Scanner as color densitometer

[Steven Lee]

Basically the subject says it all. Knowing the formula for density, using linear sensor data, and having a IT8.7 color profile (I have Wolf Faust targets), I see no reason why a scanner or a camera cannot be used to accurately measure HD-LD values of control strips. Color densitometers don't show up much on eBay, and when they do they're close to $1K. Besides, I worry about their accuracy due to old age. They also never offered with calibration targets, and even if they did they probably would have been faded.

So I'm thinking about devising a process here, and what I need (or I think I need) is a bunch of used control strips with known measured values written on them. Or perhaps there are color density targets I can buy?

Thanks.

 

[Bill Burk]

Search eBay for: Xrite i1 Scanner Professional Color Management Software & Photoshop Elements 10 and buy the twenty dollar item

 

[Steven Lee]

@Bill Burk But isn't this X-Rite version of what I already have? (Wolf Faust IT 8.7 target with Lumariver). I can make color profiles, but that's not the same as getting accurate RGB/CMY density values. Basically, I am stuck at the stage when I have a linear TIFF scan of a control strip, I compute RGB densities from the 16-bit RGB values, and now I need a known reference to error-check the result. I know that Kodak control strips come with a properly developed reference strip, but that's just one sample. I think I need variability to make sure my method is accurate.

 

[Bill Burk]

I overlooked the part where you mentioned you had a target.

 

[koraks]

Basically the subject says it all.

Yup, and I had the same thought, same Faust target, tried it out and...it doesn't work at all. Readings are still miles off. Don't bother with this approach if it needs to be anything close to accurate.

Search for a used xRite i1Pro. Yes, they cost a pretty penny, but mine I got for about half the price you mentioned and it came with everything - including excellent support from xRite when setting it up. Despite mine device being second hand, purchased from a private individual and likely several years old, xRite helped me out getting the correct software to run and the license key required to do this. It's also way more convenient (once you get the hang of it) than using a scanner for the same job.

If you really want to think along the lines of the scanner thing: I'd hold better hope for a profiled output from a decent (modern) digital camera.

 

[Steven Lee]

@koraks I have a camera as well, in fact the targets were used for scanning slides with it. Maybe I'll just test the feasibility of the approach with B&W, because I have a great B&W densitometer I can use for reference.

 

[koraks]

Give it a try. I'm skeptical if it makes much sense to use B&W as a benchmark for color; in an RGB system it would mean you're still likely averaging out errors on the three channels and as soon as you mvoe to color, these errors will bite you.

I had an 'interesting' experience where I compared the cyan I got from a PB15:3 pigment with readings I found in one of Grier's manuals. Mine was off by a mile and seemed to be far more blue. Turns out the scanner was the problem and my cyan was just fine.

A fundamental issue is that I doubt very much how well the scanner (or camera) will perform well given the fact that it's still a three-stimulus capture and not an actual hue bin capture like a proper photospectrometer. This will create problems with especially saturated hues that don't align perfectly with the 3 sensitivity peaks of the sensor elements in a scanner or camera.

 

[Alan Edward Klein]

@Bill Burk But isn't this X-Rite version of what I already have? (Wolf Faust IT 8.7 target with Lumariver). I can make color profiles, but that's not the same as getting accurate RGB/CMY density values. Basically, I am stuck at the stage when I have a linear TIFF scan of a control strip, I compute RGB densities from the 16-bit RGB values, and now I need a known reference to error-check the result. I know that Kodak control strips come with a properly developed reference strip, but that's just one sample. I think I need variability to make sure my method is accurate.

I have a NEC calibratable monitor (PA242W) with the Spectraview II app and puck (I think the puck is from X-Rite). I also have an Epson V850 with the target chrome slide for the scanner. Can my setup be used?

 

[koraks]

Can my setup be used?

