mm? I will not let this go; people are at best being confused, at worst, mislead.

05-07-2012, 01:42 AM	#24
m42man Veteran Member Join Date: Jan 2010 Location: Southern England Posts: 623	Reivax, sorry about this, your thread seems to be getting somewhat hijacked! Newarts, you don't seem to be able to grasp what my analysis was about: the idea is to come up with a focal length value which ends up with the same amount of SR for the zoom lens at max and min. Then if you were to operate the zoom from one end of the range to the other, you would obviously find the SR effectiveness increase, until the actual FL matches the value that you entered into the camera. Beyond that the SR effectiveness will gradually fall, until you get to the opposite end of the zoom range. At no point will the SR mechanism fail to provide some degree of effectiveness. If you follow the link in my first post in this thread, this leads to a forum page, at the bottom of which you'll find my post (number 10) which explains my workings (and you'll need to follow the link provided). If you don't understand or disagree with the analysis, say so, and tell me which bit you have any problems with. Perhaps the key aspect of this is that, if the value you set is more than 2x the actual FL, then the camera will apply sufficient sensor movement that the blur is magnified (but in the opposite direction to the blur you'd get with SR off - not that this fact matters). If you don't understand this, then you'll always be barking up the wrong tree. And do note that your formula will run into problems of this nature (at least for zoom ranges greater than 3x). And no more talk about statistics - this is all about simple geometry and algebra. As I say, if you don't follow any aspect of my reasoning, just tell me.

05-07-2012, 06:42 PM	#26
reivax Veteran Member Join Date: Jun 2011 Location: California Posts: 826 Original Poster	I've got to admit, even though I understand none of this, it's still pretty entertaining to pretend I do.

05-07-2012, 08:14 PM	#27
RioRico Veteran Member Join Date: Jun 2008 Location: Limbo, California Posts: 11,263	This is one of those unresolved no-consensus issues. There are various ways of calculating a focal length to tell the SR'bot. I am unhappy with all such calculations -- they work OK near the calculated value, an not-so-good elsewhere. This is especially true with zooms with longer focal lengths and zoom ranges. Whatever midpoint I use with a 70-210 (3x) or 60-300 (5x) MF zoom, the extremes suck. Whatever midpoint I use with a 19-35 (2x) zoom is pretty acceptable. So if I want SHARP images, I turn off SR and shoot in bright light using the 1/(5FL) rule. For acceptable sharpness, I use the 1/FL rule with SR off. Are MF zooms worth using? That depends on the lens, and on our preferences. Some have absolutely gorgeous optics or other features () and are worth whatever labor is entailed. Most are completely outclassed by modern AF zooms, which is why of my current 235 lenses, only a dozen are MF zooms. And I'll probably off some of those soon. ___________________________________ () My favortie MF zooms: Vivitar Series 1 70-210: V1/3.5 (Komine) and V3/2.8-4 (Kiron) * Vivitar (Cosina) Series 1 19-35/3.5-4.5 (also in AF) * Schneider Betavaron 50-125/4-5.6 enlarger zoom * Sears (Tokina) 55-135/3.5, wizard for portraits Some are tolerable and I'll probably keep them because they're A-type: * Sears (Samyang) 80-200 and 70-200, both f/4 * A35-80/4-5.6, but ONLY reversed for macros * Promaster (Tamron) 60-300/4-5.6 * Tokina SD 35-200/4-5.6 My other non-A-types just aren't worth bothering with IMHO. Your mileage may vary.

05-07-2012, 09:04 PM	#28
newarts Inactive Account Join Date: Dec 2008 Location: Ames, Iowa, USA Photos: Albums Posts: 2,965	Originally posted by RioRico This is one of those unresolved no-consensus issues. There are various ways of calculating a focal length to tell the SR'bot. I am unhappy with all such calculations -- they work OK near the calculated value, an not-so-good elsewhere. This is especially true with zooms with longer focal lengths and zoom ranges. Whatever midpoint I use with a 70-210 (3x) or 60-300 (5x) MF zoom, the extremes suck. Whatever midpoint I use with a 19-35 (2x) zoom is pretty acceptable. There is a theoretically (and practically) best SR setting; the one in the middle of the range you'll be using - if you will be using less than the full range of a zoom, input the middle of the range you expect to be using. The unfortunate part is that not-to-far away from that setting, things get bad pretty fast. Simple theory says SR error increases in proportion to the difference between actual focal length and SR setting. As you say, the extremes suck and the farther the extremes are from the SR parameter, the more they suck. Quote: So if I want SHARP images, I turn off SR and shoot in bright light using the 1/(5*FL) rule. For acceptable sharpness, I use the 1/FL rule with SR off. Wow, I wish I were that steady... I find SR helps me a lot for low light & long lens hand-held photos.... also, if I'm concerned about shake, I'll cycle the camera off-on so I can input the middle of the focal length range I'm going to use. Dave Dave

05-07-2012, 11:20 PM	#29
reivax Veteran Member Join Date: Jun 2011 Location: California Posts: 826 Original Poster	Originally posted by RioRico * Vivitar Series 1 70-210: V3/2.8-4 (Kiron) I keep hearing great things about this lens. If you don't mind me asking, what's your favorite use for this lens? Good for portraits? Video?

05-07-2012, 11:54 PM	#30
RioRico Veteran Member Join Date: Jun 2008 Location: Limbo, California Posts: 11,263	Originally posted by reivax Originally posted by RioRico: * Vivitar Series 1 70-210: V3/2.8-4 (Kiron) I keep hearing great things about this lens. If you don't mind me asking, what's your favorite use for this lens? Good for portraits? Video? The Series 1 70-210 was built in at least 4 different versions -- read about them here: http://www.robertstech.com/vivitar.htm. The V1 is a legendary NASA design; mine is M42 and lives on an advanced M42 camera, a perfect longer zoom from portraiture to action to near-macro closeness. My V3 in PK-A mount arrived just a few weeks ago whilst I was on the road and hasn't yet been exercised much. Due to technical constraints (my travel computer kept crashing), I haven't even examined its shots. I need to do some shootouts vs my other 70-210 and 80-200 MF zooms. I can't really report on it yet; sorry.