Adventures in film characteristic analysis

[Bill Burk]

Of course since we are possibly off by a full stop, we don’t know how that stop is distributed… the 0.15 intervals can’t be 0.15 intervals, this may reveal that you have higher contrast than we measured

 

[Bill Burk]

Another thing could be light leak in the device allowing some of the bright tubes to influence exposure of the low tubes. Do you have good segregation between each of the tubes? Could the circuit board itself be piping light?

 

[radiant]

Here’s a paper graph. I measure 0.57 CI and just a tad short of ASA parameters.
@radiant you and I have been discussing where I think exposure falls vs where your nominal device exposure is designed to fall.

The 0.57 CI might be caused by the mess with developer concentration. I've targeted to 0.62. It is not a big difference of course.

I continue to estimate the exposure which would cause HP5+ to meet 0.1 above base plus fog is -2.70 log mcs. The curve crossed 0.1 above base plus fog right between two of your tubes. But I estimate that actual exposure is 0.01 or 0.02 higher because you didn’t hit ASA parameters.
So I think the tube you labeled -3.056 is -2.76 or -2.77 and the tube you have labeled -2.9031 is -2.62 .
We can surmise why the effective exposure seems to be higher than the engineered exposure, but I think the factor(s) will reveal themselves as you do more and more tests.

The lowest density tube should be getting 4.4ms at 0.2 lux. I'm quite sure it should get that. That is 0.00088 lux seconds and in log scale it is -3.055. I will double check that. Maybe I need to measure the pulse, too.

Spectrum reaching the film could be “more actinic” than daylight through glass lenses of a camera (UV could be hitting the film? Does your diffuser pass UV?).

I have no block for UV. But on the other hand WS2812b leds do not go that down, lowest at 465 nm based on manufacturer data sheets. https://cdn-shop.adafruit.com/datasheets/WS2812B.pdf

You have a chance to try R G and B separately.
Have you electronically isolated the LED’s so there’s no noise or trace voltage that could be weakly energizing the LED’s that are supposed to be off?

RGB-separation is on the list for sure!

The other LEDs are fully off as those are digitally controlled. There is no light coming from the leds. But there can be some transmission between my slots/tubes. But that should be very very low. I cannot see any leakage with my eyes.

It is actually quite interesting to look in dark as the slots are deep. If you don't look straight over the slot when the program is running, you cannot see any light. It feels like "broken" until you adjust the slot to be directly pointing at your eyes. It feels a bit strange actually because of so directed light / no light anywhere else. A bit same feeling when you try to look for light leakage on bellows.

I think this is interesting when I have now reached to the toe are exposure and understand how demanding that is for proper speed analysis. CI analysis can be done quite accurately on this data, it seems.

Next time I build densitometer I will just do a fast controllable small light table and stick Stouffer on it

 

[radiant]

Of course since we are possibly off by a full stop, we don’t know how that stop is distributed… the 0.15 intervals can’t be 0.15 intervals, this may reveal that you have higher contrast than we measured

Yes I was also calculating one stop off. That seems very very stange while the slots are half stop apart.

Another thing could be light leak in the device allowing some of the bright tubes to influence exposure of the low tubes. Do you have good segregation between each of the tubes? Could the circuit board itself be piping light?

Of course there is this possibility. It is hard to measure / notice. The leakage would be in relation to next slot time, the brighest tube is hardly leaking to the lowest because of the mechanic block and because of distance.

Should I just build the quick flasher light-table

 

[Bill Burk]

Tape off the bulbs and pour a black ooze over the board.

Same with the backing behind the film.

Look at all the parts with your IR device. Are they “white”?

TMY2 is relatively insensitive to IR, there is a hard cutoff in near IR. I don’t know about HP5+

 

[radiant]

Tape off the bulbs and pour a black ooze over the board.

Same with the backing behind the film.

Look at all the parts with your IR device. Are they “white”?

TMY2 is relatively insensitive to IR, there is a hard cutoff in near IR. I don’t know about HP5+

I will probably run one test with exposing a bit longer on slot and skipping the exposure on the next one - leaving one blank between exposing slots. That way it is easy to see if there are any light leaking. I cannot measure any light piping/leaking on the exposure surface at all - there is no sign of exposure between slots.

There is a good chance that I've just messed up my calculations because of this is the first time I did two strip test. I will crosscheck later.

 

[Bill Burk]

At that rate, do a tight series of only the low exposures 0.1 apart

What do the pieces look like under infrared viewer?

One thing about your series that makes me think it’s electronic and not light leakage… is that the toe drops at a rate that’s typical for film. Flare makes an elongated toe.

Can you mount the gadget to a fan blade and spin it around and take a picture to see the light duration?

 

[radiant]

edit: I'm getting lost ..

 

[radiant]

I've now cross-checked my log-lux-second calculations and came up with same results few times. So those are correct, I'm sure about that.

Bill is on right marks when asking if my device is leaking light + dropping like film. Shutting down the LEDs takes about 0.65 milliseconds which I haven't taken into account on the timing. On longer times it doesn't matter but on shorter times it takes a bigger role.

