Reciprocity misbehavior.

I am not suggesting that there is a direct analogy or that the same mechanism is at work. I am saying that the idea that the reciprocity relationship may have a shape part and an amplitude part. The shape part can be exponential of the type tm ^ K that is common to all films. That is what I am seeing, not theorizing. I am wondering if there is any theory that would explain what I am seeing. There seems to be that possibility because there is a similar observation to be made of the transfer fundtion of a junction diode. I don't think I am stupid or off my rocker for asking and I do see a relevance. I read your web site, and you called silver halide a form of semiconductor.

Gainer, please do not misquote me. Silver bromide is an indirect gap semiconductor but this does not mean that silver bromide and pn silicone junctions behave the same or obey the same physical law.

Anyone who studied semiconductor devices would know that the kinds of carriers present are very important, perhaps more so than the material itself. You are comparing very different things here.

Again I do not see the relevance.

Patrick - selenium is a semiconductor too. Does that mean when we tone our prints with selenium that we are making electronic components?

Toooo much oversimplification I think.

Kirk Keyes said:

Patrick - selenium is a semiconductor too. Does that mean when we tone our prints with selenium that we are making electronic components?

Toooo much oversimplification I think.

Click to expand...

Well, that is your oversimplification, not mine. I know for a fact that I can use the properties of a silicon junction diode, a couple of op-amps, a couple of pots and some resistors to make an analog of the reciprocity behavior I saw and analyzed. If I knew a lot about the theory of semiconductors, I would not have asked the question. I was hoping someone would know if there is a more direct analog. A simple "No" would have sufficed.

We have been using light meters with calculators that are analogs of the reciprocity law that says every film requires a certain number of meter-candle-seconds for correct exposure in a direct relation such that the product of illumination and exposure time is a constant for any given film. If we put an aperture before the sensor, as it is in a camera, the light falling on the sensor would be the same as the light falling on the film, assuming we use the meter correctly. There could then be a more direct transfer function from sensor to readout as seconds of exposure, and it could be any function that strikes our fancy, including the exponential function that I like.

Ryuji said:

Gainer, you are bringing up irrelevant issue here. What you are talking about is forward voltage drop at the pn junction. Reciprocity law failure is mostly due to the loss of the carrier or unstable latent subimage. All these are described on my webpage and references therein in greater details. The former is negligible in the exponential region of the pn junction, and the latter concept does not exist in pn junction.

Click to expand...

Your web page describes the mechanism of reciprocity by theoretical concepts but does not use these concepts to form an algorithm by which the real reciprocal relationship can be predicted for any particular film. If it is possible to do so, why not do it? We would be glad to call it by your name. We would like something we could program into a light meter to convert illumination measurements into exposure time for any given film.

gainer said:

Well, that is your oversimplification, not mine. I know for a fact that I can use the properties of a silicon junction diode, a couple of op-amps, a couple of pots and some resistors to make an analog of the reciprocity behavior I saw and analyzed.

Click to expand...

Well, you go from talking about AgX being a semiconductor to discussing diode junction electronics. Too big of a jump for me. Now you are talking about modelling the properties of silicon diodes using op amps. Getting even further afield.

Now if you are trying to say that some carefully constructed set of op amps model something, well that's nice. You could do it with a digital computer as well. Nothing special there. There are may ways to model things.

Ultimately though, I think you are trying to defend your reciprocity equation. OK. I really have no problem with the modelling that the equation does. I think it looks like it does a reasonable job to matching Bond's data. So now, let's test it against some more data and see how it holds up. And it does sound like more checking needs to be done to verify the novelty of it.

But I do have an issue with all the scientific mumbo-jumbo. "Art is real. Science is imaginary." ? Come on!

I still don't see any significance to phi with your model. I think I calculated the standard deviation of the results and it was at 4%. Perhaps your model is really built on 34/21 and not phi, afterall, they are not that much different. And then bringing up the Boltzmann constant? Maybe there is a correlation, but I think a lot of work needs to be done first.

"If I knew a lot about the theory of semiconductors, I would not have asked the question. I was hoping someone would know if there is a more direct analog."

