Ceci n'est pas une fleur bleue.

[JMC]

Not sure how important this is here, but will throw it into the discussion. Some of the Nikon cameras I've measured the spectra response for tend to have a bit of red sensitivity even well down into the blue wavelength range.

 

There is a graph for the d800 shown in here, which is unfortunately behind a paywall;

https://www.spiedigitallibrary.org/journals/Journal-of-Electronic-Imaging/volume-25/issue-6/061609/Objective-colorimetric-validation-of-perceptually-realistic-rendering--a-study/10.1117/1.JEI.25.6.061609.short?SSO=1&tab=ArticleLinkReference

 

Here's the graph, which I'm comfortable sharing as I've referenced the source article.

 

The d800, d850 (and the Sony A7III) I've measured all showed this behaviour - red channel response even in the blue region. Not all cameras behave the same though - for instance my Canon EOS 5DSR does not show the same type of response from the red channel in the blue region.

[colinbm]

I wish they would make these graphs out to say 300nm.

[JMC]

Colin, don't forget this is done with all the internal filters in place (hence nothing above 700nm). With those there everything is flat down to 300nm. When I do my measurements I typically do a with and without the filters to see what the effects of conversion are. Unfortunately the d800 data I have is for a client and I cannot share the graph at the moment due to confidentiality.

 

But you can see my graphs for the Sony A7III here which show the sensitivity curves for that before and after conversion (the two links to my website) - https://www.ultravioletphotography.com/content/index.php/topic/3608-sony-a7iii-conversion-to-multispectral/page__view__findpost__p__31578

[dabateman]

My point is to look at lee congo blue filter #181 and compare it to Tokyo blue filter #071. See how #071 is much more blue and the main difference is the loss of the red peak.

 

 

https://www.leefilters.com/lighting/colour-details.html#181&filter=cf

[colinbm]

Thanks Jonathan

The second part has the graph,

https://jmcscientificconsulting.com/imaging/project-mirrorless-part-2-the-camera-conversion/

down the bottom of the page, that I want to see, & is labelled "Spectral sensitivity modified Sony A7III between 280nm and 420nm".

This shows how the false colours, in UV photography, are made from with the mixing of the RGB responses.

[Andrea B.]

All sorts of good responses after I left last night !

 

I'm off this morning to make further tests to try to find the reason for the purple Blue Flax.

*****

 

I don't think a color profile can correct colors when you have captured too much deep red.

 

Agreed. For the current topic we are only wanting to correct any visible drift off the normal. If you are capturing near-UV and near-IR, then all bets are off when making CC Card profiles in Photo Ninja.

[Andrea B.]

Experiment:

 

This post will contain 3 sets of photos made about 1 hour apart. The purpose is to check the temperature "drift" and color profiling at 3 different times when using a converted camera for visible photos with a Baader UV/IR-Cut Filter. Additionally, the purpose is to try to get an accurate color for Blue Flax flowers.

 

Gear:

• Nikon D610-full spectrum conversion + Coastal Optics 60/4.0 + Baader UV/IR-Cut Filter in Sunlight
  
• Spectralon 5"x5" rectangle
  
• Color Checker Passport
  

.

Three main photos per set:

• CC Card against Spectralon rectangle used for creating Custom Light profile in Photo Ninja.
  
• Blue Flax flowers with CC Card & Spectralon background.
  
• Blue Flax flower close-up.
  

.

Profiles:

Custom Light profiles were made in Photo Ninja for the 3 different shooting times.

This included white balance measurement against the Spectralon.

• Blue Flax Profile #1 made at 8:11 AM Mountain Daylight Time (UTC-6)
  
• Blue Flax Profile #2 made at 9:07 AM Mountain Daylight Time (UTC-6)
  
• Blue Flax Profile #3 made at 10:06 AM Mountain Daylight Time (UTC-6)
  

.

Doofus Move:

The profiles were actually made at 9:11 AM, 10:07 AM, and 11:06 AM. The times are wrong in the files because I failed to reset my D610 when Mountain Daylight Time began. I only just realized this.

 

"It's always something."

^{Roseanne Roseannadanna, 1970s.}

 

 

Conversion:

The file was opened in Photo Ninja. I only made basic adjustments.

• Color Correction: The appropriate color profile for each set was selected.
  
• Exposure & Detail: Black & white points checked and adjusted if needed. Detail +15. Illumination slider for increased light when needed.
  
• Color Enhancement: Plain +60.
  

Resizing and labeling was done in Photo Mechanic. No sharpening or other edits were made after the resize. I didn't want to accidently skew any color.

 

 

Comment:

The Blue Flax color is much better in these three sets. My tentative conclusion is that the Blue Flax flower in the very first post simply has a variation in its color. You will note immediately that one of the flowers in today's photographs also is not quite as blue as the other flowers.

 

That said, I still don't think that the biggest blue flower in today's photos is quite as blue as it should be, but it is close enough. It is still a teensy bit "off". But I can live with the results in these three sets.

 

IMPORTANT:

The suggestions above that the Baader UV/IR-Cut Filter might be passing too much red seems more possible to me now that these results rule out a problem from making a profile photo to apply to a photo made two hours later.

 

 

SET 1

---

 

SET 2

 

Note natural variations in the flower colors.

---

 

SET 3

I had to move the set-up a few inches to the right for Set 3 because a tree shadow was encroaching.

 

Note natural variations in the flower colors.

---

 

 

SANITY CHECK

Make some snapshots with unconverted stock gear.

• Gear = Nikon Z7 + 24-70/4S.
  
• White Balance = Natural Light Auto.
  
• Picture Control = Standard (Sat +2)
  
• Exposure = f/11 for 1/250" @ ISO-100
  

These photos are JPGs saved from the NEF.

