SR and Focal Length

theunartist · 04-20-2012, 03:03 PM

Does anyone know how inputting the incorrect Focal Length on Manual Lenses as per Shake Reduction impacts the Focus Accuracy?

I noticed the other day, I had a 35mm on the camera and had the SR incorrectly set on 120mm, most of the images where a bit soft and out of @ 30 shots, 2 were usable. This got me thinking about zoom lenses, would it be better to turn off SR on Zoom lenses?

Thanks!

Adam · 04-20-2012, 03:09 PM

What you're describing sounds like motion blur, and yes, you're likely to see it if you set the focal length to something longer than your lens.

RioRico · 04-20-2012, 06:29 PM

So I put a manual lens on my camera, maybe a 50mm. I tell the SR'bot that it's 100mm. The camera expects me to shake+shiver+quiver with a 100mm lens so it compensates for that. But that's over-compensation for how the 50mm moves. Result: motion blur.

It's bad enough with MF primes; MF zooms are even worse. There's an algorithm for picking a midpoint in a zoom range to tell the SR'bot. Supposedly if I tell it that my Vivitar 19-35 is 24mm, it evens-out the compensation over the lens range. But it doesn't do it well.

I can't lie to the SR'bot. Oh bother...

demp10 · 04-20-2012, 06:42 PM

For MF zoom lenses you have to change the focal length as you change zoom settings either from the menu, or by cycling the camera OFF/ON and choosing the correct value on the prompt.

It is a pain especially if you zoom around all the time for composition. Your best option other than that is to use high ISO to get appropriate shutter speed so SR is unnecessary, or use a flash if the conditions justify it.

**PPPPPP42** · 04-20-2012, 08:53 PM

Whoops, and all along I just set it to the highest focal length the zoom was capable of (I read it somewhere, maybe in the manual) always works ok for me so far and I have almost exclusively MF zooms.

sterretje · 04-20-2012, 09:50 PM

Check for a post by m42man; he did some calculations as to what the most universal choice would be for SR; I don't know if he is correct as I don't use MF zooms.

For me it's the reason to only buy MF primes and if I ever end up with an MF zoom, I will try m42man's suggestions or switch SR off.

RioRico · 04-21-2012, 12:20 AM

The algorithm for figuring what to tell the SR'bot is: (2*Min*Max)/(Min+Max) where Min and Max are the focal-length range. For a 70-210mm it's 105mm; I round it to 100. For my 19-35mm it's 25mm, rounded to 24. I suspect that the longer lens will suffer more with the FL mismatch (210:105) than with the shorter lens (35:25).

As it happens, I recently bought two MF Vivitar Series 1 zooms in PK-A mount with just those focal lengths. I've shot them a lot the last couple weeks. Using a computed SR-FL, the short lens *IS* sharper than the longer one. I can use that 24mm number with the 19-35 quite satisfactorily. But I must switch SR off, or constantly recalibrate, with the 70-210 because the 100mm number just doesn't work when zoomed out. Those shots are too mushy.

So with the 19-35 I can use the 1/FL rule to get SHARP shots at 1/25 second. But with the 70-210 I must use the 1/(5FL) rule (SR off) and shoot at 1/350-1/1000 for the same sharpness. I can use the 1/FL rule (SR off) with the 70-210 for shots of acceptable sharpness, but that's not SHARP. I keep this stuff in mind when I'm using my few MF zooms.

newarts · 04-21-2012, 12:02 PM

For small errors, shake error is proportional to the deviation from true focal length (actually, distance from lens to sensor which is close to focal length for normal photos.) If you know what focal length you'll use, that's the length to tell the camera.

If you don't know what focal length you'll use until just before you take the photo, cycle the camera of/on if you've enough time to enter the actual focal length.

If there's not enough time, RioRico's formula seems reasonable but I don't know if it is optimum. I'll try to check it out.

Dave

EDIT: RioRico's equation must be wrong - because if the minimum focal length is zero the result is zero.

I found the least squares minimum solution to the problem ( Fzoom minimizes the total squared error for random shots between Fmax and Fmin. The answer is:

Fzoom = (Fmin+Fmax)/2 ie the average.

theunartist · 04-21-2012, 01:25 PM

Thanks for all the replies!