Yes, but:
1: Your calibration target is a chrome, so suitable for profiling your scanner for transparencies. If you want to measure reflective densities, you need a calibration target to construct a profile for reflective scans as well.
2: My comments concerning the (un)suitability of a scanner as a color densitometer remain the same. I'm skeptical this will work well. You'll get values, but they may be rather meaningless. Why bother if you can't trust the outcome anyway?

Btw, it's nice to have a calibrated monitor, but it's not relevant to this issue.

 

[Bill Burk]

Are you trying to use a scanner as a color analyzer like a Beseler PM type?

 

[Steven Lee]

@Bill Burk I just want to know if my C-41 and E-6 temperatures are within spec and I am on a budget. Control strips alone are $50-100 and I wouldn't want to spend more. Besides, it seems like a fun project.

 

[koraks]

The scanner approach won't be sufficient for this.

 

[Steven Lee]

@koraks There's always brute-forcing via interpolation. Say I assemble a dataset of known RGB densities and map them to the corresponding 16-bit RGB values taken from a camera RAW file produced with a known WB and a light source, I can extrapolate and map any future 16-bit reading to a guessed density.

The accuracy of such guessing will only depend on the dataset size. Maybe I should call my local lab...

 

[koraks]

You could indeed statistically approximate whether you're in a certain bracket. Of course, when you end up outside that bracket, you don't know exactly by how much. Still, it may be adequate.

But in all honesty, I'm not sure if it's necessary to go to these lengths. For slide film, a visual assay is sufficient to determine if your processing is good enough - after all, if there are no visible deviations from the desired output, there's no problem, even if you happen to be slightly outside the box. Shooting a test chart / scene once in a while or even on every roll should suffice, IMO.

For C41 film, I'd apply a similar logic: if it prints and/or scans fine, don't worry about it.

This is a deep hole you're going to dig if you want to do it 'right', and from a practical viewpoint, I really don't think there's much justification for digging this deep. Well, that's my 2 cts; please do let all of us know how you proceed with this. If you figure something out that works in an amateur setting, I'm sure many of us will find it informative!

 

[Radost]

Silver fast has one. Don’t know if it is any accurate.

 

[Bill Burk]

@Bill Burk I just want to know if my C-41 and E-6 temperatures are within spec and I am on a budget. Control strips alone are $50-100 and I wouldn't want to spend more. Besides, it seems like a fun project.

Ok! I think you can perform C-41 and E-6 process control with a scanner.

Make your own test strips using a sensitometer: (lots written here about that. I would jerry rig one from electronic flash and Stouffer scale T2115).

Occasionally send a test strip exposure off to a good quality lab and let them develop the film. This will be your comparison standard to meet.

Put a test exposure at the start of occasional rolls and finish the roll in camera.

Scan the standard and estimate the color densities. Scan a test strip that you exposed and compare the differences. Try to make them match.

Diagnosing problems will take another advisor’s experience because I believe the process control documentation is based on using standard test strips, you’ll have to extrapolate your own process control limits.

May as well get a pH meter and note the chemistry pH along with temperatures. It’s to see if something is out of control, not to adjust pH but to spot a possible culprit when things go wrong.

 

[Bill Burk]

Silver fast has one. Don’t know if it is any accurate.

VueScan also has a densitometer reading tool.

I don’t know how to get accurate density readings with a scanner. I tried the tool and it was way off.

 

[Alan Edward Klein]

VueScan also has a densitometer reading tool.

I don’t know how to get accurate density readings with a scanner. I tried the tool and it was way off.

Bill: Epsonscan software has a densitometer as well. I found the explanation here for an Epson Expression XL1640 scanner. The manual for my Epson V600 and V850 does not go into details but I assume the feature must be similar to this scanner.

The luminosity value is calculated by the following formula according to the explanation: 0.2R+0.6G+0.2B.

Does this explain anything of value that could be used?

Using the Densitometer

 

[Steven Lee]

@Bill Burk excellent advice!

