Shutter times in relation to what the eye can see

[JMC1969]

I was not particularly sure where to post this thread so hopefully the people interested will find it here.

Recently I have been trying to sell a couple of Minolta x-570 camera bodies here on APUG. Obviously the question came up as to their condition so I put them to the test of the shutter tester. Which, I had not had the chance to really sit down and figure out how to use. Now I have and have some question as to what is expectable when shutter speeds are not right on the money. Funny enough, these two camera bodies are almost dead on. I thought this was just about impossible, so I got out my old standby camera I have been using for years and never had a problem with. Canon F-1. This camera's times seem way off and I have never had a bad exposure due to shutter malfunction. I assume the big difference here is, the Minolta's are electronically fired and The Canon is mechanically fired, being that the Minolta needs the battery to trip shutter and the Canon does not.

The up hill battle begins down the mathematical road in trying to figure out exactly what the time differences mean in an F-stop scenario.

This is the chart I am using as a reference for correct shutter speeds. These times and tests are based on the camera mounted without a lens so there is no iris to calculate for, just straight speed of curtain travel.

Next is the test I ran on the Canon F-1. Four (4) consecutive shutter trips, then averaged. The percentage you see to the far right is the percentage of difference from the correct speed the shutter should be traveling.

In my previous job, I dealt with digital capturing of art work for reproduction. Correct color is a must as to match the original piece. Through this, I learned quite a bit about color and how it is translated to "computer talk". This, including "Delta E". Not to get too into this subject, but simply for where I am going with this. One (1) measurement of Delta E is said to be the equivalent of one (1) shade difference in color that the human eye can differentiate between.

So in relevance to this subject, does it apply to density?

What percentage of correct shutter speed can the human eye detect?
What does that translate into as fraction of F-stops?

Can you actually see the difference in a shutter that is 25% (from perfect) or a 1/4 stop (either way) different? Is it 20%/ 1/5 stop? 10%/ 1/10 stop?

I have searched high and low to find the answer over the past few days. Maybe I am not posing the question correctly in my searches. I have re-familiarized myself with LUX, Lumens, Footcandles and still can not really come up with a definitive answer.

Any help/discussion is greatly appreciated.

Jody

 

[nick mulder]

We learned about this stuff as 'JND's':

http://en.wikipedia.org/wiki/Just-noticeable_difference (as dull as the article may be it's fascinating stuff huh)

stops don't work like that by the way - log base2 calculators:
http://www.tutorvista.com/math/logarithm-base-2-calculator

Are you talking about seeing the result of a shutter difference in a print/neg or by looking/hearing the shutter itself ? Assuming the neg and everything else equal (i.e. a mental exercise) then, um, yeh no idea on the number - but yes, there is a number... Opinions will vary on its value, regardless of the fact that the number is best calculated as an average of actual empirical testing :wizard:

 

[steven_e007]

Hi,

I've recently built my own shutter tester, so have got quite involved with this.

(When I say involved - I was measuring the width of slots in aluminium discs being rotated at carefully measured revolutions to calibrate my tester, type involved!)

There are two issues to consider here - the accuracy of your shutter - and the accuracy of the test.

With a focal plane shutter there will always be an error in the tester.

What type of shutter tester are you using?

If the shutter tester has a single sensor (photodiode or phototransistor) then this has a size, a width. As the slot in the shutter passes this sensor, after a certain portion has been uncovered the output will rise until the circuitry registers this as 'on'. As the trailing edge covers the sensor at some point it will register as 'off'. There is likely to be a difference between these two points - and it could be a fair chunk of a millimetre or so. Obviously the calibration of the tester will allow for this - but the problem is that different shutters have different curtain speed and slot sizes. So... if the sensor is effectively 0.5mm, say, and a particular shutter uses a 2mm slot at a certain curtain speed for, say 1/250s, then the sensor is 25% the size of the shutter slot. If a different camera uses a faster curtain speed to get the same exposure (say a vertically moving titanium film rather than a horizontal cloth curtain) with a bigger slot, say 3mm, then the sensor is more like 15% of the shutter gap. The correction factor applied would need to be quite different.