As I said in my OP, I thought something was very wrong with my S-Tak 35/3.5, every single image looked soft on my computer screen and was only able to salvage 1 photo out of 30 or so. I was so worried, I changed out the focus screen that same day

But I think, it turned out I had the SR at 120mm and using a 35mm. I would never had thought that the SR would actually cause blurring/softness but this may explain why I sometimes find my A 35-105 inexplicably soft at times.

If it's better to have the actual FL in the SR, I will probably do that... in most cases and when I remember

theunartist · 04-21-2012, 01:29 PM

Originally posted by PPPPPP42

Whoops, and all along I just set it to the highest focal length the zoom was capable of (I read it somewhere, maybe in the manual) always works ok for me so far and I have almost exclusively MF zooms.

You've never noticed any problems? I usually set zooms right in the middle if possible and above middle if not...

Philoslothical · 04-21-2012, 03:05 PM

Originally posted by newarts

EDIT: RioRico's equation must be wrong - because if the minimum focal length is zero the result is zero.

If the focal length is zero, you should probably mount a lens.

The widest zoom people are likely to encounter is what, 10mm? (thinking the 10-20mm here) or 17-18mm for anything else.

I agree that I'm unsure of his calculation though. It sounds like the safest, but not necessarily the most accurate, depending on the lens used.

newarts · 04-21-2012, 03:13 PM

Originally posted by Philoslothical

If the focal length is zero, you should probably mount a lens.

The widest zoom people are likely to encounter is what, 10mm? (thinking the 10-20mm here) or 17-18mm for anything else.

I agree that I'm unsure of his calculation though. It sounds like the safest, but not necessarily the most accurate, depending on the lens used.

Theory very clearly says that the average focal length is the best choice to minimize error in the "least squares" sense.

Since his formula gives a result between the maximum and minimum focal lengths it'll be better than something outside those limits or nothing at all, but it won't be as good as the simple average in the "least squares" or "maximum likelihood" sense.

How does it "sound[s] the safest"?

RioRico · 04-21-2012, 03:45 PM - **1 Like**

Just to be clear: I didn't devise the algorithm. I found it here, on the forums, some time ago. I haven't tested it, except as I noted that it works better with a short than a long zoom. I don't know if that's because the short zoom (19-35) is ~2x and the long zoom (70-210) is 3x. I haven't tested it with a 35-200 (~6x) zoom. I don't plan rigorous tests. I just know that with my few MF zooms, I'd rather have SR off, or dialed to the focal length used.
___________________________________

Addendum: Yes, I stole the formula. Intellectual theft. Or maybe I just borrowed it. It would be plagiarism if I claimed it as my own, which would be silly. But I am reminded of Tom Lehrer's about the fast-fingered mathematician Lobachevskii:

PLAY-giarize
Let no one else's work e-VADE your eyes
Remember why the good lord MADE your eyes
And plagiarize, plagiarize, plagiarize
(But be certain always to call it:
Research)

newarts · 04-21-2012, 04:41 PM

Originally posted by RioRico

Addendum: Yes, I stole the formula. Intellectual theft. Or maybe I just borrowed it. It would be plagiarism if I claimed it as my own, which would be silly. But I am reminded of Tom Lehrer's about the fast-fingered mathematician Lobachevskii:

PLAY-giarize
Let no one else's work e-VADE your eyes
Remember why the good lord MADE your eyes
And plagiarize, plagiarize, plagiarize
(But be certain always to call it:
Research)

No, you didn't plagiarize; you just made fair use of an equation you saw elsewhere. Qualifying words like "I read that" would have made it the situation clearer but it is not common practice to do so. I doubt that few, if any, thought you were claiming to be the source of the equation, especially given its complexity.

I'm going to check it out in more depth; 'til then I'll stick by the average as being a good approximation since I know what it is based on.

For those interested it is based on the fact that if the camera rotates off-axis to a certain angle the image moves across the sensor a certain distance image.error where:

image.error = angle x focal.length

Note: this is the same simple equation that gives rise to the shutter.speed ~ 1/Focal.length rule. When macros are involved focal length should be multiplied by (1+magnification) - ie, the actual distance between lens and sensor.

