Adventures in film characteristic analysis

radiant · Jan 27, 2022

I came up an excuse to build myself film strip exposing device because I wanted to know first-handed how film developing affects to film characteristics. As if this was not described thousand times here and everywhere. Anyways, good reason to design the thing.

The device can expose 16 "strips" at once with any kind of step interval as it is controlled by my own code. The apparatus is 3D printed and has two diffusers to get even illumination to each of these strip-slots.

I've calculated it to give the 18% gray tone to slot number 8 and then it goes up and down in stops I have defined. The light illumination is 0.2 lux at each slot. I control the exposure by time.

Short description.. now to the test in next post.

radiant · Jan 27, 2022

First test candidate: HP5 developed in Rodinal 1+25 for 6 minutes, 1 minute agitation. Very standard.

Exposed at half stops. The strip with a notch is the 18% tone. So I think I got 2-3 stops under the 18% and 4 stops over. Please note that the highlight range was not enough to reach any sort of maximum density. You can see the max density in the lower part of the image.

But I cannot find the toe shape. It seems to be pretty linear or something is not right.

I tried to analyse the negative with my enlarger meter but it failed a bit; the film must be measured from directly under the enlarger head and I got bad readings. Then I analyzed with Epson V600 scanners "pick color" tool. That way I got something happening in the toe area, but that might have been just measuring error. Now measuring from this scan output file, it seems to be really linear:

So is HP5 this linear in reality? Or do I have bug somewhere? The exposure times are based on my measurements dead on.

Kino · Jan 27, 2022

You have a linear graph of a log progression!

albada · Jan 27, 2022

radiant said:
I came up an excuse to build myself film strip exposing device because I wanted to know first-handed how film developing affects to film characteristics. As if this was not described thousand times here and everywhere. Anyways, good reason to design the thing. The device can expose 16 "strips" at once with any kind of step interval as it is controlled by my own code. The apparatus is 3D printed and has two diffusers to get even illumination to each of these strip-slots. I've calculated it to give the 18% gray tone to slot number 8 and then it goes up and down in stops I have defined. The light illumination is 0.2 lux at each slot. I control the exposure by time. Short description.. now to the test in next post.

Interesting idea!
Could you post a photo of the device?
Did you use a decoder-chip to drive the 16 LEDs with the microcontroller's output-pin? Or does the microcontroller have 16 outputs?

Also, if each tile (strip) is illuminated by a different LED, then you will need to calibrate the tiles because LEDs don't have perfectly equal output. How did you calibrate them?

Mark Overton

radiant · Jan 27, 2022

Kino said:
You have a linear graph of a log progression!

*facepalm*

radiant · Jan 27, 2022

albada said:
Interesting idea!
Could you post a photo of the device?
Did you use a decoder-chip to drive the 16 LEDs with the microcontroller's output-pin? Or does the microcontroller have 16 outputs?

Also, if each tile (strip) is illuminated by a different LED, then you will need to calibrate the tiles because LEDs don't have perfectly equal output. How did you calibrate them?

It uses WS2812b LED strip. That was my base idea, no need for external drivers and low amount of soldering. I cross-checked few leds and they output really same amount of light. I could of course calibrate, but for now the accuracy is enough for me. I'm going to use this to really extreme tests so that level of accuracy is probably not needed.

The actual device is very black blob so I think 3D render is much better:

It is just a bunch of chimneys to get distance to the leds. Pretty simple actually.

ic-racer · Jan 27, 2022

To test your sensitometer you could use a common 21 step wedge and contact it with the same film. The overall shape of the curve should be similar if your sensitometer is behaving correctly. Even the most expensive sensitometers use a step wedge, however. In fact I have never heard or read of a device such as yours. I think you are finding out now why it (multiple separate light sources) has never been used in a senstiometer design.

Light Capture · Jan 27, 2022

ic-racer said:
To test your sensitometer you could use a common 21 step wedge and contact it with the same film. The overall shape of the curve should be similar if your sensitometer is behaving correctly. Even the most expensive sensitometers use a step wedge.

+1 on this.
It's hard to make measurements without reference. Calibrated 21 step wedge will give you measured values for each step to be able to compare.
Uncalibrated one is the same but it doesn't come with measurements for each step.
Step wedge can also be scanned together with film strip and then it can be compared in Epson software.

radiant · Jan 28, 2022

I have Stouffer T2115, I will scan next to the film and lets see what it looks like!

koraks · Jan 28, 2022

ic-racer said:
I think you are finding out now why it (multiple separate light sources) has never been used in a senstiometer design.

Not necessarily. Back when equipment for film was being developed at a large scale we had to make do with slow and unreliable incandescent bulbs (for purposes of dimming or very short time exposures) and/or awkward shutters, so different approaches were more attractive. There's no doubt in my mind that most engineers today would do something quite similar to what @radiant did.
One of the problems I see a lot among people who are enthusiastic about the technicalities of film photography is a very deeply rooted flavor of conservatism. I suppose it comes with the territory. But in fact, film photography and modern semiconductor electronics are an absolutely fantastic combination.