So - unless the shutter tester is calibrated differently for different shutters, the accuracy will vary.

Of course, with a film it is individual photosensitive grains in the emulsion that 'see' the patch of light passed by the shutter - and these are more or less a dimensionless point, compared to the shutter testers sensor.

A different approach is to measure the curtain speed using two sensors (both detect the leading edge of the curtain - and do so in the same way with the same sensor - so the errors cancel out). This is more accurate and can give an accurate measure of curtain speed. You then calculate the exposure knowing the slot size.

BUT! There is also an inherent error here, too. The patch of light on the film is not exactly the same size as the shutter slot, with nice sharp edges. Because the shutter must sit at least a short distance clear of the film, the light patch will tend to be a little larger, with fuzzy edges. Again, this will vary with different cameras. My tester was built for use with a Speed Graphic. Here the shutter is so far forward of the film (to clear the sides of the dark slide) that the patch of light seen on the film (or ground glass) is MUCH bigger than the shutter slot!

I hope that makes some sort of sense

In the end I calibrated my shutter photographically using 'traditional' methods (Photographing a grey card and measuring the resulting densities) and the used the shutter speeds I determined to calibrate my shutter tester!

 

[nick mulder]

If the shutter tester has a single sensor (photodiode or phototransistor) then this has a size, a width. As the slot in the shutter passes this senor, after a certain portion has been uncovered the output will rise until the circuitry registers this as 'on'. As the trailing edge covers the sensor at some point it will register as 'off'. There is likely to be a difference between these two points

There's also hysteresis to account for...

From wikipedia again: (I know, yes I'm lazy)

:munch:

"Hysteresis can be used to filter signals so that the output reacts slowly by taking recent history into account. For example, a thermostat controlling a heater may turn the heater on when the temperature drops below A degrees, but not turn it off until the temperature rises above B degrees. Thus the on/off output of the thermostat to the heater when the temperature is between A and B depends on the history of the temperature. This prevents rapid switching on and off as the temperature drifts around the set point."

Hysteresis is additive in the math, it wont scale with the period changes so it will cause more and more relative error the shorter the period it is you are testing...

I haven't thought about it enough yet to be %100 its a concern - need to draw some pictures, but its a hunch - I'm trying out the new tactic that my hunches are enough for me to go on nowadays

 

[steven_e007]

There's also hysteresis to account for...

Yep!
There the rise and fall times of the sensor, too. Photodiodes are very fast - but phototransistors are usually quite a bit slower to react. Unfortunately, steps taken to reduce the sensor size error (e.g sticking a pinhole over them) is likely to reduce the current, slowing them down. It will still be fast - a lot less than a millisecond probably, so as a percentage of the time measured at slow speed it will be quite negligible - but as the shutter speed gets faster - approaching 1/1000s or more - it can start to become a significant error.

You can correct for some of these errors using a 'fudge factor' - but the big problem is when the conditions change with a different shutter... you really need to change the fudge factor.

 

[Blanc]

Well.
Your shutter seems to work really good...
Most high end companies standards were 20% on slow speeds and 35% on high speeds.

For the test conditions, there are a lot of variations in a non controlled environment (back torsion due to the opening of the door, air moisture and temperature...) plus the fact that the first fire on a camera in the day is generally slower than the others, the phenomenon is worst when the camera was kept cocked...

For a true answer to your question: camera makers know that a spring can not behave linearly, so most of them tricked the cells to match their speeds.
But the reason why your films seems good is just because film have a huge overexposure tolerance.
You can for example give an HP5 a difference of 2.5 fstops to really see a difference. And this tolerance is the reason why most fineprint shops keep telling us to overexpose and under develop. I'ts a rule since the begining of photography and wih a good developper it's quite and hard game to give too much exposure.