Philoslothical · 04-21-2012, 04:46 PM

Originally posted by newarts

How does it "sound[s] the safest"?

It doesn't. I derped. Just ran a few different zoom lengths through it, and it seems too heavily biased by the minimum focal length.

(2*Min*Max)/(Min+Max)
18-55mm - 3x zoom: 27.1 (rounded, 28mm)
70-210mm - 3x zoom: 105 (rounded, 100mm)
18-250mm - 14x zoom: 33.5 (rounded, 35mm)
10-20mm - 2x zoom: 13.3 (rounded, 15mm)

(Min+Max)/2
18-55mm - 3x zoom: 45.5 (rounded, 50mm)
70-210mm - 3x zoom: 175 (rounded, 180mm)
18-250mm - 14x zoom: 143 (rounded, 150mm)
10-20mm - 2x zoom: 15mm

Really, neither method looks optimal, does it? I guess it's a good thing that (anecdotally) it doesn't matter too much, as long as it's somewhere in the range.

04-20-2012, 03:03 PM	#1
theunartist Inactive Account Join Date: Oct 2010 Location: Baltimore Posts: 2,542	SR and Focal Length Does anyone know how inputting the incorrect Focal Length on Manual Lenses as per Shake Reduction impacts the Focus Accuracy? I noticed the other day, I had a 35mm on the camera and had the SR incorrectly set on 120mm, most of the images where a bit soft and out of @ 30 shots, 2 were usable. This got me thinking about zoom lenses, would it be better to turn off SR on Zoom lenses? Thanks!

04-21-2012, 12:02 PM	#8
newarts Inactive Account Join Date: Dec 2008 Location: Ames, Iowa, USA Photos: Albums Posts: 2,965	For small errors, shake error is proportional to the deviation from true focal length (actually, distance from lens to sensor which is close to focal length for normal photos.) If you know what focal length you'll use, that's the length to tell the camera. If you don't know what focal length you'll use until just before you take the photo, cycle the camera of/on if you've enough time to enter the actual focal length. If there's not enough time, RioRico's formula seems reasonable but I don't know if it is optimum. I'll try to check it out. Dave EDIT: RioRico's equation must be wrong - because if the minimum focal length is zero the result is zero. I found the least squares minimum solution to the problem ( Fzoom minimizes the total squared error for random shots between Fmax and Fmin. The answer is: Fzoom = (Fmin+Fmax)/2 ie the average. Last edited by newarts; 04-21-2012 at 12:45 PM.

04-21-2012, 03:45 PM - 1 Like	#13
RioRico Veteran Member Join Date: Jun 2008 Location: Limbo, California Posts: 11,263	Just to be clear: I didn't devise the algorithm. I found it here, on the forums, some time ago. I haven't tested it, except as I noted that it works better with a short than a long zoom. I don't know if that's because the short zoom (19-35) is ~2x and the long zoom (70-210) is 3x. I haven't tested it with a 35-200 (~6x) zoom. I don't plan rigorous tests. I just know that with my few MF zooms, I'd rather have SR off, or dialed to the focal length used. ___________________________________ Addendum: Yes, I stole the formula. Intellectual theft. Or maybe I just borrowed it. It would be plagiarism if I claimed it as my own, which would be silly. But I am reminded of Tom Lehrer's about the fast-fingered mathematician Lobachevskii: PLAY-giarize Let no one else's work e-VADE your eyes Remember why the good lord MADE your eyes And plagiarize, plagiarize, plagiarize (But be certain always to call it: Research) Last edited by RioRico; 04-21-2012 at 03:52 PM.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Other Challenge Focus on Focal Length	jheu02	Mini-Challenges, Games, and Photo Stories	10	12-01-2010 08:11 PM
focal length	pentax k-x	Photographic Technique	9	11-22-2010 12:41 PM
Focal length input	K(s)evin	Pentax SLR Lens Discussion	6	10-27-2010 04:03 PM
OPS and Focal Length	SimonC	Troubleshooting and Beginner Help	2	08-14-2010 01:06 AM
focal length	ewig	Troubleshooting and Beginner Help	11	07-23-2010 09:32 PM

04-20-2012, 03:09 PM	#2
Adam Administrator Site Webmaster Join Date: Sep 2006 Location: Arizona Photos: Gallery \| Albums Posts: 51,597	What you're describing sounds like motion blur, and yes, you're likely to see it if you set the focal length to something longer than your lens.