The last visible/measurable slot is exposed for 2.21 ms. Next ones are 3.1 and 4.2ms. 0.65 milliseconds is 27% - 13% of the total time. That probably causes the toe to have more details. And because the shut-down-time is proportinal it makes the curve smooth?

 

[Bill Burk]

very cool! subtract .65 milliseconds from each exposure and away you go!

 

[radiant]

very cool! subtract .65 milliseconds from each exposure and away you go!

Yup! Bill, do you think the curve is otherwise fine? The toe has a bit of incorrect shape but that isn't a big deal. My question leads to next..

I've been adding features to my online graph visualizer at https://jouni.kapsi.fi/filmgraph/ - and I would like to hear opinion about this. Are the points visualized in correct positions for example?

You can change the ISO for the analysis in the text box and see what happens. I think it works logically and one can see where the Zones hit on different exposures with that tool by chaning the ISO. Also it calculates negative density range which helps understand what part of the exposure you would use when printing. At least for me this has been very informative, but I would like to hear analysis if the software works correctly ..

 

[radiant]

This might be really stupid question but I will risk it (again):

From my previous results we can see that the slope of H+D stays pretty much constant along the workable part of exposure. If I change my EI to 800 or even 1600 the negative density range (for 7 stops SBR) still stays in the linear part of the film. There is plenty of room to play.

My question is that why would I change my development time if I get same contrast with EI 800 and 1600 ? Of course my overall density changes but it only changes printing exposure times, right? Ok, I got CI 0.62 at 400 and CI 0.58 at 800 and 1600, but I think that is so small difference that there isn't need to compensate that. Printing is still within normal grades.

Or have I understood something completely incorrectly?

EI 400:

EI 800:

EI 1600:

 

[Bill Burk]

You’ll notice how close you are to the toe when you consider 7 stops of subject luminance range at 1600. The longer development time helps improve image quality down in the toe. A photograph taken at the 1600 setting might place shadow exposure lower than you planned.

You could definitely take some actual photos and plot parts if the pictures on the graph of its development curve, then you will see how well your planned exposure fit the model.

I did that here:
https://beefalobill.com/imgs/tmxaim.jpg

 

[radiant]

You’ll notice how close you are to the toe when you consider 7 stops of subject luminance range at 1600. The longer development time helps improve image quality down in the toe. A photograph taken at the 1600 setting might place shadow exposure lower than you planned.

Yes, at 1600 I'm close to toe, there isn't much room for error in exposure. On lower stop range, low contrast scene, then I'm more safe and on linear part. Of course this is possible becuase of this film+dev combo being really,really good.

It is good that my thinking is not completely wrong, I start to understand this. I think my plotting tool has helped me to understand everything as it draws everything at the same time.

I've done some shots and prints with new dev times that you Bill analysed for me. I've been using 800 as EI for HP5+ to have a bit more speed - and now looking at the graph that is good choice. What has really changed is that I'm printing now on G2-G3 area and the prints seem to get much better. Also printing is more enjoyable and the results give positive energy back.

One of my latest prints: http://kuvau.tuu.fi/kebu-3/ (the three latest posts are from same setup). I've have done few other unpublished prints that just work great.

Today I shot one 6-7 SLR/SBR scene with EI 400/800/1600 just to see if my theory is right that the film is working in linear. I overexposed a bit just to keep the EI 1600 from hitting the toe, but it doesn't really have negative effect as I'm just testing the linearity. If my analysis is correct I can print three similar prints with only adjusting the base time and while printing on same grade.

I've also kept logbook of my prints and analysed the printed negatives. I'm not sure what I can do with the data but as I grow the log I think I can see patterns. I have a bit same kind of data that you have in your graph, actually. Not that many points but at least meaningful shadow / highlight information so I know what has been the negative density range for the exposure chosen on paper.

 

[Stephen Benskin]

Yes, at 1600 I'm close to toe, there isn't much room for error in exposure. On lower stop range, low contrast scene, then I'm more safe and on linear part. Of course this is possible becuase of this film+dev combo being really,really good.

Today I shot one 6-7 SLR/SBR scene with EI 400/800/1600 just to see if my theory is right that the film is working in linear. I overexposed a bit just to keep the EI 1600 from hitting the toe, but it doesn't really have negative effect as I'm just testing the linearity. If my analysis is correct I can print three similar prints with only adjusting the base time and while printing on same grade.

I've also kept logbook of my prints and analysed the printed negatives. I'm not sure what I can do with the data but as I grow the log I think I can see patterns. I have a bit same kind of data that you have in your graph, actually. Not that many points but at least meaningful shadow / highlight information so I know what has been the negative density range for the exposure chosen on paper.

Shooting a 400 film at 1600 is the same as rating the film at 400 and stopping down two. Rating a film at 1600 and opening up isn't the same as shooting it at 1600. You will be misinterpreting the results.