 

[JMC1969]

We learned about this stuff as 'JND's':

http://en.wikipedia.org/wiki/Just-noticeable_difference (as dull as the article may be it's fascinating stuff huh)

Are you talking about seeing the result of a shutter difference in a print/neg or by looking/hearing the shutter itself ? Assuming the neg and everything else equal (i.e. a mental exercise) then, um, yeh no idea on the number - but yes, there is a number... Opinions will vary on its value, regardless of the fact that the number is best calculated as an average of actual empirical testing :wizard:

Nick,
1st, I'm omitting the log base2 thing, because I am just not on that level.lol

Ultimately I am looking for the expectable value of change in exposure that any one individual can visually detect a difference and relate it to a fraction of "Stops". Quick, short story. Many years ago when I purchased my first shutter, lens combination (which I thought was for LF and turned out not, different story), I took it to the local camera repair shop. He took it back to a piece of equipment out of my view, click, click, click and brought it back and said "it's about a 1/3 stop of, just adjust your exposure at the time of imaging". I understood this in a relation as to how it effected my photography. Had he handed me a sheet of paper with a bunch of numbers on it shows me actual shutter speeds and percentages, I believe I would have dropped it in the nearest trash. It would have meant nothing to me at the time and now I want to relate the two.

Further links from the link you provided did give me this:

"It was found that the emission of only 90 photons was required in order to elicit visual experience. However, only 45 of these actually entered the retina, due to absorption by the optic media. Furthermore, 80% of these did not reach the fovea. Therefore, it only takes nine photons to be detected by the human eye."

Not that I have any perception how to use the measurement of 9 photons to determine an answer for myself. I'm also not sure it would really help to break it all down to the wave length of light in nm, but, do understand that out visual region is between 400nm - 700nm.


Steven,

What type of shutter tester are you using?

Here it is

Bottom Lt is the sensor I used. Sorry for the camera phone pics, it's the only digi I have. Basically I ran through this quickly and found what was working for my needs at the time. I don't completely understand all the workings of this thing yet. I used the one with the diffuser in it first w/ a 50mm lens and got results, but decided I would rather test w/o a lens.

If I understand the rest of your statement here correctly, I'm not sure these variables have a bearing on the outcome of the test I am performing. Though the differences from camera to camera and how they have determined their ratio of slot sizes and curtain travel is camera specific, the overall result should be the same when measuring the amount of time it takes for light to pass through that ratio and onto the tester sensor and/or film in milliseconds. In the end, a 1/4 of a second is a 1/4 of a second regardless of how the camera is set up, right?:confused:

Blanc,

Your shutter seems to work really good...
Most high end companies standards were 20% on slow speeds and 35% on high speeds.

So that would be approx 1/5 stop in slow speeds (either direction) and approx. 1/3 stop in fast. I was scared that I might be in need of a new Canon F-1 (not F-1n). I am going to need to sit down and really learn the functionality of this shutter tester. Hopefully in the long run it will work for older leaf shutters as I like to repair TLRs' and LF shutters for a hobby.

Thanks for the replies guys,
Jody

 

[RalphLambrecht]

... What does that translate into as fraction of F-stops? ...

f/stop difference =LOG(actual time / target time )/LOG(2)

Let's take your first average measurement example for 1 s.

f/stop difference =LOG(0.738/1.0)/LOG(2)
f/stop difference =LOG(0.738/1.0)/LOG(2)
f/stop difference =0.13194/0.30102
f/stop difference =0.44 stops

The standard for shutter speeds is quite liberal and allows different deviations for different temperatures. For B&W work I consider red, yellow, green approach (see attached) with a tighter tolerance towards underexposure.

Also, please note how I chart the ranges plus the average. A shutter set to 1 s and exposing for 1/2 s one time and 1 1/2 s the next has a perfect average of 1 s but a terrible performance.

