Five IR-Blockers Tested for Visible Color Use on Full Spectrum Camera

[Andrea B.]

Update: 21 Aug 2021 Added links.

Update: 26 Aug 2021 Minor edits for clarity. Title improvement. Added Goal and Test statements.

---

 

 

CONCLUSION: Post #38

Don't go there immediately.

Read thru the experiment and enjoy the s*u*s*p*e*n*s*e, yes?

 

 

---

Camera: Nikon D610-full spectrum conversion

Lens: UV-Nikkor 105mm f/4.5

Light: Ambient afternoon skylight

Filters: In alphabetical order, Baader UV/IR-Cut, BG38, BG39, BG40, S8612.

This is not the order in which the filters were tested.

 

GOAL:

Which of these 5 filters is best for blocking IR light to give the best Visible color when used with a full-spectrum converted camera?

 

 

FIRST TEST:

During conversion of the photos, apply white balance only to the white or grey patch but do not make any color correction profiles.

 

There follow below 6 Visible light photos. Five (5) were made with IR-blockers and one photo is unfiltered.

 

Conversion was in Nikon Capture NX2 so that a stock Nikon D610 color profile would be used.

 

Capture NX2 can read the picture control settings. Those were Neutral 0. That means there was no extra contrast, brightness or saturation applied when making the photos. And no hue set was given any kind of preferential boosts as can happen with blue and green when using Landscape picture controls.

 

I used three edits only.

• The Capture NX2 white balance marquee was applied to the white patch (upper right).
  ADDED: 3 Apr 2024 Subsequent experiments have shown it is best to use one of the grey patches for WB.

• The black and white points were set using the LCH Tool. That is, I moved the histogram end points as needed.

• The brightness of the black patch (upper left) was set to 5% using the Black "Luminosity" dropper. The brightness of the white patch (upper right) was set to 95% using the White "Luminosity" dropper. My thought was that uniformity in the darkest and brightest tones would overcome any minor exposure differences and make color analyses easier.

Summary: White balance + stock camera profile.

 

 

Which photo has the best Visible colors?

Define "best" however you like.

 

#1

 

 

#2

 

 

#3

 

 

#4

 

 

#5

 

 

#6

[Andrea B.]

Camera: Nikon D610-full spectrum conversion

Lens: UV-Nikkor 105mm f/4.5

Light: Ambient afternoon skylight

Filters: In alphabetical order, Baader UV/IR-Cut, BG38, BG39, BG40, S8612.

This is not the order in which the filters were tested.

 

 

SECOND TEST:

During conversion of the photos, apply both white balance and a color correction profile.

 

There follow below 6 Visible light photos. Five (5) were made with IR-blockers and one photo is unfiltered.

 

Conversion was made in Photo Ninja.

A neutral picture control, Plain 50, was set.

While balance was measured over the white patch.

Then a Custom Light Profile was created for each photo.

 

Note that: A custom profile which incorporates white balance is good only for the particular camera, lens, filter and lighting used during a given photo session. My experience has been that moment to moment variations in Sunlight do not affect the application of a profile made at the beginning of a session to photos taken within a limited time range of that profiling step. However, if your outdoor shooting session is a long one, then the CC Passport should be re-photographed a few times during the day.

 

After application of the Custom Light Profile, the photo was exported as a TIF. I then sent all six TIFs through Capture NX2 to apply Black & White Point adjustments and the two Luminosity Droppers as explained above.

 

Summary: Custom color profiles incorporating white balance.

 

Which photo has the best colors?

 

The filter order is the same as in the first post.

For example the filter used in photo #1 above is the same filter used in photo #11 here.

 

#11

 

 

#22

 

 

#33

 

#44

 

 

#55

 

 

#66

 

 

---

 

Are most of the photos in this second set more alike one another

than those in the first set?

 

---

 

I did not know for certain-sure exactly how this IR-blocker filter comparison would come out. It is possible that custom color profiles cannot rescue every IR-blocker.

 

---

 

Impatient readers can skip to Post #31 for the IDs of the filters.

 

.

[Andy Perrin]

Are we supposed to guess which photo goes with which IR blocker? Like a blind taste test?

 

I like the colors in numbers 1 and 2, maybe a slight preference for 1. Color blindness may be an issue here…

 

By the way, all the color science stuff says you are supposed to white balance against the 18% gray, which is the second square from the white, not the white patch itself?

 

Counterintuitive, but that’s what the directions say.

[Stefano]

I would like to find which photo has the most accurate colors. Does "best" mean more vibrant/saturated? That may not be equivalent to "most accurate". But the only way to do this is to chech the subject with your eyes.

[dabateman]

3 is my favorite, but not enough red patches. Do you have an IT8 color target?

Those are better.

[Andrea B.]

Andy, guessing the filter might be fun. But I don't think you could do that in the 2nd set.

 

The point I've tried to make for at least the last 1000 years (snork, snork) is that white balance alone is not enough to restore good color to a Visible photo made with some kind of IR-blocker on the lens of a converted camera. So I rustled up 5 different IR blockers and ran them through the two described processes to see what would happen.

 

CLEARLY......the first set shows that in the absence of color profiling for camera + lens + filter + light, some of the IR-blockers are better than others when only white balance is used in an attempt to restore the original Visible color of the converted camera. And remember that in the first set I used a Nikon converter to ensure that a Nikon D610 stock profile was applied. That did not seem to help fully restore some of the colors when using some of those filters.