The example you have is a good looking image, but it has stylized lighting. With exterior scenes, we have a internal idea of how they should look in sunlight and in overcast light. Interior light, especially stylized lighting, doesn't have such a criteria and consequently harder to judge the relationship between exposure and quality. That you are able to obtain a desired print with less effort does indicate better exposure / processing control.

 

[radiant]

Shooting a 400 film at 1600 is the same as rating the film at 400 and stopping down two. Rating a film at 1600 and opening up isn't the same as shooting it at 1600. You will be misinterpreting the results.

Maybe I was not clear enough on my explanation. I of course understand what happens on 400->1600 change.

I think what happens is:
- my NDR area moves to the left, but stays on linear part where gamma is pretty much the same all the way
- without compensating on developing time I just get thinner negatives and I have to shorten my print times
- printing grade doesn't change
- with normal or low SBR/SLR I won't be hitting the toe, close of course and no margin for error

Those are my findings and I would like to hear did I misintepret the results?

The example you have is a good looking image, but it has stylized lighting. With exterior scenes, we have a internal idea of how they should look in sunlight and in overcast light. Interior light, especially stylized lighting, doesn't have such a criteria and consequently harder to judge the relationship between exposure and quality. That you are able to obtain a desired print with less effort does indicate better exposure / processing control.

Ok, I understand. That wasn't actually direct example to this, just to show that I'm doing other things than creating H+D curves

That particular negative was shot on the edge of if there is enough light. It printed very well, easy to print. Not that much of play because it is quite contrasty but still in class "fun to print". There are other frames on that film that didn't make it, too much smoke and shakyness.

But; easyness and outcome of that print (from EI 800 HP5+) was in line what I have seen from the curves, so it is kind of example. I have other good frames already and the results are very promising.

 

[Stephen Benskin]

Maybe I was not clear enough on my explanation. I of course understand what happens on 400->1600 change.

I think what happens is:
- my NDR area moves to the left, but stays on linear part where gamma is pretty much the same all the way
- without compensating on developing time I just get thinner negatives and I have to shorten my print times
- printing grade doesn't change
- with normal or low SBR/SLR I won't be hitting the toe, close of course and no margin for error

Those are my findings and I would like to hear did I misintepret the results?

You are correct. For a shorter than normal LSLR, underexposure will generally not have the lowest point of exposure fall far into the toe. A normal LSLR is another matter. As the speed point is located in the toe, normal shadow exposure already falls in the toe. Here's a two quad example.

Exactly where the deepest shadow falls depends on the degree of flare, but even without any flare, the shadows will fall at approximately the fractional gradient speed point. Flare gives about a stop underexposure latitude.

BTW, brightness is a psychological term and is no longer used. Only those who follow Phil Davis' BTZS still use it, and I believe he kept it only because he didn't want to change his program. Luminance is psychophysical.

 

[radiant]

You are correct. For a shorter than normal LSLR, underexposure will generally not have the lowest point of exposure fall far into the toe. A normal LSLR is another matter. As the speed point is located in the toe, normal shadow exposure already falls in the toe. Here's a two quad example.

Hmm, maybe I don't understand something (yet).

EI 400 Zone V is at -1.7 on film MCS. If we go 3.5 stops down (half of "normal" 7 stop SLR) we are then at -2.3. ISO 400 speed point is at -2.7. I don't see that falls into toe yet? Or well, it depdends on description of "toe" but for example my development at -2.3 there is good distance to toe. At -2.3 I'm on very linear part of the negative, right?

 

[Stephen Benskin]

Hmm, maybe I don't understand something (yet).

EI 400 Zone V is at -1.7 on film MCS. If we go 3.5 stops down (half of "normal" 7 stop SLR) we are then at -2.3. ISO 400 speed point is at -2.7. I don't see that falls into toe yet? Or well, it depdends on description of "toe" but for example my development at -2.3 there is good distance to toe. At -2.3 I'm on very linear part of the negative, right?

Maybe it's how you define toe or how it's drawn in your curve. 0.10 over Fb+f (Hm) is 1.0 log-H below the metered exposure (Hg) and not 1.05. Shadow without flare falls about 1.25 log-H. From a paper by Jack Holm,

 

[Bill Burk]

You can add a bottom line to Paper LER, then slide the pair up and down like you slide the Subject Luminance Range to understand greater and less exposure.

 

[radiant]

Maybe it's how you define toe or how it's drawn in your curve. 0.10 over Fb+f (Hm) is 1.0 log-H below the metered exposure (Hg) and not 1.05. Shadow without flare falls about 1.25 log-H. From a paper by Jack Holm,

Yes, this is about how to define toe. Basically we are trying to say what is the area where shadows start to lose details. Maybe 1.0 log-H is a good rule of thumb from toe to measured exposure. On my graphs the toe exposures are a bit off, but I believe HP5+ can get a bit lower than that, but that is a just this use case.

 

[radiant]

You can add a bottom line to Paper LER, then slide the pair up and down like you slide the Subject Luminance Range to understand greater and less exposure.

Yes, my web-tool does that. It is very informative.