Printalyzer - Darkroom enlarging timer & exposure meter

[radiant]

..

 

[dkonigs]

I'm also CS embedded engineer so I tried to not to make a project about it. I needed to remind me that my hobby is darkroom printing, not darkroom timer/analyzer building

I probably should put more smileys over and around every single sentence for extra impact, for some reason.

Anyways, if you asked me about a project like this last year, I'd have probably said the exact same thing as you. I normally have an aversion to spending all my time on making tools for a hobby, when I could actually be working on that hobby itself. However, this year my actual shooting has basically taken a nosedive (for reasons that are probably obvious) and the project has given me something to help preserve my sanity among all of this. I also do really enjoy working on embedded electronics projects, but am terrible at actually coming up with good ideas for those projects.

When you do look around in our niche, there is tons of opportunities for new contraptions. There just isn't enough commercial potential to support most of them as a real business venture. The equipment that does exist is essentially a relic of the 90's (at best) and may still work, but could be better or more reliable/supportable.

 

[albada]

Make sure you darkroom-test your display.

Thanks for the warning about that. Luckily, I copied your idea and put a blackout switch on the controller, so if worst comes to worst, I'll change the settings, turn off the display, and then pull out the paper.
Also, the fact that Roscoe #19 solves the fogging problem suggests that it would also make a good filter for a safelight.
Mark Overton

 

[dkonigs]

Just a little taste of what I've been working on this past week, in the form of a picture of my cluttered desk...

Basically, I've decided to go ahead and start working on my densitometer interface, as the first step of improving my paper profile generation process.
Of course before I can do that, I need to actually be able to interface my device with densitometers. For older units, like the X-Rite 810, that means using a USB<->Serial adapter to connect to the serial port on the back of it. For newer units, like the Heiland TRD-2, its a direct USB connection. However, internally, the TRD-2 does this using a USB<->Serial chipset, so its really the same from the perspective of my device.

But wait, there's more!
In order to do this, I first needed to actually make my microcontroller capable of talking to these USB<->Serial adapters. Support for that was actually not built into the platform's USB host stack. And obviously, there are many kinds of such adapters out there. So I spent the better part of this past week figuring out how to get the FTDI FT232R working (most common chipset, also used internally by the Heiland TRD-2). I then spent the last day or two working on the PL2302, and next after that is the CP2102. Hopefully this covers enough of the bases for now.

With any luck, it won't be long before I'm actually writing real code to speak the protocols of these densitometers (that's actually the easiest part of this whole project). Once that's wrapped up, maybe in another week or so, I'll finally be able to start working on densitometer-assisted paper profile creation.

For now, I'm going to focus on the Heiland TRD-2 and the X-Rite 810 (and similar models) as that's what I have to work with. If there are any other popular densitometers out there with serial or USB ports on them, and they're worth adding support for, please let me know.

But until then, here's a picture of using that serial interface to make my X-Rite 810 display custom text on its LCD!

 

[ic-racer]

Very nice. One of my X-rites is the 890 automatic densitometer with an rs-232 intended to go to a modem for two way communication. The interesting thing about the 890 is you just feed the strip (transmission or reflection) into the slot and it determines if it is transmission or reflection and reads and records (or transmits) all the patches automatically.

 

[dkonigs]

Very nice. One of my X-rites is the 890 automatic densitometer with an rs-232 intended to go to a modem for two way communication. The interesting thing about the 890 is you just feed the strip (transmission or reflection) into the slot and it determines if it is transmission or reflection and reads and records (or transmits) all the patches automatically.

Looking over the manual of the 890, it seems as though its designed specifically to read a variety of common formats of manufacturer-provided process control strips.
Can you actually use this on a strip you exposed yourself, using one of the common off-the-shelf step wedges (like Stouffer sells)? If so, how does it work for that?

 

[ic-racer]

There are two manuals for it. The second manual describes the film formats.

 

[dkonigs]

There are two manuals for it. The second manual describes the film formats.

Just found the second manual. I'm guessing the part that applies is the section on "Generic Strips", but I'm not sure whether a typical step wedge paper exposure would qualify or if you'd have to make some jig to make it expose with the region separation that's recommended there.

Have you used your 890 for measuring things that aren't off-the-shelf control strips, and if so how well has it worked for that?

 

[dkonigs]

In the middle of all my research for this project, I actually have stumbled across another product that's very similar in its goals to what I'm trying to do:
DLG Enlarging Meter / Timer

I even found a forum thread here talking about it:
https://www.photrio.com/forum/threads/enlarging-meter-timer.161117/

The preliminary manual is also quite interesting, as it gives a good idea of what they're trying to accomplish. But unfortunately, it seems like they put the whole project on indefinite hold back in 2019 and don't seem to have said anything about it since.

