scanner as densitometer?

[markbau]

I'm getting back into B&W after a 10 year break and was wondering if anyone has figured out how to use a scanner and photoshop as a densitometer. I would think that once the scanner was properly calibrated it would be a simple matter to relate the values from 1 to 255 into neg densities.

Mark

 

[Anon Ymous]

Mark, assuming that you've found a way and you're 101% sure that the software acompanying the scanner doesn't play any funny games, then you'd also need a point of reference. You'll need something with know density, take a reading with that, then you might be able to use it as a densitometer. Keep in mind though, that some scanners have low Dmax. That could be another limiting factor.

 

[numnutz]

Vuescan available from www.hamrick.com has a densitometer facility. It seems to be reasonable accurate according to the only calibrated test strip I have.
But I think a lot depends on the scanner you will be using. I am using an Epson 4990 for scanning 5 x 4 and 10 x 8 B&W Negatives.

nn

 

[BetterSense]

You could scan a calibrated step wedge, and compare tones. To me that would seem to be about the only reliable way to go about it.

 

[domaz]

Vuescan does seem about right to me. You have to turn on Density display in the Preferences and then hold Ctrl and hover over the final scanned image to see the values. Probably just as accurate as a hacked up spot-meter densiometer

 

[Photo Engineer]

There are several recent threads on this subject that you might want to look through.

PE

 

[gainer]

My hacked up meter was considerably more sophisticated, but didn't start with a spotmeter. Unfortunately, my PC board negative along with instructions got lost when I went to the hospital with meningoencephalitis and my kids thought I might not make it back. I still use the meter, however, as an easel photometer for my enlarger.

I can tell you this: integrated circuit op amps are available that are near perfect in that they have nearly infinite input resistance and extremely high gain. The base-collector junction of a silicon transistor has a near perfect exponential voltage-current transfer function, which can be used to convert the op amp into a near perfect amplifier whose output EMF is, for all practical purposes, a constant times the logarithm of its input current. We now have phototransistors whose output current is a linear function of the illumination falling on a very tiny piece of silicon. We use two opamps in a bridge circuit to balance out most of the effects of temperature. One of these receives the output of the phototransistor, the other receives a voltage from the arm of a potentiometer which allows calibration. Now you have to figure out how to read the differenc between the outouts of the two opamps. It could ba an old fashined analog meter, or a self illuminated digital panel meter. These are handy for daarkroom work.

As to power supply, a small transformer with 24 volt center tapped secondary and a couple of IC voltage regulatos, various resistors and capacitors and a box to put everything in are all you need.

 

[gainer]

I forgot to say that I did all this because I found that the use of scanner output to measure density was unsatisfactory because sensitivity varied over its range and the range was too small. We want to be able to analyze a negative that has in its total range more than printing paper can hold, such as when you want to see what is outside a window as well as what is in the room.

 

[BetterSense]

A clever way to obtain a logarithmic reading via analog means. Another way would be to get linear readings straight from the phototransistor and do the math separately, like with a calculator.

I'm trying to build densitometer functionality into my microcontroller-based shutter speed tester/enlarging lightmeter, using the same photodiode. I can just do the math in software. Long term stability/calibration shouldn't matter much, if you calibrate to a clear piece of your film first, correct me if I'm wrong. It's in working condition now, but I have no way to verify its accuracy. I need another densitometer, or a calibrated piece of film.

 

[RalphLambrecht]

You could scan a calibrated step wedge, and compare tones. To me that would seem to be about the only reliable way to go about it.

That works reasonably well for relative densities, but one needs to remember to turn all auto functions off, otherwise, the scanner software will shift tones around and mess it up.

 

[glarsson]

My humble Epson 4490 works quite well as a densitometer over the density range I need, 0.15 to about 2.2. I did an initial calibration by scanning a pre-calibrated step wedge and creating a Photoshop curve to convert scanned values to densities. Scanned densities agree with a dedicated film densitometer to within +/- 0.02 which is plenty for me. Density readings from the scanner are repeatable to within the same tolerance (+/- 0.02) as long as I let the scanner warm up for about 20 minutes first.

As others have said, it is critical to turn off all auto functions. I use VueScan's raw mode.

I scan at 16 bits per channel to capture enough bits in the densest areas (lowest scanned values). Eight bits per channel won't cut it. Applying a gamma=3 in Photoshop before applying the calibration curve expands the range of the lowest scan values (greatest density) and compresses the highest values (least density). Otherwise the low values are too close together for a good calibration curve. I set up my curve so that pixel values of 0-255 represent densities of 0.00-2.55 (although the scanner doesn't actually reach the extremes). I have a Photoshop file with the layers ready to go so I just copy in the scanned image and read out densities.

I also have Photoshop layers that convert negative density to print values for various contrast filters based on a standard printing exposure. I simulate different printing exposures by adding or subtracting a constant to density values. For example, adding +30 to all pixels is equivalent to a density increase of 0.30 which is one stop less printing exposure. This method helps me visually come up with a good starting point (contrast filter and printing exposure) to make better use of precious darkroom time. It was a bear to set up but comes in handy now.

 

[RalphLambrecht]

My humble Epson 4490 works quite well as a densitometer over the density range I need, 0.15 to about 2.2. I did an initial calibration by scanning a pre-calibrated step wedge and creating a Photoshop curve to convert scanned values to densities. Scanned densities agree with a dedicated film densitometer to within +/- 0.02 which is plenty for me. Density readings from the scanner are repeatable to within the same tolerance (+/- 0.02) as long as I let the scanner warm up for about 20 minutes first.

As others have said, it is critical to turn off all auto functions. I use VueScan's raw mode.

I scan at 16 bits per channel to capture enough bits in the densest areas (lowest scanned values). Eight bits per channel won't cut it. Applying a gamma=3 in Photoshop before applying the calibration curve expands the range of the lowest scan values (greatest density) and compresses the highest values (least density). Otherwise the low values are too close together for a good calibration curve. I set up my curve so that pixel values of 0-255 represent densities of 0.00-2.55 (although the scanner doesn't actually reach the extremes). I have a Photoshop file with the layers ready to go so I just copy in the scanned image and read out densities.

I also have Photoshop layers that convert negative density to print values for various contrast filters based on a standard printing exposure. I simulate different printing exposures by adding or subtracting a constant to density values. For example, adding +30 to all pixels is equivalent to a density increase of 0.30 which is one stop less printing exposure. This method helps me visually come up with a good starting point (contrast filter and printing exposure) to make better use of precious darkroom time. It was a bear to set up but comes in handy now.

Sounds like a good approach.