I like asking questions too! But it just seems to me that a lot of needless speculation and even misinformation gets put into these Internet forums that it just confuses and misleads a lot of people. Especially when it comes from a respected person like you, Patrick. Afterall, not everyone has quite as technical background as you and it will make it hard for them to find the real gems of information from all the surrounding chaff.

Kirk Keyes]Well, you go from talking about AgX being a semiconductor to discussing diode junction electronics. Too big of a jump for me. Now you are talking about modelling the properties of silicon diodes using op amps. Getting even further afield."
I'm not modelling the diodes. They are part of the model.

"Now if you are trying to say that some carefully constructed set of op amps model something, well that's nice. You could do it with a digital computer as well. Nothing special there. There are may ways to model things. "

Sure you can use a digital computer, but you can get two op-amps in an 8-pin 1/2" square package that run off a few volts.


"Ultimately though, I think you are trying to defend your reciprocity equation. OK. I really have no problem with the modelling that the equation does. I think it looks like it does a reasonable job to matching Bond's data. So now, let's test it against some more data and see how it holds up. And it does sound like more checking needs to be done to verify the novelty of it."

Defend, deschmend, as some of my friends would say. If I design and sell a successful product using it, I won't even tell anyone that it's in there. If it helps someone else to make art, so much the better. If you didn't attack it, I wouldn't defend it. Your attacks seem more to be based on what you think I know than on what I did. Now I cannot publish data that I porduce by my own experiments for fear of being accused of forging the results. If you want to see more data, go to the AZO forum.

"But I do have an issue with all the scientific mumbo-jumbo. "Art is real. Science is imaginary." ? Come on!"

Art becomes art through the senses, visual, auditory and tactile. Hawings can do science without enough physical ability to hold a pen. It is real in the neural pathways of his brain, but it means nothing to most people. It is in fact a work of art that can be appreciated by very few.

If scientists did not imagine that there is a way to describe in words and pictures the workings of Nature, there would be no science. If we did ot have first the ability to create art, we could not have created science.

It is not mumbo-jumbo to say that every system of axioms at least as rich as arithmetic is either incomplete or inconsistent. If it is, it is not my mumbo- jumbo but that of Kurt Godel.

"I still don't see any significance to phi with your model. I think I calculated the standard deviation of the results and it was at 4%. Perhaps your model is really built on 34/21 and not phi, afterall, they are not that much different. And then bringing up the Boltzmann constant? Maybe there is a correlation, but I think a lot of work needs to be done first."

You guys really don't have a sense of humor, do you? A statistics book I once read pointed out that there was a high correlation between illegitimate births in a certain area and the number of men leaving. Cause and effect? maybe so, maybe no. Statistics won't tell you either way.
What does your 4% mean in terms of f-stops? What the heck are you talking about when you talk about phi? I know it's Greek, and I have seen it used in a lot of places, but what does it have to do with the unweighted method of least squares? Same goes for 31/21.

If you have a set of independent observations relating the same unknown constants and there are more observations than constants, you can solve for the set of constants that give the minimum square of the deviations from the curve represented by the equation that uses the constants. Traditionally, the probable error of fit is 0.67 times the mean square error. That number is based on the assumption that the error distribution is random and normal. If you don't know if the distribution is normal, the best you can do is assume normality. After you're done, you may know more about the distribution if you have enough observations.


"I like asking questions too! But it just seems to me that a lot of needless speculation and even misinformation gets put into these Internet forums that it just confuses and misleads a lot of people. Especially when it comes from a respected person like you, Patrick. Afterall, not everyone has quite as technical background as you and it will make it hard for them to find the real gems of information from all the surrounding chaff."

If someone does not ask foolish questions, we will all think that the existing foolish ideas are absolute fact. Like the well known facts that tomatos are deadly poison and that rubbing garlic on a permanent magnet will destroy its magnetism.

I should also refer you to the real definition of science. Science is knowledge. Don't confuse it with the scientific method, whic is a way of getting that knowledge, and is as much an art as singing or dancing.

May I make the observation; I think both Kirk and Ryuji have missed the point that Mr Gainer is making. The key to Patrick's (healthy imo) exploration here, is in his statement quoted:

gainer said:

I am not suggesting that there is a direct analogy or that the same mechanism is at work. I am saying that the idea that the reciprocity relationship may have a shape part and an amplitude part. The shape part can be exponential of the type tm ^ K that is common to all films. That is what I am seeing, not theorizing. I am wondering if there is any theory that would explain what I am seeing.