04-20-2012, 06:29 PM	#3
RioRico Veteran Member Join Date: Jun 2008 Location: Limbo, California Posts: 11,263	So I put a manual lens on my camera, maybe a 50mm. I tell the SR'bot that it's 100mm. The camera expects me to shake+shiver+quiver with a 100mm lens so it compensates for that. But that's over-compensation for how the 50mm moves. Result: motion blur. It's bad enough with MF primes; MF zooms are even worse. There's an algorithm for picking a midpoint in a zoom range to tell the SR'bot. Supposedly if I tell it that my Vivitar 19-35 is 24mm, it evens-out the compensation over the lens range. But it doesn't do it well. I can't lie to the SR'bot. Oh bother...

04-20-2012, 06:42 PM	#4
demp10 Veteran Member Join Date: Jun 2011 Location: Atlanta Photos: Albums Posts: 602	For MF zoom lenses you have to change the focal length as you change zoom settings either from the menu, or by cycling the camera OFF/ON and choosing the correct value on the prompt. It is a pain especially if you zoom around all the time for composition. Your best option other than that is to use high ISO to get appropriate shutter speed so SR is unnecessary, or use a flash if the conditions justify it.

04-20-2012, 08:53 PM	#5
PPPPPP42 Site Supporter Join Date: Jan 2012 Location: Wisconsin Photos: Albums Posts: 947	Whoops, and all along I just set it to the highest focal length the zoom was capable of (I read it somewhere, maybe in the manual) always works ok for me so far and I have almost exclusively MF zooms.

04-20-2012, 09:50 PM	#6
sterretje Veteran Member Join Date: Dec 2008 Location: Roodepoort, South Africa Photos: Gallery \| Albums Posts: 3,561	Check for a post by m42man; he did some calculations as to what the most universal choice would be for SR; I don't know if he is correct as I don't use MF zooms. For me it's the reason to only buy MF primes and if I ever end up with an MF zoom, I will try m42man's suggestions or switch SR off.

04-21-2012, 12:20 AM	#7
RioRico Veteran Member Join Date: Jun 2008 Location: Limbo, California Posts: 11,263	The algorithm for figuring what to tell the SR'bot is: *(2MinMax)/(Min+Max)* where Min and Max are the focal-length range. For a 70-210mm it's 105mm; I round it to 100. For my 19-35mm it's 25mm, rounded to 24. I suspect that the longer lens will suffer more with the FL mismatch (210:105) than with the shorter lens (35:25). As it happens, I recently bought two MF Vivitar Series 1 zooms in PK-A mount with just those focal lengths. I've shot them a lot the last couple weeks. Using a computed SR-FL, the short lens IS sharper than the longer one. I can use that 24mm number with the 19-35 quite satisfactorily. But I must switch SR off, or constantly recalibrate, with the 70-210 because the 100mm number just doesn't work when zoomed out. Those shots are too mushy. So with the 19-35 I can use the 1/FL rule to get SHARP shots at 1/25 second. But with the 70-210 I must use the 1/(5FL) rule (SR off) and shoot at 1/350-1/1000 for the same sharpness. I can use the 1/FL rule (SR off) with the 70-210 for shots of acceptable sharpness, but that's not SHARP. I keep this stuff in mind when I'm using my few MF zooms.

04-21-2012, 01:25 PM	#9
theunartist Inactive Account Join Date: Oct 2010 Location: Baltimore Posts: 2,542 Original Poster	Thanks for all the replies! As I said in my OP, I thought something was very wrong with my S-Tak 35/3.5, every single image looked soft on my computer screen and was only able to salvage 1 photo out of 30 or so. I was so worried, I changed out the focus screen that same day But I think, it turned out I had the SR at 120mm and using a 35mm. I would never had thought that the SR would actually cause blurring/softness but this may explain why I sometimes find my A 35-105 inexplicably soft at times. If it's better to have the actual FL in the SR, I will probably do that... in most cases and when I remember

04-21-2012, 01:29 PM	#10
theunartist Inactive Account Join Date: Oct 2010 Location: Baltimore Posts: 2,542 Original Poster	Originally posted by PPPPPP42 Whoops, and all along I just set it to the highest focal length the zoom was capable of (I read it somewhere, maybe in the manual) always works ok for me so far and I have almost exclusively MF zooms. You've never noticed any problems? I usually set zooms right in the middle if possible and above middle if not...