 

[steven_e007]

Nick,

Steven,

If I understand the rest of your statement here correctly, I'm not sure these variables have a bearing on the outcome of the test I am performing. Though the differences from camera to camera and how they have determined their ratio of slot sizes and curtain travel is camera specific, the overall result should be the same when measuring the amount of time it takes for light to pass through that ratio and onto the tester sensor and/or film in milliseconds. In the end, a 1/4 of a second is a 1/4 of a second regardless of how the camera is set up, right?:confused:

Blanc,

No, it is not the same. The reason is that the way the film behaves and the way a sensor behaves are completely different.

I don't know what type of sensor you are using - I can't tell from that photo - but whatever it is it will be a photoelectric device of some sort with a particular size, probably a few millimetres across. Your shutter consists of a slot moving across this sensor. As soon as light illuminates the edge of the device - it will start to produce a voltage (or allow current to flow or whatever... depending on the device type. Let's say it produces a signal). As the slot exposes more of the sensor's surface to light, the signal gets bigger. At some point, decided by the designer, the electronics must be triggered to register this as 'on' or 'being exposed'. After the electronics have been triggered to record the signal as 'on' - it will stay in this state as the shutter continues to move - until the sensor starts to be covered again. In the same way - at some threshold the falling signal level will be regarded as 'off' or 'shutter closed'.

The important thing is that the distance between the point at which the tester decides is 'open' and the point at which it decides it is 'closed' will be of the order of something like 1/2 a millimetre or more, maybe. This could be 25% of the width of the slot you are measuring. But - on a different shutter making the same exposure - it might be 10% or 30%. So - the error changes with the slot size.

Now consider a photographic film. This consists of millions of microscopic of photosensitive grains. The size of each grain is totally negligible compared to the size of the slot - any slot.

Consequently the way the sensor sees the slot and the way the film grains see the slot are totally different.

That is just one scenario - with a single sensor. With a diffused sensor the situation is different - and with multiple sensors different again (here you are measuring the curtain speed, which is better for small focal plane shutters) but in every case the way you detect the moving shutter slot has characteristics which are quite different to the way film works. You have to take these factors into consideration if you want to get accurate readings.

I suspect that most manufacturers know what curtain speed relates to what effective shutter speed they require for a given slot size - and measure for that.

 

[John Koehrer]

All the speeds are close enough for government work except 1/2000 which is a stop off. That's not at all unusual.
Your slow speeds are fast, not slow until you get to 1/8 which should be 125ms and you're at 113ms that's mighty fine.

 

[nick mulder]

And jumping the gun here a little does anyone care to let the reciprocity issues of film come into the equation ?

Remembering it is a factor with very small exposures as well as long ...

 

[RalphLambrecht]

Most of the 'working' speeds are fine, but the 'slow' speeds are causing almost 1/2 stop underexposure. For B&W work, I would correct for that by rating the film 1/3 stop slower.

 

[RalphLambrecht]

And jumping the gun here a little does anyone care to let the reciprocity issues of film come into the equation ?

Remembering it is a factor with very small exposures as well as long ...

I wouldn't worry too much about reciprocity failure in this range, but you are right, the underexposure at the slow speeds will be worsened through reciprocity failure.

 

[nick mulder]

Yeh, don't get me mistaken - I'm talking completely in terms of mental wankery here - trying to determine where, how and when our thought experiment begins and ends ...

If all other things are going to be equal, we need to define 'not other'

 

[steven_e007]

Yeh, don't get me mistaken - I'm talking completely in terms of mental wankery here - trying to determine where, how and when our thought experiment begins and ends ...

If all other things are going to be equal, we need to define 'not other'

Mental what? Sorry - obviously a technical term with which I am unfamiliar....

Here's a bit more mental w... er, a bit more food for thought:

Take a completely different shutter... say a leaf shutter.

