[eles]

Thinking about the new K3III Monochrome camera, and also about using filters with the new K3 Monochrome, some questions came to mind:

1. How different is the spectral sensitivity of the K3III Monochrome sensor (same sensor as the K3 color but without a Bayer array) from the spectral sensitivity of the K3III Color sensor with its Bayer array?

2. What color-compensating or temperature-correcting filter would bring K3 Monochrome sensor sensitivity closest to that of the K3 Color sensor? For reference, this link provides diagrams for sensitivities of two sample digital camera, with and without Bayer arrays: Figure - PMC

3. How different is the spectral sensitivity of the K3 Monochrome from that of the panchromatic film Tri-X 400? Will filters designed for black and white film produce similar effects on a digital monochrome camera? For reference, here is a link to Kodak's daylight spectral sensitivity curves for Tri-X 400 - scroll to the bottom of the page: https://imaging.kodakalaris.com/sites/default/files/files/resources/f4017_TriX.pdf

4. What color-compensating or temperature-correcting filter would bring the K3 Monochrome sensor sensitivity closest to that of Tri-X 400 film?

I don't have any way to answer questions 3 and 4, but I made a start on questions 1 and 2, thanks to the generosity of Luigi Barbano for allowing downloads of his sample files from his excellent article "Pentax K3 Monochrome vs K3 mkIII Color" (Pentax K3 Monochrome vs K3 mkIII Color - Luigi Barbano Photography). I compared the LCh "Lightness" values from the x-rite Color Checker shown in these two files:
* Color, aka "Katie per the nickname mentioned by Barbano": File "108 418.DNG" in zip file "K3MkIIITestRAW_Part1.zip"
* B&W, aka "Kimono": File "_IMG2378.DNG" in zip file "KimonoTestRAW_Part1.zip"

The Pentax Forums has kicked my attempt to post to a "not found" page three times now, so I'm trying one last time, this time breaking the post up into parts. Hopefully that will work.

[eles]

Continuing, I put together a spreadsheet showing the "Katie" LCh Chroma and hue values for the more saturated colors in the Color Checker chart in Barbano's file, and the LCh Lightness values for these same colors for both cameras. Specifically I sampled the squares in the first column on the left side of the Color Checker and the first three columns on the right side of the Color Checker. Counting the 8 columns in the Color Checker from left to right, that would be the first, fifth, sixth, and seventh columns, labeled "1", "5", "6" and "7" respectively in the spreadsheet.


The screenshot shows the values in the spreadsheet re-ordered by the LCh "hue" (https://handprint.com/HP/WCL/color16.html) values from the "Katie" shot, with magenta at the top of the spreadsheet, through red, orange, yellow, green, blue, violet and red-violet at the bottom. As already noted, there are two columns for LCh "Lightness", one for "Katie" and one for the Monochrome "Kimono". The "delta" column shows the difference betwen "Katie" and "Kimono" Lightness values. The last column indicates whether any given "Katie" square is lighter or darker than the corresponding "Kimono" square, and where the maximum deltas are located. Here is a screenshot of the spreadsheet:

---------- Post added 06-10-23 at 03:33 PM ----------



---------- Post added 06-10-23 at 03:31 PM ----------

Continuing on, here are some observations based on the spreadsheet:

* Tonal range for the more or less saturated colors: Looking at the bottom of the spreadsheet screenshot at the Max and Min Lightness values for Katie and Kimono, I was surprised at how much narrower the Range of tonalities for the saturated colors was for Kimono: Max minus Min equals 33.5 for Kimino, compared to 52.5 for Katie.

* Tonal range for the Neutral dark/grey/light colors: The spreadsheet does not show the Lab Lightness values for the Neutral colors in Column 8 (I did not measure values for Columns 2, 3, and 4). But before measuring any LCh values I did a slight exposure compensation on the Katie shot to bring the Lightness value for the brightest square (far right column of Neutrals from dark to light) up to match the same square in the Kimono shot. I used GIMP to do this, electing to use Lightness blend mode to avoid modifying Chroma or Hue. To my surprise, this adjustment made all six of the Neutral squares in Column 8 have very close to exactly the same values.

* Generalizing, the closer the scene is to containing all neutral colors, the less discrepancy there will be in the Lightness values when shooting the same scene with both cameras.

* The correlation between LCh hue and the "delta" betweeen "Katie" and "Kimono" LCh Lightness values is almost perfect, with positive delta peaking at hue 89, and negative delta maxing out at hue 287, that is, very close to the top and bottom of the LCh color wheel. And the minimum differences between Katie and Kimono are near Magenta (hue=0/360) and Cyan (hue=180) along the left-right axis on the color wheel. Sadly, these minimum differences are for colors (cyan, magenta) that are relatively uncommon in natural objects.

So what filter might stand the best chance of allowing the Lightness values for the K3III Monochrome to more closely match those of the K3III Color camera?

* According to the handprint link I gave above, the wavelength of yellow light near 90 degrees is 575 nanometers, which on the CIE 1931 color space diagram is pretty much where yellow stops being green and starts being warm (File:CIE1931xy blank.svg - Wikipedia). So maybe a yellow "CC" filter would "hold back" blues and violets, allowing yellow-green/yellow/orange/red to catch up, bringing Kimono sensitivity closer to Katie sensitivity.