There were two reasons cited for this:

• Lack of a big enough market due to similarities with the RH Analyser (not a concern for me, but probably a concern for them).
• Concern over infringing on the RH patent.

They clearly didn't do any research on the patent issue, or they'd have realized that the RH product was designed in the 90's and patents don't last forever. I actually found the RH patent (GB2299688B), and observed the following about it:

• It was filed in 1995
• It should have expired in 2015
• It actually expired in 2001, likely due to nonpayment of a renewal fee
• It reads more like a user manual for a product, than like a conventional patent covering a technology

 

[Nicholas Lindan]

In reading a patent you really only have to pay attention to the claims. There is only one independent claim in the RHD patent:

1. A darkroom light meter having means for sensing the intensity of the light at a point in an image to be projected onto printing paper and a precalibrated scale for providing a visual representation of the print density of that point on the resulting print after exposure and development of the printing paper under the same parameters as employed during calibration.​

The patent only covers the little grey scale above the readout. All other claims are dependent on claim 1. If you don't have the grey scale the patent doesn't apply even if your product is otherwise a direct clone of the RHD product.

The popular conception is that a patent exists to protect the inventor. Nothing could be further from the truth. A patent trades a limited period of exclusivity in exchange for revealing everything about the invention to the point that someone 'skilled in the art' could duplicate the invention. After 20 years your product is free for the taking. Kodak was famous for gaming the system by breaking an invention into pieces, obfuscating and individually patenting the pieces, and then covering the whole with a thick layer of red herring irrelevancies. Then if someone did happen to come up with the a similar product Kodak would extract the relevant bits, put them back together again and claim infringement. Then there was the IBM approach: "We have more money to spend on lawyers than you do." There was a saying "Never dance with IBM."

 

[dkonigs]

A patent trades a limited period of exclusivity in exchange for revealing everything about the invention to the point that someone 'skilled in the art' could duplicate the invention. After 20 years your product is free for the taking.

And its been more than 20 years, so this product is free for the taking

Kodak was famous for gaming the system by breaking an invention into pieces, obfuscating and individually patenting the pieces, and then covering the whole with a thick layer of red herring irrelevancies. Then if someone did happen to come up with the a similar product Kodak would extract the relevant bits, put them back together again and claim infringement.

I think any large corporation with lawyers to write their patents does exactly this. Heck, I have a patent (from work) that the lawyers laced with so much nonsense that I don't think anyone could actually figure out the invention from reading the patent by itself.
Though nowadays they don't just use them to stop similar products. They also use them to maintain a permanent state of stalemate (or M.A.D.) with other large corporations where everyone infringes on everyone else's patents. So its better for them to play nice with each other, because any suit could be met with a counter-suit.

 

[albada]

The DLG product contained a built-in reflection densitometer. According to the thread linked above, the user would slide a test strip through a slot in the machine. But the DLG manual shows the densitometer as a separate plug-in device. Either way, that made the product a complete system because no additional device would be needed, which would be a strong advantage over RH Designs.

An employee at the prior large company I worked for mentioned that the company uses patents as "trading cards".

 

[Nicholas Lindan]

An employee at the prior large company I worked for mentioned that the company uses patents as "trading cards".

Yup. Patent trading is all over the place in the semiconductor and software industries. Texas Instruments held many key patents in digital logic, they licensed them to anyone for a $1/year. They, however, insisted on their dollar - they wanted the kowtow. The late 60's to the late 80's were the golden age of IP sharing in Silicon Valley. Then AMD decided to get greedy...

 

[dkonigs]

Okay, I think its time for another update... Now that I'm able to directly feed densitometer readings into my device (got it working for both the Heiland TRD-2 and the X-Rite 810), I've moved on to doing something useful with them. Specifically, creating paper profiles from step wedge exposures! (This is basically a process detailed as very manual and notes/paper/spreadsheet oriented in the back of the RH Analyser calibration guide, except that I can do it all on-device.)

The process I envision is basically:

• Setup the enlarger and take a light reading (taking note of a lux-second based value the Printalyzer will give you)
  
• Insert contrast filter, expose a strip of paper under a step wedge, and process it
  
• Measure each patch with a densitometer, feeding the numbers into the Printalizer
• Push button, out comes a full characteristic curve and calculated values for all three reference-points (mentioned in my previous posts) necessary to define the paper profile.

That being said, here are some screenshots for the pieces of this I've assembled so far. It all works, though there's certainly room for user-interface improvements in the future.