As we probably all know they suffer from 'efficiency' problems. When the shutter is fired and starts to open - it acts like an iris - starting as a pinhole and growing larger until it gets to full aperture. It then sits fully open for a while - then closes. During the opening and closing times the shutter only passes part of the light. At slow speeds the open and close times are negligible. At the fastest shutter speed then there may never be a pause at the 'fully open' point - the shutter starts to close as soon as it is opened. At this speed the shutter is 'less efficient' as the film only sees the full light from the lens for an instant. This is why most leaf shutters do not exceed 1/250 or maybe 1/300 of a second. Even the very fastest are restricted to about 1/500th. Usually this is achieved with a (very powerful) supplementary spring that only comes in on the highest speed or two - and the shutter takes a lot of thumb pressure to cock.

So... Consider another shutter, say the Thornton Picard roller bind shutter. Here a curtain passes the back of the lens with a slot in it just slightly bigger than the aperture it covers. As it opens the light is passed through almost a crescent moon shape - which grows through all the phases of the moon until just for an instant it is full - then it waxes (or is it wanes?) until it is closed. If in 'Instantaneous mode' - i.e: not held open - then this is by design a very inefficient shutter. There is never a period where the full aperture is held open, unlike the later leaf shutters.

Plot a graph of light intensity against time and you get a triangle - sloping up and then sloping down. With a compur type shutter you get a slope up - a flat top - then a slope down. The faster the shutter speed the shorter the flat top and the lower the efficiency.

So - stick these two on a shutter tester and what do we have? Well, at some point the tester is triggered 'on' by the rising light intensity. It doesn't matter at which point - as long as it is the same for both (except it might be different for the film!) The point is - if the tester measures the shutter speed - the actual exposure received by the film, especially at lower speeds will be much higher for the compur - since it will have chance to sit fully open for maybe the big majority of this time. For the film to see the same exposure from the TP rollerblind - the actual time from when the shutter actually opens to when it closes would have to be much longer - maybe nearly twice as long? (I have a horrible feeling to be sure would involve... calculus :eek:. Let's not go there!).

Another factor is that with any leaf shutter - the efficiency varies with the aperture (because at f22 the tiny aperture is obviously opened much more quickly than the full aperture of the lens). Another factor to consider.

The only point I'm trying to make here is that you can't measure 'shutter speed' and expect it to accurately relate to film exposure. It isn't like measuring temperature or even light intensity - you have to consider what is going on.

The OP said "In the end, a 1/4 of a second is a 1/4 of a second regardless of how the camera is set up".

Well, no, sorry. I humbly disagree. A 1/4 of a second exposure where the shutter is, on average, only passing half the light from the lens and 1/4 second exposure from a shutter that snaps open and shut at speeds that are negligible compared to the exposure time are certainly NOT the same. With the FP shutter - the design is almost 100% efficient at any aperture - again a different set of factors to a leaf shutter - but here the characteristics of the sensor need to be considered.

 

[Q.G.]

Steve,

Less well 'advertised', but focal plane shutters too have an efficiency problem.
It is caused by the distance between the focal plane shutter and the film (the same thing you describe about behind the lens shutters, but a bit less dramatic).
And it is (part of) the reason why focal plane shutter's fast speeds usually vere towards the boundaries of the tolerance range, i.e. it can quite significant, a long way off your almost 100%.

 

[steven_e007]

Yes indeed, that's why I include the word 'almost'. Maybe I should have said 'sometimes almost'!

Another effect I noticed on my Speed Graphic shutter is that the gap between the shutter and film is so big that the slot 'image' has very fuzzy out of focus edges. Quite noticeable - a sort of penumbra, if you like. I suspect this could have the effect of giving the film a slight 'pre-flash' as the shutter transverses the film (and a post flash too, for that matter). Whether it has any real effect on the exposure or not I don't know - but I would have though that with a slow curtain speed and a fast film - it might well do!