* Maybe a temperature-correcting filter would do the job: Light on an overcast/cloudy day carries considerably more blue and less red than direct sunlight during the middle of the day, which is turn carries more blue and less red than direct sunlight at sunrise and sunset. So a color-temperature filter of an appropriate mired value (one of the warming filters) might bring Kimono sensitivity closer to Katie sensitivity.

* On the one hand adding a filter would decrease the amount of light reaching the sensor, negating the extra sensitivity of the K3III Monochrome sensor compared to the K3III Color sensor. On the other hand, adding the right filter (if there is one, testing required) would help with skin tones (reds are depressed by the Monochrome camera relative to the Color camera) and colorful flowers.

[eles]

The graphs I've seen for monochrome silicon sensors show a peak sensitivity more or less near the blue-green 500nm wavelengths, trailing off towards 0/near zero for lightwaves that are longer or shorter than 500nm. This "bare sensor sensitivity" is not anywhere close to human vision spectral sensitivity. Nor is it close to the spectral sensitivity of Tri-X film, which has three peaks of sensitivity, just like human vision. That's the whole point of panchromatic black and white film, to produce tonal values from lightwaves that more or less match the tonal values in the scene as seen by the standard observer.

A bare silicon sensor has a more extended range of wavelengths to which it is sensitive, than did the original "blue light only" early black and white film. But it is still a long way from matching the tonalities that humans see.

It seemed to me that a color photograph of a scene captured by the K3III Color camera with its Bayer array, after conversion to black and white using a straigtforward "Luminance" conversion, *necessarily* will have different tonal values than the same scene captured by the K3III Monochrome camera with its bare sensor.

I started this thread because I was curious about "just how different" the tonal values captured by the K3III Monochrome camera were, compared to the K3III Color camera. I was expecting to find that peak Monochrome camera Lightness values would be at colors close to 500nm (blue-green) and get progressively darker (relative to the Color camera Lightness values) as the color hues (in the scene and as captured by the Color camera) got further away from blue-green. If this turned out to be the case, it would suggest that a CC Magenta filter might produce Lightness values for the Monochrome camera, that were a closer match to those produced by the Color camera.

My expectation turned out to be completely wrong. In fact peak positive deltas between the Monochrome and Color camera was for blues and violets, with the warm colors from yellow-green to orange being noticeably darker than the Lightness values produced by the Color camera. I was quite surprised at how uniformly smooth the transition was from "hardly any difference near cyan and magenta" to maximum positive differences for blue-violet colors and maximum negative differences for yellow colors, referring of course to the colors on the Color Checker target chart in Barbano's raw files.

Nothing says one can't make great black and white images irrespective of the spectral sensitivities of the recording medium. But clearly the effect of the standard filters one might use for Tri-X film, won't be the same for a digital monochrome sensor. It would be pretty nice if there was a filter that brought the sensitivity of the Monochrome camera closer to that of the Color camera, allowing to take pictures with the Monochrome camera without warm colors being darker compared to how the Color camera captures them and violet-blue colors being brighter.

I was sort of hoping that someone with both K3III cameras, a tripod, a standard (Color Checker/Spyder/IT8/other) color chart, some appropriate CC or Temperature-correcting filters, some free time and ideally access to full-spectrum lighting preferably not too far from D50, might be interested in making some test shots to see if in fact one of the standard filters might bring the tonal values of the Monochrome camera closer to that of the Color camera. I could help with processing the raw files and digging out the Lightness information, if anyone wants to take some test shots.

[nightfly]

I've been interested in this question as well. Considering that panchromatic films record into the red end of the spectrum, and this makes use of red filters to dramatically alter rendering of a blue sky and clouds. Likewise for other color filters and different subjects.

One application of the K-3 Monochrome would be tri-color photography. By combining the three primary colors of three individual exposures taken through filters, one could assemble a color image. Understanding the sensor and it's response to light would be desirable in such pursuits.

Likewise in near infrared photography, as well as astrophotography.

Where would we find such information ?

Has Pentax published such a document ?

[Mack53]

I don't think Pentax has published the spectral response of the Monochrome sensor. Would be helpful - maybe.

I think the best way to use this camera is to try and separate the colors with regards to the tonal shades. I found Purples/Blues seemed to lower the contrast for some reason - and even removed a lot of shine on people's faces - and the Orange (B+W #040 Orange) seemed best to separate all the colors and supply more tonality overall. I shot the ColorChecker with sundry filters to see what and where the colors went to in the tones.


Trying to duplicate a film look I find often futile. Even editing software comes up with a different look between brands for a given film. Some seem close, and others not so much. And then it can differ greatly among viewers who act as film judges. There are 10 pages of people trying to emulate a "Kodachrome film look" on the m43 forum with over 129K views here: My "kodachrome/classic chrome" Pen-F settings (update) | Mu-43 and they still cannot agree on a "Kodachrome look" in SEVEN YEARS of trying to make suitable copy

Probably best to call it what you like - or what you think it resembles - and move on. Even if Pentax published a spectral response, people would no doubt say, "It looks more like Agfa developed in HC-110 for 10 minutes than Tri-X 400 in Rodinal," etc. Too many B&W variables compared to a color film that had control strips - and the digital emulators still cannot agree on that.