First, there's the various screens for setting up your step wedge:

You can pick from a variety of known configurations, then tweak/calibrate each patch if you want to be extra-accurate. I'll probably target the Stouffer T2115 as my default (its what RH uses too, so I have one), but something physically larger would probably be easier to measure from (especially with a chunkier densitometer like the X-Rite untis).

Then we have the intro screens to a paper profile, showing the top-level properties and the properties for a specific contrast grade:

When you select "Measure From Step Wedge," it then takes you down to a screen with some basic high-level properties and a list of all the steps to input measurements for:

The "Paper Dmin" is a convenience value if you have a densitometer that's meant for absolute readings and isn't so practical to "zero" on the paper base (e.g. the X-Rite units). The "Paper Dmax" is basically the "max net density" value from the earlier screen, shifted by that Dmin value. The "Paper Exposure Value" is that lux-seconds based number I've mentioned before, and is auto-populated based on the last meter reading. However, it can be changed.

Finally, when you select "Calculate Profile," it does all the math and spits out the results:

The way the math works, basically involves combining the step wedge properties and the "Paper Exposure Value" to determine the amount of light that actually fell on each patch of the step wedge exposure. It then interpolates the graph of this versus the reflection densitometer readings of the paper, and finds the relevant intercept points.

 

[albada]

Nice!
The bottom screen has [OK] and [Cancel] buttons on the screen. Is this a touch-screen perchance?

Because the Printalyzer controls the enlarger, I suggest that you allow the user to make his own test-strip the conventional way instead of using a Stouffer wedge. This has the following advantages:
* Eliminating the Stouffer means there's one less object the user must procure and store.
* The user is likely to have lost the little Stouffer wedge 1-3 years later when he needs it.
* Patches will be larger, making measurements easier.
* Stouffer wedges are somewhat inaccurate, drifting a surprising amount over 10 or so patches. So controlling patches yourself can give more accurate results. To minimize the effects of ramp-up and ramp-down times, tell the user to use a 50mm or shorter lens, stop it down to f/16 or smaller, and raise the head high -- all causing exposures to be longer and thus more accurate.

Anyway, we appreciate the update!

 

[ic-racer]

Okay, I think its time for another update... Now that I'm able to directly feed densitometer readings into my device
The way the math works, basically involves combining the step wedge properties and the "Paper Exposure Value" to determine the amount of light that actually fell on each patch of the step wedge exposure. It then interpolates the graph of this versus the reflection densitometer readings of the paper, and finds the relevant intercept points.

That is very nice. Do you do the interpolation with triangulation? Or do you try to get the curve formula?
My spreadsheet that does all this automatically for film measures a slope of a tiny portion of a curve by assuming the line is straight between the two reference points straddling the value to be computed.

Nice thing about computers, once you have the dataset entered, you can go crazy with the math. My spreadsheet for film calculates ASA (ISO), Delta-X and W-speeds and contrast index and draws the graph with all the calculated points.

 

[dkonigs]

The bottom screen has [OK] and [Cancel] buttons on the screen. Is this a touch-screen perchance?

No, its not a touch screen. That's just one of the prompt display styles used by the display library I'm using (u8g2) that I haven't shown much in the shared screenshots.

Because the Printalyzer controls the enlarger, I suggest that you allow the user to make his own test-strip the conventional way instead of using a Stouffer wedge.

You can already do this. I just haven't built out the functionality to do complex calibration calculations based on it, but I do provide all the information from which could could nail down the calibration numbers yourself with that method.

Just keep in mind that duplicating a step-wedge exposure with conventional test strips has some disadvantages:

• To duplicate the exact process, you'd need to be able to accurately make a test strip with 14-21 separate patches you can distinguish. Good luck doing that without some jig you don't have.
• To get the same results with reasonable-sized test strips, you may have to trial-and-error many test strips to ensure you cover the paper's entire contrast range.

I could provide a mode to eliminate some of the trial-and-error in the RH approach to doing this (for the D=0.04 reference point) using densitometer measurements. But at that point, one can assume the user has a densitometer. And if the user has a densitometer, they can just do the damn step wedge exposure. And if they're worried about the step wedge drifting, they can measure it first.

Regardless, at the end of the day this is all software. And software can be easily updated over time. Unlike older devices, I have no need to burn a "final" version into some ROM chip that's fixed in stone and will never change. Making the firmware user-upgradable is a very important feature I want to have, such that all of this nonsense could be revisited by me (or anyone else) years from now and improved.