04-21-2012, 03:05 PM	#11
Philoslothical Veteran Member Join Date: May 2011 Posts: 1,889	Originally posted by newarts EDIT: RioRico's equation must be wrong - because if the minimum focal length is zero the result is zero. If the focal length is zero, you should probably mount a lens. The widest zoom people are likely to encounter is what, 10mm? (thinking the 10-20mm here) or 17-18mm for anything else. I agree that I'm unsure of his calculation though. It sounds like the safest, but not necessarily the most accurate, depending on the lens used.

04-21-2012, 03:13 PM	#12
newarts Inactive Account Join Date: Dec 2008 Location: Ames, Iowa, USA Photos: Albums Posts: 2,965	Originally posted by Philoslothical If the focal length is zero, you should probably mount a lens. The widest zoom people are likely to encounter is what, 10mm? (thinking the 10-20mm here) or 17-18mm for anything else. I agree that I'm unsure of his calculation though. It sounds like the safest, but not necessarily the most accurate, depending on the lens used. Theory very clearly says that the average focal length is the best choice to minimize error in the "least squares" sense. Since his formula gives a result between the maximum and minimum focal lengths it'll be better than something outside those limits or nothing at all, but it won't be as good as the simple average in the "least squares" or "maximum likelihood" sense. How does it "sound[s] the safest"?

04-21-2012, 04:41 PM	#14
newarts Inactive Account Join Date: Dec 2008 Location: Ames, Iowa, USA Photos: Albums Posts: 2,965	Originally posted by RioRico Addendum: Yes, I stole the formula. Intellectual theft. Or maybe I just borrowed it. It would be plagiarism if I claimed it as my own, which would be silly. But I am reminded of Tom Lehrer's about the fast-fingered mathematician Lobachevskii: PLAY-giarize Let no one else's work e-VADE your eyes Remember why the good lord MADE your eyes And plagiarize, plagiarize, plagiarize (But be certain always to call it: Research) No, you didn't plagiarize; you just made fair use of an equation you saw elsewhere. Qualifying words like "I read that" would have made it the situation clearer but it is not common practice to do so. I doubt that few, if any, thought you were claiming to be the source of the equation, especially given its complexity. I'm going to check it out in more depth; 'til then I'll stick by the average as being a good approximation since I know what it is based on. For those interested it is based on the fact that if the camera rotates off-axis to a certain angle the image moves across the sensor a certain distance image.error where: image.error = angle x focal.length Note: this is the same simple equation that gives rise to the shutter.speed ~ 1/Focal.length rule. When macros are involved focal length should be multiplied by (1+magnification) - ie, the actual distance between lens and sensor.

04-21-2012, 04:46 PM	#15
Philoslothical Veteran Member Join Date: May 2011 Posts: 1,889	Originally posted by newarts How does it "sound[s] the safest"? It doesn't. I derped. Just ran a few different zoom lengths through it, and it seems too heavily biased by the minimum focal length. *(2MinMax)/(Min+Max)* 18-55mm - 3x zoom: 27.1 (rounded, 28mm) 70-210mm - 3x zoom: 105 (rounded, 100mm) 18-250mm - 14x zoom: 33.5 (rounded, 35mm) 10-20mm - 2x zoom: 13.3 (rounded, 15mm) (Min+Max)/2 18-55mm - 3x zoom: 45.5 (rounded, 50mm) 70-210mm - 3x zoom: 175 (rounded, 180mm) 18-250mm - 14x zoom: 143 (rounded, 150mm) 10-20mm - 2x zoom: 15mm Really, neither method looks optimal, does it? I guess it's a good thing that (anecdotally) it doesn't matter too much, as long as it's somewhere in the range.