Click to expand...

It appears from his observation that many film types appear to be able to have a mathematical model with a shape component and an amplitude component fits the characteristics of film speed performance whilst in reciprocity failure. This is analogous to signal amplification most commonly seen in electronics, and incidentally is a model commonly seen in nature also.
Cellular growth of trees.
Bird wing flap frequency.
Crickets chirping.
etc, etc ....
(Try to argue against that.)

The reason it is a healthy exploration is because when a mathematical model can be successfully applied to a closed system, then that system becomes more predictable. This is tremendously useful to us laymen.
Sorry for the venting of a little frustration; but argueing the method rather than result is ........ well ....... it's just not cricket dear chap!

Wow, what a thread.

When I'm out making a photograph shortly after sunset and the light is decaying, I often think that there must be a moment at which one could start an exposure that it would take all the light you could get (for the rest of the day) to complete the exposure. That is that because the light is decaying while I am exposing the film, my exposure needs to increase and by the time I've reached my predicted exposure, the light has faded more. Do you all think you can add this effect to your graphs?

If this is true, then there must be a point at which one cannot complete an exposure no matter how long the exposure is (assuming that the ambient light is fading forever). You would have started too late .

Just a thought.

Loose Gravel said:

Wow, what a thread.

When I'm out making a photograph shortly after sunset and the light is decaying, I often think that there must be a moment at which one could start an exposure that it would take all the light you could get (for the rest of the day) to complete the exposure. That is that because the light is decaying while I am exposing the film, my exposure needs to increase and by the time I've reached my predicted exposure, the light has faded more. Do you all think you can add this effect to your graphs?

Click to expand...

I don't think this could be easily modelled, as there are too many variables, including things you couldn't know, like sky conditions over the horizon after sunset. Heavy clouds over the horizon could kill the light faster than expected.

Also, the sun sets at different angles at different latitudes at different times of the year. So the rate at which it drops below the horizon changes, which changes the length of twilight according to season and location.

There are good working models of twilight, so you could do something with that, dependant on latitude and time of year, but I think combining that with unknown sky conditions over the horizon would be too complex and involve variables that the photographer couldn't observe in the field.

Lee

Has anyone come up with tc,1 coefficients for additional films like color print, transparency and the multitude of other films that have appeared on the market in recent years?

Thanks

Murray

Hello all,

That one must be my first post here but, I watch the forum for some time now.

I have tried to gather as much info as i can and create a calculator for film reciprocity times...

The programme is still under dev and might be unstable but you can get the general idea...

I implemented the equation

tr = tc,1*(tm^1.62) + tm

where tc,1 coefficients are the following

KODAK TMY T-MAX 400 ........0.061
KODAK 400 TX TRI-X 400 ......0.169
KODAK TMX T-MAX 100 ........0.069
ILFORD HP5 Plus 400 ............0.101
ILFORD 100 DELTA ...............0.046
ILFORD Pan-F Plus 50 ...........0.140
ILFORD FP4 Plus 125 ............0.110
KODAK PLUS-X 125 ..............0.210
KODAK Technical Pan ...........0.140
ILFORD SFX 200 ..................0.450
ILFORD XP2 Super 400 .........0.050
KODAK T400 CN ..................0.030
KODAK TMZ T-Max P3200 .....0.310

for EFKE CHS ART I used ADOX suggestion:

1/2s -> +0 stop
1s -> +1/6 stop
10s -> +1/3 stop
100s-> +2/3 stop

the general equation I used for that case is

tr = tm + (tm * 2^log(tm)) / 6


dev language C#.Net
requires Microsoft Framework .Net 2
win xp or later

in order to upload the file I changed the extension to .pdf
after you download the file just replace the .pdf part by .rar, unzip the file, paste both contained files in one folder and run the "film_reciprocity.exe" file

if you think that it can help I might also develop it for mobile phones...

I would appreciate any feedback

If you have any additional equations for more films I will be glad to add them

have fun...