 

[koraks]

I can confirm that the densitometer applet is also part of the older 3.04E Epson Scan software that works with the 4990 scanner:

Since it's 8 bit, it's not very precise. I doubt it'll suffice for color film calibration, but it's worth a try, I guess.

 

[Bill Burk]

I am going to call it cultural appropriation. The electronic use of the word “Densitometer” has been taken to mean “underlying measurement raw values”.

These are not density measurements in logarithmic terms.

VueScan’s is an attempt to estimate density in logarithmic terms, so it’s closer to what we would expect for this purpose, but I have not found its readings reliable.

 

[Bill Burk]

@Bill Burk excellent advice!

Thanks!

What kind of stuff have you already got?

I have toyed with color analyzers because they are so cheap. For instance Beseler PM type analyzers would have good linearity and a short density scale… enough to measure differences between standard and sample.

It’s hard to use them as absolute densitometers but I have a PM2M that has an analog scale to 2.0 (in 0.05 increment). I use it to do black and white contrast and print base time quick checks.

And all of the Beselers have a logarithmic density scale because color filtration always used density measurements.

I’d say the lowly PM1 could do it, But there are two PM3L on eBay right now, we’re talking thirty bucks and one has a module included. The module could be where you store the standard grays and then as you check your middle gray patch, you could read the difference quickly without having to get both negs out.

 

[Steven Lee]

@Bill Burk I have a digital cameras with a macro lense, an Epson scanner, Wolf Faust scanner calibration targets (transparencies) with profile-building tools, X-Rite color targets (for calibrating digital cameras) both color and a B&W scale, a Metz flash. I've been using the X-Rite greyscale patches for creating color inversion "profiles" (recorded PS adjustment layers) with good success, so what you're advising is very similar in principle. Also I have a Heiland B&W densitometer which can probably help with overall density control but not with temperature-related color crossover.

 

[Bill Burk]

Ok for a color analyzers you would need an enlarger. Skip that thought.

You could use the color reflection target with a macro stand and shoot tests in-camera with flash. Leave the rig setup all the time. Maybe dedicate a camera to the purpose and after popping off a few frames, rewind leaving the tail out and mark the leader so you know the film has a test on it. (I mark it “6 for sens” to tell me I already put a sensitometry exposure on it and I should wind to frame six before shooting. I am inconsistent and sometimes I mean I rewound on frame 30 leaving six at the end to add a sensitometry exposure. I don’t know how I remember which it means.)

Like before, send one roll off to a lab to get your standard.

 

[pwadoc]

I think it should be possible to built a homemade densitometer. You could potentially use a digital camera, but that might actually be overkill, and you would really need to carefully characterize the response to color response of the sensor. If you think about what a densitometer does in the simplest form, you'll see how a digital camera makes it more complicated. A transmission densitometer simply emits light of a given wavelength (for film we'd be using Status M, so ~440, 540, 660), and then uses a sensor on the other side of media you are measuring to detect how much light made it through. I believe in the case of the densitometer I use, the XRite 810, it shines white light through the negative, and then uses filters over the sensors to detect how much of the specific wavelengths made it through, but you could also just use LEDs that emit the required wavelengths. I've actually recently built a light that uses those exact wavelengths, so I happen to know that Luxeon makes emitters in those exact wavelengths in its 2835 color line.

I think the tricky part with a digital camera is the color lenses that cover the individual photo sites and compose the bayer array. The color response of those lenses determines the native response of the sensor, and you would have to carefully characterize how light of a the wavelengths you care about is measured by the camera sensor. It would be easier to use a camera without a bayer array, and easier yet to use a small, cheap black and white camera module like the ones they make for Arduino. Resolution doesn't really matter for the use case, you're just shining light through a spot in the negative and you only care about how much light of each color makes it through the negative. Hell, you could do away with a camera of any kind and just use a photo-transistor or some other type of cheap sensor. The most of the work is in calibrating the thing to get a result that would match what you would expect from a commercial device.