UPDATE: The following article has been revised to reflect newly-optimised Lightroom settings and example images. For a limited time, you can also download the CCD Effect preset for Lightroom HERE.


Get "that CCD look" with the K-3 / K-3II and Lightroom

Many people feel that the colour rendition from Pentax's earlier CCD sensor cameras is superior to that of later CMOS sensor models. Although the K-3 and K-3II are my day-to-day cameras, more recently I've been shooting extensively with a GX-10 - Samsung's clone of the Pentax K10D - and I've become a huge fan. Whether it's down to the sensor, the camera's colour profiling or a combination of the two, I really can't say for sure; but, the results are very appealing - punchy, saturated, almost film-like images, with very little post-processing needed to achieve great-looking results.

So fond am I of the GX-10's output that I decided to develop a Lightroom preset that would re-create "that CCD look" for my K-3 and K-3II photos.

I started by taking two photographs of an X-rite ColorChecker Passport, one with the GX-10, the other with the K-3. Both cameras were fitted with the same model of lens (the Pentax-F 28-80 f/3.5-4.5), to avoid optical differences in contrast and colour reproduction. The photos were taken in RAW format using the DNG file type (which, importantly, embeds a copy of the camera profile).

I imported both photos into Lightroom 6 and ensured all settings were at Lightroom's default values, with no presets or user defaults applied. I then selected the "Embedded" profile for each photo (to use the camera profiles rather than Adobe's), and set the white balance for both images using the eye-dropper tool on the same mid-grey square of the Passport.

From this point, all adjustments would be to the K-3 image alone, trying to match against the GX-10 "master" image as closely as possible.

I fine-tuned the exposure level so that the mid-grey tones were at the same luminosity for both photos (within +/- 0.5, as there was some variance of values across the square).

I adjusted the contrast so that the "darkest black" and "brightest white" squares had the same luminosity. This required some minor tweaking of exposure to keep the mid-tones at the right level, as contrast adjustments appeared to have a non-linear effect on the tone curve. Now, each of the grey-scale squares showed luminosity values very close to those in the GX-10 image.

The remaining adjustments would deal with colour reproduction - specifically hue, saturation and luminosity for each colour.

In the Camera Calibration section of Lightroom, I adjusted the hue and saturation of the Red Primary, Green Primary and Blue Primary channels to get those primary colours as close as possible. Since there is no luminosity adjustment for these, it's impossible to get them exactly right - but we can place them in the ball park.

At this point, a casual comparison of the colours in both images was already much closer than before.

Next came the really time-consuming part...

In the Color section of Lightroom, I adjusted (and re-adjusted!) the hue, saturation and luminosity of each colour to achieve a close match for each coloured square in turn. As you'd expect, adjustments for each colour had a knock-on effect to one or more of the others, and there was a great deal of back-and-forth fine-tuning required. Small, incremental adjustments were vital to avoid significant impact on related colours, which had to be tweaked to counteract any minor changes.

I was unable to get every coloured square matching exactly, but it was very close.

I saved the adjustments as a user preset, checking the Contrast, Color Adjustments, Process Version and Calibration boxes to ensure all the relevant settings were saved.

Then, on a sunny day with no clouds (rare in my part of the world!), I took some test shots of real scenes side-by-side with both cameras, and tried out the new preset on the K-3 files. The results were good, though not quite as good as I'd hoped - in particular, the green and yellow balance wasn't quite right, and the saturation and luminosity of light-blue skies wasn't what it should be. So, I went back to the ColorChecker Passport images and fine-tuned the adjustments before re-applying them to the test shots. After many iterations over several days, I eventually reached a point of diminishing returns; the results were so close that further tweaking would have little additional benefit and more than likely be detrimental.

Here's a "before" and "after" example of the preset being applied to an indoor test shot of some coloured pencils:

What's interesting about this example is how little difference there is in the greens, as vivid greens are something the CCD sensor cameras are revered for. From my testing in normal shooting conditions, it seems this is largely due to luminosity and saturation of yellow and orange rather than a radical difference in green tones.

It's worth studying that example image in detail. At a glance, you'd be forgiven for thinking the adjusted image is merely brighter, with a little more contrast. But if you look at each pencil individually, you'll see some quite significant changes in hue, saturation and luminosity, while the white, black and grey shades are (as they should be) nearly identical in both shots

Since completing this exercise, I've applied the preset to a number of K-3, K-3II, and even some K-5 images from my Lightroom library. The outcome is just what I'd hoped for... the photos have more of "that CCD look" I've come to appreciate so much. Greens and browns are warmer (great for landscape work), blues are richer, light-blue skies are somewhat deeper and better defined against clouds, yellows and oranges are brighter, while reds are a real treat - more orange than scarlet and not so over-saturated.