Christos

Christos,

Whose reciprocity data did you use for each of the Kodak and Ilford films?

Thanks,
Lee

Hello Lee,

Actually I used the data you uploaded at (there was a url link here which no longer exists).

Regards

Christos

Christos,

Thanks for the reply. The data from Reeves referenced in that post has been taken off the web, and that link in my post is dead. Reeves is no longer using film, and the test data from his page is about 10+ years old. The post of mine that you drew from is now almost 5 years old. So some of the films are likely to have changed a bit. TMY has been reformulated, although the few who have mentioned it say reciprocity is similar to the older version.

Lee

Lee L said:

Christos,

Thanks for the reply. The data from Reeves referenced in that post has been taken off the web, and that link in my post is dead. Reeves is no longer using film, and the test data from his page is about 10+ years old. The post of mine that you drew from is now almost 5 years old. So some of the films are likely to have changed a bit. TMY has been reformulated, although the few who have mentioned it say reciprocity is similar to the older version.

Lee

Click to expand...

Quite possibly, but for the moment that is all the data I have. I guess I could run my own tests but it'll take ages... Maybe I should develop another version with the ability the users to add their own films and coefficients.

regards

celetron said:

Quite possibly, but for the moment that is all the data I have.

Click to expand...

I'm not implying that there's better data around now, only that people who read this need to confirm or adjust through personal testing. If there is better data, I haven't seen it. And I'm disappointed that Reeves took down his color and B&W film information.

Bond's methods were exhaustive. Reeves uses a rather short and simple test that I use. It's outlined in his and Michael Covington's books on film Astrophotography. (Be aware that the Covington book 2nd ed. has a typo that I found last year in one equation used to calculate reciprocity, so check his website for the errata page.) This method uses a 1/8 second exposure, then 128 second exposures under the same conditions but with a 10 stop neutral density filter and some bracketing. The amount of bracketing in stops that produces the same density at 128 seconds (with filter) as the 1/8 second exposure (without filter) is used to calculate a Schwarzschild exponent. This is a check at a single data point, but it's very good at giving a useful indication of the relative degree of reciprocity failure in a film. You could also use other ND filters of different values to get more data points, but 128 seconds is a good start.

This testing takes only one reference frame and then however many bracketing shots you feel you need, at whatever step rate in stops would be useful to you. Four frames and reasonably careful work will give you a very good start.

Lee

Just an observation: I have the original Howard Bond article, "Black-and-White Reciprocity Departure Revisited," and in it, Bond states clearly that Kodak provided him with sheet film. Since 400TX does not come in sheets, he must have tested the 320TX, not the 400-speed-roll-film version that seems to be mentioned everywhere in the thread. Unfortunately, the usually careful Bond only identifies the film he tested as "Tri-X," with no other details. Since it was sheet film he was testing, however, we must assume that it was the 320 version and, therefore, revise all the earlier references to "400TX" in this thread to read "320TX."

Nevertheless, this is a fine thread with much useful information.

Best,

Doremus Scudder

Interesting

Hope it's not a faux-pas to revive old threads...

I'm trying to figure out the reciprocity failure for the EB/RA Carestream blue-sensitive x-ray film that I use in my 4x5 because I'd love to do some night work with it. I use a Wratten #44 on my spot meter to approximate what the film is sensitive to. Wondering if there's a way that I could use my equipment to experimentally figure out the important number for this formula. Would I need a densitometer to figure it out?

Just pull your dark slide in 1" steps and set a time for each step. Process and see what yo get.

Oh, it's good to see Mr. Gainer again!

You are over thinking this. Forget about the filter. Just go out and shoot. The film is cheap enough to do so.

The filter is to allow me to use the spot meter without having to guess. The amount of red light in natural sky light, or fluorescent bulbs, etc. differs enough, even from moment-to-moment, and imperceptibly, that I do not want to leave my exposures up to chance. I find that using the wratten 44 for my metering permits me to get far more consistent results that I had been getting before using it.

As for this thread: I want to nail down the tc,1 coefficient, but I'm not sure exactly how I'll know whether my negatively was "properly" exposed at 1 second with any given aperture. I figure it's important if I'm trying to accurately map the reciprocity failure of the film.