 

[Q.G.]

Yes indeed, that's why I include the word 'almost'. Maybe I should have said 'sometimes almost'!

Another effect I noticed on my Speed Graphic shutter is that the gap between the shutter and film is so big that the slot 'image' has very fuzzy out of focus edges. Quite noticeable - a sort of penumbra, if you like. I suspect this could have the effect of giving the film a slight 'pre-flash' as the shutter transverses the film (and a post flash too, for that matter). Whether it has any real effect on the exposure or not I don't know - but I would have though that with a slow curtain speed and a fast film - it might well do!

Yes, it has.

That is exactly the thing that causes the efficiency problems with focal plane shutters.
They do not produce a sharply defined 'slot' of light on the film, but one with fuzzy edges. The degree of fuzz, of spread, depends (mostly) on the distance between shutter and film.
That spread means that the exposure is in fact a bit longer than it would have been were the transition from dark to light instantaneous. And with the fuzzy edge being the same width no matter how broad the slit, the effect of it on the total exposure is larger when the slit formed by the shutter curtains gets smaller, i.e. with shorter speeds.
The spread also depends on the aperture. But the effect is (i believe) less than that of the aperture on leaf shutter's efficiency at short speeds.


Leaf shutters get (deservedly) 'bad press' when it comes to efficiency. But focal plane shutters do not deserve being spared the same "bad press" the way they are.

 

[steven_e007]

Hmmmm. Interesting. Thing is though, because the spread is fuzzy and acts like a pre-flash, it doesn't just affect the exposure - it will surely affect the film gradation as well? I wonder if any Speed Graphic users (or any camera where you can use both leaf and FP shutters) have ever noticed a difference in gradation (better shadow detail) between the two shutters on the same lens and film?

I must try it myself some time.

 

[Q.G.]

Hmmmm. Interesting. Thing is though, because the spread is fuzzy and acts like a pre-flash, it doesn't just affect the exposure - it will surely affect the film gradation as well? I wonder if any Speed Graphic users (or any camera where you can use both leaf and FP shutters) have ever noticed a difference in gradation (better shadow detail) between the two shutters on the same lens and film?

I must try it myself some time.

It's just part of the total exposure.
It will have an uplifting effect on shadow detail, because the shadows (like the rest of the film) get to be exposed a little bit more than they should.

 

[nick mulder]

I have a horrible feeling to be sure would involve... calculus :eek:. Let's not go there!.

No please, lets !

Perhaps just work with a couple of triangles and a rectangle for focal plane ?

But the integrals involved with leaf shutters ? hmmm

 

[steven_e007]

No please, lets !

Perhaps just work with a couple of triangles and a rectangle for focal plane ?

But the integrals involved with leaf shutters ? hmmm

Be my guest... as long as I don't have to work it out myself!!! :eek:

The overall point of all this, I think, is that you can't just measure shutter speed with a tester like you measure temperature. Really, you need to test for the effect you want photographically. I've seen many people complain that compur style fast shutters only achieve 1/350sec when set to 1/500 - as if that is a huge error. In reality you should do a reciprocity test, on film, against 1/250 and 1/125 (stopping the lens down each time). Quite probably the slow speed compensates for the low efficiency. As long as the 1/500s setting gives us about have the exposure of 1/250 - then as photographers we should be happy?

 

[RalphLambrecht]

I don't know, but I don't like the idea of relying on two wrongs making a right.

 

[steven_e007]

I don't know, but I don't like the idea of relying on two wrongs making a right.

But I think with the inherent shortcomings in certain shutter designs, there necessarily has to be a few wrongs, so maybe better to try to get them to cancel each other out?

 

[RalphLambrecht]

But I think with the inherent shortcomings in certain shutter designs, there necessarily has to be a few wrongs, so maybe better to try to get them to cancel each other out?

Try to cancel them out?

What if they add up?