Of course, no preset or any amount of post-processing can re-create the fun of using a particular camera. For those who've never owned one of Pentax's CCD-sensor models (or their Samsung cousins), I highly recommend picking one up at the right price, since - at lower ISO settings - they produce wonderful images. I'll continue to use my GX-10 regularly, as it's a great bit of kit - but these adjustments provide a solid basis for reproducing at least some of that signature CCD look with the K-3 and other Pentax CMOS sensor cameras.

For those who'd like to try this out, here are some screen captures of the relevant settings (remember to save them as a preset so you can apply them all in one go!). And, if you do try it, please let me know how you get on!

... and, just to finish, here's one more test shot from the K-3, taken in my back garden, with the CCD preset applied (and no other adjustments, save for exposure and white balance sampled from a known grey area). It certainly has the warmth and saturation I was hoping for

Thanks for reading!

[BigMackCam]

I use DxO OP 11 as well. As a quick and dirty test, I re-exported a photo that I had processed with the default camera profile (K-S2 in this instance), with only one change: selecting the camera profile for the K10D.

First photo is using the K-S2 profile (which, curiously, is also for the K-5ii, K-5iis and K-500, but not the K-S1, which has the same sensor). Second is using the K10D profile. The blues are the same to my eye, but the yellows and reds are more subtle and realistic in the latter. Green is a little different too, of course. I like the K10D profile - I'll use it again.

Interesting... thanks for posting this, Des.

I'd have expected the reds in the first photo to have a far more magenta-like hue than in the second, but that's not the case

I see a much greater difference in reds between my K-3 and GX-10 using Lightroom and the embedded profiles... the K-3 (and K-5 too; even the Q7) tend to produce a saturated and distinctly scarlet hue, while the GX-10 produces orange-like reds with greater luminance but less saturation. It's one of the more extreme differences in the colour rendering, in my experience.

Here's a side-by-side comparison of reds using the embedded profiles:



There's a big difference in the reds here, but I'm not seeing anything like that with the K-S2 example you posted. It makes me wonder if DXO is actually providing normalisation between its own camera-optimised profiles, rather than matching between the various cameras' native embedded profiles. Just a hunch; I could be wrong...

And now, that same comparison with my "CCD look" adjustments added to the K-3 image:



The reds are pretty well matched now, and the greens are very similar, if not quite spot-on...

---

Last edited by BigMackCam; 08-12-2017 at 02:08 AM.

[Dartmoor Dave]

Fascinating article, Mike, and a deeply interesting conversation thread too. You're making me wish I used Lightroom to try out your profile.

I always use custom profiles for my own cameras created using a Macbeth (Colorcheck) card and Adobe Profile Editor, so I'm convinced that the different look of CCD sensors is baked into the hardware somehow, since it's still there even with my own profiles. But the samples posted in this thread look like you're getting darn close to replicating the CCD look with CMOS sensors. And that's an unequivocably good thing!

I think I might be downloading the Lightroom trial version over the weekend. . .

[BigMackCam]

Fascinating article, Mike, and a deeply interesting conversation thread too. You're making me wish I used Lightroom to try out your profile.

I always use custom profiles for my own cameras created using a Macbeth (Colorcheck) card and Adobe Profile Editor, so I'm convinced that the different look of CCD sensors is baked into the hardware somehow, since it's still there even with my own profiles. But the samples posted in this thread look like you're getting darn close to replicating the CCD look with CMOS sensors. And that's an unequivocably good thing!

I think I might be downloading the Lightroom trial version over the weekend. . .

Thank you kindly, Dave - that's great feedback from a dedicated K10D user

[Des]

It makes me wonder if DXO is actually providing normalisation between its own camera-optimised profiles, rather than matching between the various cameras' native embedded profiles. Just a hunch; I could be wrong...

You could be right too, Mike. It does seem strange to me the way DxO bunches the profiles. One profile is common to these: K-5, K-7, K20D, K-30 and K-01. Whereas the profile for the K200D (which you would expect to be the same as the K20D) is the same as for the K-m, K-x and K-S1. I'd be surprised if the rendering from the K20D was really the same as that from the K-30, for example. The profiling does start to look a bit dubious.

[luftfluss]

You could be right too, Mike. It does seem strange to me the way DxO bunches the profiles. One profile is common to these: K-5, K-7, K20D, K-30 and K-01. Whereas the profile for the K200D (which you would expect to be the same as the K20D) is the same as for the K-m, K-x and K-S1. I'd be surprised if the rendering from the K20D was really the same as that from the K-30, for example. The profiling does start to look a bit dubious.

I have no idea how DXO came to some of their decisions wrt camera profiles, but I believe they are based on the camera's jpeg output. Some of the profiles fit the stereotypes - Canon pushing reds, Oly with cyan skies, Minolta strong green and blue, etc. - but I would hardly trust them to be exact. I guess the important thing is that once you settle on a profile you like, it should be able to create similar color response across multiple cameras (obviously lenses will effect this beyond what DXO can account for). In my experience, the profiles work consistently among my K-01, K-5 II, and Oly EM10, but all three cameras have pretty much the same sensor tech so there's not much deviation to account for, I think.