Bill Burk · Jan 28, 2022

You could vary the steps, use smaller difference near the toe (e.g., 0.05 for a few steps), then go to greater steps (e.g., 0.15 for a few steps) and then to greater still (e.g., 0.3 for the final steps). That way you can graph 3.60 or more range with the most information in the toe where speed is determined. The usual step wedge is only 3.0 range and that’s more a limitation of accuracy of film density over 3.0

radiant · Jan 28, 2022

The led turn on-off cycle is so fast that only the logic inside the chip is the limit. But single WS2812b LED can react to 40uS commands, that isn't the limiting factor either. The variance in manufacturing the led might cause variation but what I've measured the output level of these LEDs are very very equal.

radiant · Jan 28, 2022

Bill Burk said:
You could vary the steps, use smaller difference near the toe (e.g., 0.05 for a few steps), then go to greater steps (e.g., 0.15 for a few steps) and then to greater still (e.g., 0.3 for the final steps). That way you can graph 3.60 range where the T2110 is only 3.00 with the most information in the toe where speed is determined

That is exactly my plan. I was planning to do 24 step design but my 3D printer limits that.

Bill Burk · Jan 28, 2022

Instead of varying you can make paired exposures… 16 low exposure and then next higher exposure and up for the second 16

Bill Burk · Jan 28, 2022

So do 32 of .10 for the usual 3.x range and have another set of pairs 32 of .15 for a 4.8 range

so the switch would select .10 or .15 increments, and it would always make pairs.. you make two strips every test

radiant · Jan 28, 2022

Is there something off here, I scanned Stouffer + first test strip together in single scan:

And plotted together (I left linear Y-axis as it shows the "problem" much better):

Orange=Stouffer, Blue=film

Data here (first film, then Stouffer):

178 254
168 251
154 223
143 190
123 162
111 135
101 106
88 79
70 48
61
49
41
38
35
32

otto.f · Jan 28, 2022

radiant said:
the highlight range was not enough to reach any sort of maximum density

So a picture with this film/developer would look quite Ilfordy, isn’t it? In the sense that in the final print deep rich blacks will be missed?

radiant · Jan 28, 2022

otto.f said:
So a picture with this film/developer would look quite Ilfordy, isn’t it? In the sense that in the final print deep rich blacks will be missed?

Hmm, I don't quite get this. I was talking about maximum density of negative = highlights in print. The exposure of the film was not enough to reach the maximum density of the film, that I was just pointing out. The exposure of the last slot is +4 stops above 18%.

radiant · Jan 28, 2022

Bill Burk said:
Instead of varying you can make paired exposures… 16 low exposure and then next higher exposure and up for the second 16

You mean two strips, so that the whole film range is spread on 32 slots? Yes. That is also my plan.

Bill Burk said:
So do 32 of .10 for the usual 3.x range and have another set of pairs 32 of .15 for a 4.8 range
so the switch would select .10 or .15 increments, and it would always make pairs.. you make two strips every test

4.8 = 16 stops, right? Yes that would be interesting too. I want to go over the shoulder for sure, yes.

bernard_L · Jan 28, 2022

radiant said:
*facepalm*

Think twice before posting such abrasive comments. Kino's question was legitimate.

Basic rule about graphs: they should show for each axis which quantity is plotted, and the unit, as in: distance (m) or duration (s). Your graph shows neither.

Plus, you get numbers out of your scanner. They are expressed in some color space, and their relation to the luminous flux passing through the film is (a) not logarithmic; (b) non-linear and complex, with many different forms floating around. See e.g. : http://www.brucelindbloom.com/index.html?Math.html; I give below a sample at random:

radiant · Jan 28, 2022

bernard_L said:
Think twice before posting such abrasive comments. Kino's question was legitimate.

Cool down. That facepalm was for myself. I corrected the graph to logarithmic as result so I guess that is obvious I'm the one with the mistake.

bernard_L · Jan 28, 2022

"Corrected" graph was so obviously unlike a logE-D graph that it gave the impression you were poking fun.

My other comments stay. You plot graphs but not clear what is on the vertical axis.

radiant · Jan 28, 2022

Some of my own analyis from previous scan:

Stouffer wedge is half stop and the plot of it is really linear so I think we can use that. I calculated that half stop is 29 in grayscale value. Also the data from film scan looks pretty linear so I think we can trust that too.

I then calculated the difference between my film slots in stops and with some filtering I got result that the slots are about 0.22 stops apart (or 0.066 in density). So the compression in density is 0.44 (from "real stops" to how the film represents this).

Did I get this right?

radiant · Jan 28, 2022

bernard_L said:
"Corrected" graph was so obviously unlike a logE-D graph that it gave the impression you were poking fun.

My other comments stay. You plot graphs but not clear what is on the vertical axis.

I understand that. I cannot afford poking anyone here because I'm super interested in your input and ideas.

The units are missing because I guessed those are obvious. Y-axis is just the grayscale value and X-axis is the strip position. So no "real" values, just relative to the scan itself.

radiant · Jan 28, 2022

One thing that is horribly wrong is the exposure amount. There are only maybe two-three stops visible under the 18% gray - position. Andrian Bacon writes " If you expose the film at EI 400 or 500 you will get very usable images with a solid 4-5 stops of shadow detail below a correctly exposed 18 percent grey card before film base plus fog and 5+ stops of highlight density with an easy 2-3 stops of over exposure latitude if you need it, though, you don’t really need to give it that much exposure." here https://adrianbacon.com/simple-photography-services/simple-film-lab/films/400hp5/

So maybe my exposure is about 2 stops too low. What do you think?

Adventures in film characteristic analysis

Moderator

Privacy & Transparency

Privacy & Transparency