 

[dkonigs]

That is very nice. Do you do the interpolation with triangulation? Or do you try to get the curve formula?
My spreadsheet that does all this automatically for film measures a slope of a tiny portion of a curve by assuming the line is straight between the two reference points straddling the value to be computed.

My initial test programs for this did something similar to what you're doing, by assuming the curve is a series of straight line segments. However, that's not how I'm doing it now. Basically, I'm using a process called "cubic spline interpolation" that gives me an actual curve formula. (Well, technically its a series of piecewise curves, but it can be treated as one continuous curve for most of my purposes.) One disadvantage of this approach is that I cannot easily calculate the point intercepts directly. However, I can easily calculate the density value for every single exposure point (within the number of significant figures I care about) and find the closest match. It might seem a little brute-force, but the range of values is small enough that its still a fairly quick process.

I'm not sure how easy it would be to do this inside a spreadsheet (without some add-on package), but that's one advantage I have by not using a spreadsheet. General purpose mathematics software does have this capability. Or in my case, I found a vector math library for my specific microcontroller (CMSIS-DSP) that has functions for doing exactly this.

At the end of the day, my numbers might more accurate at those intermediary values than with the line-segment method. However, given enough data points, the difference is likely small enough that it doesn't actually matter all that much for practical purposes.

Of course once I've gone through this whole exercise to find the 3 reference points (Hm, Hs, Ht), I do a much simpler polynomial interpolation between those points to create the tone graph that's actually used for print metering. This works because the characteristic curve doesn't really flatten out between those points, and I don't care about plotting exposures outside of the paper's spec-defined contrast range at that stage.

Nice thing about computers, once you have the dataset entered, you can go crazy with the math. My spreadsheet for film calculates ASA (ISO), Delta-X and W-speeds and contrast index and draws the graph with all the calculated points.

Indeed you can. I've certainly done a lot of tinkering between spreadsheets, math software, and standalone test programs myself in the process of doing all of this.

 

[ic-racer]

Yes, the spreadsheet I made does something similar. It 'looks' for the value (like the 0.1 point) between all the known data points then, when it finds them I connect the two datapoints with a line and find the estimate that way. Just that function (your curve fitting) makes your device valuable, even if it didn't have an exposure meter!\
Continuing to follow with interest...

 

[Nicholas Lindan]

Beware when using a step tablet for sensitometry. The tablet may be accurate but the light source won't be uniform, especially if the light source is an enlarger. Problems really arise if you are using a projected step tablet because all sorts of effects now come into play.

To mitigate the problems you can:

• Scan a projected step tablet with a light sensor to determine the exposure given at each step and use these numbers for the exposure. Do not rely on the tablet's published/calibrated density.
• Scan the illumination light for a contacted step tablet and correct the tablet's density values with the measured illumination at that step.
• Do without a step tablet and instead use a test strip printer that moves a strip of paper past an aperture. Use a precision timer to make the exposures. Exposures around a second will have some error unless you have a timer that can compensate for the lamp's on/off time. I use the Durst test strip printer https://www.pinterest.com/pin/371828512968305479/ I slice an 8x10 sheet into 4 2x10" strips and feed them through.

The last method gives the best results. It also allows you to use whatever exposure interval you wish.

 

[dkonigs]

Beware when using a step tablet for sensitometry. The tablet may be accurate but the light source won't be uniform, especially if the light source is an enlarger. Problems really arise if you are using a projected step tablet because all sorts of effects now come into play.

I'm not particularly worried about this for the following reasons:

• The enlarger will likely be high enough to project a large image (11x14" or bigger) while the step wedge is relatively short compared to this area, so the exposure will almost entirely be "in the center"
• I do support "measuring the step wedge yourself" to deal with any spec-inaccuracies. (This is what I really mean by "calibrated step wedge", since official calibrations always have an expiration date attached to them.)
• I actually also have a fairly sophisticated "enlarger calibration" process that measures lamp behavior during exposures so that my timer can correct for the lamp's on/off time. My hope is that it'll be a major benefit to short-increment test strips.

I've also noticed some confusion between the terms "step wedge" and "step tablet" where there doesn't seem to be an authoritative source on which term is right. Its mostly a matter of opinion from whoever is selling a product, based on whether or not they care about distinguishing between transmission and reflection tools.

• Stouffer uses "wedge" for transmission and "tablet" for reflection.
• X-Rite uses "tablet" for transmission and "plaque" for reflection
• Kodak uses "tablet" for transmission and "control scale" for reflection
  
• Tiffen uses "tablet" for transmission, I think.

Meanwhile, the dictionary defines the term "optical wedge" as "a wedge-shaped filter whose transmittance decreases from one end to the other: used as an exposure control device in sensitometry" (in other words, what we're using for transmission.)