Sometimes I lose myself a bit, flipping thru all the camera & film simulations.

Anyway, I've found that the "middle of the road" DXO color renderings - generic, some Pentax, Canon, Nikon, Sony, etc - often benefit (to my taste) from an application of BMC's "CCD look". It's a real gift he's given the PF community...

[audiobomber]

You could be right too, Mike. It does seem strange to me the way DxO bunches the profiles. One profile is common to these: K-5, K-7, K20D, K-30 and K-01. Whereas the profile for the K200D (which you would expect to be the same as the K20D) is the same as for the K-m, K-x and K-S1. I'd be surprised if the rendering from the K20D was really the same as that from the K-30, for example. The profiling does start to look a bit dubious.

The K200D had a Sony CCD sensor, the K20D was a Samsung CMOS.

IMO the K20D had the best colour rendition of all Pentax bodies. The CCD bodies were too yellow (which some prefer), 16mp and 24mp Sony sensors too blue, K20D just right. Too bad DR on the K20D is so poor.

[swanlefitte]

I had a chance to try the new settings. Interesting how the settings move around to get such a subtle difference. I do believe I like this version even better. I saved both versions though.
I took my paint tube picture and added the new setting to the tops and bottoms so you can compare both the two versions to each other and to the original.
This time I saved the jpg at 100% instead of my usual 45%. I will note again this was on a K-s2.

[BigMackCam]

I had a chance to try the new settings. Interesting how the settings move around to get such a subtle difference. I do believe I like this version even better. I saved both versions though.
I took my paint tube picture and added the new setting to the tops and bottoms so you can compare both the two versions to each other and to the original.
This time I saved the jpg at 100% instead of my usual 45%. I will note again this was on a K-s2.

That's an excellent way of showing the subtle differences, Chris (especially when clicking on that image to bring up the Flickr copy) - thanks for posting this!

The newer version of the adjustments is, I believe, much more accurate, although many might find the changes between versions to be extremely subtle. At any rate, I'm very pleased with the current settings.

Frankly, I'm sure a lot of people will find the entire set of adjustments more subtle than they'd expect, but the truth is that the colour reproduction of the older cameras isn't wildly different; it's just "different enough" to have a certain something I find lacking in the newer models (though, of course, this is subjective)...

[Des]

The K200D had a Sony CCD sensor, the K20D was a Samsung CMOS.

Thanks for that. I should have checked.

I have no idea how DXO came to some of their decisions wrt camera profiles, but I believe they are based on the camera's jpeg output. Some of the profiles fit the stereotypes - Canon pushing reds, Oly with cyan skies, Minolta strong green and blue, etc. - but I would hardly trust them to be exact. I guess the important thing is that once you settle on a profile you like, it should be able to create similar color response across multiple cameras (obviously lenses will effect this beyond what DXO can account for).

It's fun, and of some value, but plainly not a way to achieve the same K10D CCD effect that BMC is developing in LR.

[Topsy]

I give up for now..

The saturation levels of the GX10 photo in C1 seem to be so far off that after matching the colours on the test shots and applying the preset to other things it really doesn't look good at all

[Dartmoor Dave]

Okay, I've obtained a Lightroom test version and given the preset a tryout, using a quick handheld GX-10 vs K-S1 comparison shot from last year that I had handy. Both shots are with the Super Takumar 20mm at identical manual exposures, and the CCD emulation effect is astonishingly convincing as far as I can tell on my Spyder calibrated monitor. I am awestruck by what you've achieved here, Mike. Your preset is utterly, staggeringly, brilliant work.

The GX-10 shot is at the top and the K-S1 at the bottom:

[BigMackCam]

Okay, I've obtained a Lightroom test version and given the preset a tryout, using a quick handheld GX-10 vs K-S1 comparison shot from last year that I had handy. Both shots are with the Super Takumar 20mm at identical manual exposures, and the CCD emulation effect is astonishingly convincing as far as I can tell on my Spyder calibrated monitor. I am awestruck by what you've achieved here, Mike. Your preset is utterly, staggeringly, brilliant work.

Dave - thank you so much for your very kind comments, and especially for taking the trouble to download Lightroom and test the preset on your photos. It's great to have confirmation from another GX-10 / K10D user with a real-world example, and I'm delighted you like the adjustments!

[Lake]

I downloaded your preset. It adds a very welcome warmth and vibrancy to my K5 images. Pale skin tones are also much improved. I'm sure I will use it frequently. Well done!

[BigMackCam]

I downloaded your preset. It adds a very welcome warmth and vibrancy to my K5 images. Pale skin tones are also much improved. I'm sure I will use it frequently. Well done!

That's great to hear - I'm so glad you like it! Thank you for the nice feedback

[savoche]

The camera is still quite new to me and I have had precious little time to use it, but initial testing seems to render nice results for the K-1 as well. I don't have any K10D images to compare with, though. I'll see if I can swap my K-5 for my father's K10D later on