After crops and resizing in Photo Mechanic, some sharpening was added in Capture NX2.

 

The slight color variation can also be seen in these stock photos. But all the flowers are bluer.

As noted above in the IMPORTANT comment, I coming around to thinking that the Baader UV/IR-Cut filter might pass too much red. But I'm not completely sure about that.

 

[Andrea B.]

I made a couple of color samples from the Z7 Snapshots to get an idea of how much variation there is in the Blue Flax flower color. Two samples do not make for any kind of grand statement. But there is an obvious minor color variation which is borne out by sampling. I'm only reporting two samples here.

 

The darker color dot in each photo shows the sampled color at full saturation. The blue flower sample reads 237°and the not so blue flower sample reads 248°. Blue is at 240°. Purple is at 270°. Remember Abert's Effect though. The 248° flower is only 33% saturated so our brain begins to interpret its color as more purple than it really is.

 

 

 

 

f

 

---

 

And here is a color sample from the D610 + CO60 + Baader UV/IR-Cut photos.

These colors are about 10° different from the Z7 samples.

 

 

So let's fix that by rolling Magenta back -30 steps in the Photo Ninja Color Enhancement tool.

Now the two flowers' colors are pretty much what they should be

Whew, lots of work......

 

As compared to the Z7 photos, brightness also needs a bit of a raise in the D610 photos. That's no big deal though. There were different angles of shooting.

[Stefano]

So, there are two options:

- use a better filter;

- use the Baader UV/IR cut and roll back the colors.

 

If I had to choose, I would choose the former.

[Andrea B.]

but what is a "better" filter?

[Andy Perrin]

Any filter that has a decent shoulder. The less correction that needs to be done the better. Cadmium has been suggesting BG38 2mm for years and it's worked fine for me. Sometimes I also add a UV cut.

[Stefano]

A filter close to a stock camera filter stack is the best. Stock cameras are meant for visible light, and are quite good at reproducing colors.

[Andrea B.]

BG 38 x 2.00 mm

 

In-camera WB only, SOOC.

 

 

Converted in PN with BG 38 color profile.

 

 

BG 39 x 2.00 mm

 

In-camera WB only, SOOC.

 

Converted in PN with BG 39 color profile.

 

 

BG 40 x 2.00 mm

 

In-camera WB only, SOOC.

 

Converted in PN with BG 40 color profile.

[Andrea B.]

Now to apply those BG profiles to the blue flower photos.

 

 

 

 

 

 

 

And for reference here is one of the Baader UV/IR-Cut profiled photos from above.

[Andrea B.]

A bit of color sampling on the BG photos to see how close we came to the goal of being a few degrees around 240° as shown in the Z7 photos above which look like the actual flowers more than any other attempt today.

 

I know some of you had faith that the BG filters would work out better, but it appears that they did not in this particular case. For these Blue Flax flowers all the BG filters are not as good as the Baader UV/IR-Cut filter.

 

BG-38 x 2.00 mm

 

BG-39 x 2.00 mm

 

BG-40 x 2.00 mm

 

Baader UV/IR-Cut (no rollback of magenta)

[Andy Perrin]

You have to do it in conjunction with the cut filter or you get UV contamination and a bit of IR. This is what I would have expected if you don’t cut the out of spectrum also. Put the Baader on in addition and let’s see how the combo fares.

[Andrea B.]

Hey, I'm stumped.

I've tried hard to make these profiles correctly and test them.

If I might have made some error, please let me know. And thank you in advance!

 

Tomorrow I can try the S8612. But I'm not particularly optimistic about it working either.

[Andrea B.]

Andy, I didn't see your post till just now. But it's not clear. What is it that I should stack? Baader UV/IR-Cut + BG glass??

Thx.

[Andy Perrin]

Yes that’s exactly what I mean because right now the BG is letting through all the UV and some IR since it’s not a sharp cut. When I do it I’ve been using my MidOpt vis cut filter, which is squareish, with the BG38.

[Shane]

Andrea, you might find this article of interest:

Color Accurate Digital Photography of Artworks by Robin Myers, Better Light Inc, 2000

It discusses some of the issues with blue.

[Andrea B.]

Thanks, Shane, for the reference. Cobalt blue is an interesting pigment. I'll have to ponder whether there is a lesson there applicable to this Blue Flax flower. Could be! It was also interesting to read how they solved the problem.

 

**********

 

I've noticed something else about this Blue Flax flower. It seems to become less blue as the day progresses.So there is something going on with the Blue Flax pigment. If not a pigment "change" of some kind, it could be that as the flower ages during the day (Blue Flax flowers only last a day), its surface characteristics change and the flower no longer either reflects/transmits light the same way it does in the morning.

 

 

**********

 

I said I would try a Baader UV/IR-Cut + S8612 stack. But conditions deteriorated before I could catch the best light. I played around with the stack anyway. Right after I made the profiling photos and placed the blue flowers in front of the lens, a gusty breeze blew most of the petals off. whee...!!!!!. It's always an "interesting" experience photographing outdoors in New Mexico.

[Andrea B.]

D610 + Novoflex 35/3.5 + Baader UV/IR-Cut + S8612x2.0 + Sunlight

 

This is *not* truly comparable to the preceding photos because I was using a different lens. But while I had it out I tried the double stack. There seems to be some improvement, but it still did not produce the bluer looking flowers quite like the Z-7 unconverted photos. (It is the Z-7 photos which I thought most matched the actual flower color.)

 

The 236/237° values are good. But what should be the lighter blue is still about 10° off (252/258°).

 

 

[Stefano]

Did you use the profile for your camera?

[Andrea B.]

yes. I make a new profile for every shooting situation.

[Stefano]

That is odd. You blocked UV, and you blocked IR with a nice slope. Did the camera have a bluer BG filter when stock?