 

[Nicholas Lindan]

The terms originated in the graphic arts industry. Step wedges were made of tinted gelatine cast in the form of a very flattened staircase and bonded to glass. Hence the name. They were placed on top of lithographic printing plates to check for correct exposure. I think the #4 step had to be solidly black (well, blue, the color of the photoresist (or maybe #4 should be solidly white/aluminum)) after developing. A density tablet was, IIRC, just a block of plastic or ceramic with painted squares white -> black.

The old terms carried on when they were replaced with photographic negative and print materials.

Exposing at high enlarger heights will help with edge fall off but will really play havoc with stray light effects. Put a mask in the negative carrier so you are only illuminating the wedge. And check with a meter. You should put alignment strips on the densitometer so you are always measuring each step on test print in the same place.

I'm not giving the advice through needing to blow off air. I have been there and done that. I'm trying to save you some grief.

 

[dkonigs]

Exposing at high enlarger heights will help with edge fall off but will really play havoc with stray light effects. Put a mask in the negative carrier so you are only illuminating the wedge. And check with a meter. You should put alignment strips on the densitometer so you are always measuring each step on test print in the same place.

I'm not giving the advice through needing to blow off air. I have been there and done that. I'm trying to save you some grief.

Such advice is always welcome, and I'll keep it in mind as I proceed. I know the RH calibration process involves finding the D=0.04 exposure point with the enlarger high up (and it takes a fair bit of trial-and-error to nail it), then moving it to a mid-way position to get better illumination for the step wedge to get the contrast range. My attempt here is to combine these two. Regardless, this is one part of my system that can always be improved and revised later. Partly with software updates, but mostly by simply improving the (as of yet non-existent) documentation for the process. That's why figuring out the system for storing paper profiles (as mentioned many posts back) is almost more important right now than the process of creating them.

Right now my goal is to essentially make my project "feature complete," then swing back around later and start improving things. Its very easy to get distracted along this journey, of course. That's why I generally feed these conversations back into my notes for things to revisit. A lot of the things I'm probably going to prioritize in the near future are parts of the project that don't directly relate to photographic features, but are still critically important. (e.g. saving/loading configurations, handling firmware upgrades off a USB stick, ironing out some hardware issues/improvements that'll feed directly into the next revision of the PCBs, etc.)

 

[dkonigs]

So today I decided to go into the darkroom and do a head-to-head test between my RH Analyser Pro and my Printalyzer prototype, at the task of making a simple test strip. Specifically, I wanted to see if all of my crazy "enlarger calibration" nonsense made a difference.

So my test setup involved building simple test strips in both "incremental exposure" and "separate exposure" modes on both devices, with the identical specified enlarger time, contrast grade, aperture setting, and patch increment. The first test was centered at a 15s exposure, and the second test was centered on a 7.21s Grade 2 exposure.


I then measured all the strips with my densitometer, to see how the two compared. Now you can't really compare the full strip between devices, because RH uses "funny math" to calculate their f-stop timing increments (best explained as a slightly-off attempt at doing fixed-point math on an 8-bit MCU). But what I was really looking for was the comparison between "incremental" and "separate" modes for each device.

With the RH Analyzer, there was a small divergence between the two modes, that seemed to grow a bit the further you got into the strip.
With the Printalizer, the two modes yielded IDENTICAL results at 15s, and almost-identical (within a D=0.01 margin of error) results at 7.21s.

I'm quite happy that my crazy deep-dive into making my device able to profile the rise/fall times of the enlarger was actually good for something.

The ultimate goal here is that when you say "Expose for 15 seconds," what you actually get is "The equivalent exposure of 15 seconds of full light output." Sure, with a 15 second exposure, this is mostly meaningless. But with a 1-4 second exposure (as you might have with incremental test strips), it does make a measurable difference.

Now I did do a lot of bench-test measurements of the RH Analyser's actual output-toggling behavior, and it does seem to stretch the exposure time by a little bit (40-80ms, and it varies a little depending on the length of the specified exposure), which may help them a little. If not for that, the difference would be even more dramatic. But it does not beat actually calibrating the output timing to the enlarger's behavior.

 

[albada]

I don't remember whether you are using the Printalyzer's meter to measure rise/fall times. Normal light is so dim that exposure times would be too long, but placing the meter on the lens with the aperture wide open would be so bright that you could take a few hundred exposures per second. Then transfer-rate would be a problem. I doubt I2C would be fast enough, but SPI might be. Or use an 8-bit MCU to buffer inside the meter itself. I know how much you love 8-bit MCUs.
Mark Overton