 

I'm still looking at the second set. Custom color profiling goes a long, long way towards restoring original colors to the converted camera.

 

I think that what I'll do next is make a photo of some colorful subject (which is not a CC Passport) using the test filter set. Then I'll apply the Custom profiles. We need to see how this works for an "ordinary" photo. I'm not sure that just looking at the CC Passport is enough.

[Andrea B.]

If you make such color restoration experiments at home, then look at the TIFs using a wide color set like ProPhoto RGB. Jamming files into an sRGB/Jpg box can mess up some of the careful work I did for brightness and white balance. But what can ya do? Jpgs are the only way we can look at stuff here. And these photo are good enough to make the point.

[Andrea B.]

I think everyone can figure out which is the unfiltered photo in the first set. But wow look at its match in the 2nd set !!

[Andrea B.]

Andy, in answer to the 18% gray comment: I'll try that and see if I get anything much different.

[Andy Perrin]

Andy, guessing the filter might be fun. But I don't think you could do that in the 2nd set.

 

The point I've tried to make for at least the last 1000 years (snork, snork) is that white balance alone is not enough to restore good color to a Visible photo made with some kind of IR-blocker on the lens of a converted camera. So I rustled up 5 different IR blockers and ran them through the two described processes to see what would happen.

 

CLEARLY......the first set shows that in the absence of color profiling for camera + lens + filter + light, some of the IR-blockers are better than others when only white balance is used in an attempt to restore the original Visible color of the converted camera. And remember that in the first set I used a Nikon converter to ensure that a Nikon D610 stock profile was applied. That did not seem to help fully restore some of the colors when using some of those filters.

 

I'm still looking at the second set. Custom color profiling goes a long, long way towards restoring original colors to the converted camera.

 

I think that what I'll do next is make a photo of some colorful subject (which is not a CC Passport) using the test filter set. Then I'll apply the Custom profiles. We need to see how this works for an "ordinary" photo. I'm not sure that just looking at the CC Passport is enough.

 

Color profiling definitely helps a lot, but even the second set is not perfect. I maintain that we should choose filters using the first set, AND then apply the profiling. That way the profiler does the smallest alterations necessary. Forcing the profiler to do all the work is bound to cause glitches under extreme conditions (like that flower you showed awhile ago whose color you couldn't capture well).

 

Re the 18% gray:

I believe that it will matter only if any of those shots were close to being over-exposed. I think that's the reasoning behind the guideline. (The other reason is that all the automatic settings are designed to force the average image exposure to be 18% gray according to the light meter, but that's not relevant in manual mode.)

[Cadmium]

I looked up the camera type that worked best with S8612 as a visible filter: Fuji X-T1IR

I will will ask about posting the example pics.

[Andy Perrin]

I looked up the camera type that worked best with S8612 as a visible filter: Fuji X-T1IR

I will will ask about posting the example pics.

Wasn't that also the camera that was particularly IR-sensitive? If that's the case then it would make sense that S8612 is needed to tone down the IR more aggressively.

[photoni]

Andrea .... you had to try the new free Nikon software "NX Studio" ... it's not bad

I am now viewing the ColorCheckers with an old retina MacBook Pro from 2012.

This monitor sees sRGB fine but not AdobeRGB or DCI-P3 ...

.

of the two programs

# 2 looks good to me

# 44 is the best .

tonight I see with a calibrated Nec monitor ... now I'm going to take pictures in the hills :)))

.

P.S. Of all the programs I have (nikon adobe PhaseOne) to manage multispectral files the best is Capture One

[Fandyus]

Now stack them

[Andrea B.]

Toni, thx for the recommendation. I plan to put the new NX Studio on my new laptop.

 

Fandy, I fixed your double post. I'll probably skip stacking all that mess!

 

Dave, I can only make color profiles in Photo Ninja with the set of Xrite Color Checkers. There is a larger one available, but I do not have it.

 

Andy writes: I maintain that we should choose filters using the first set, AND then apply the profiling. That way the profiler does the smallest alterations necessary.

 

This is my thinking also. Twisting the colors (so to speak) too far could possibly induce artifacts. If nothing else the experiment supports exactly what you say by showing that some IR-blockers with WB-only are very close to the their color profiling results.

 

---

 

SET 1: UNFILTERED PHOTO

In the first set of photos it is obvious that #6 is an outlier. I hope everyone has figured out that #6 is the Unfiltered photo. What completely surprised me is that its companion in the second set, #66, looks spot on after profiling. But how would this play out for a real world, unfiltered photo? I will try to investigate this.

 

 

SET 1: RED, PURPLE, PINK PATCHES

These seem particularly prone to misbehaving under the various IR-blockers and not particularly repairable by white balance only. I'm thinking this should not be a surprise because most IR-blockers are are blue-green and red is cyan's complement. #2 and #3 have the best red, purple & pink patches of the first set. To generalize, any patch with a red component seems best in #2 and #3.

 

 

SET 1: BAADER UV/IR-CUT FILTER

It is obvious that #1 in the first set is "different" from #2, 3, 4, and 5. And also that #2 & #3 in the first set are very similar as are #4 & #5. So some might have already guessed that #1 is the Baader UV/IR-Cut result. (That is, you can make that guess if you know I use the Baader UV/IR-Cut.)

 

 

SET 1: BG IR-BLOCKERS

So, we are left with two sets of BG filters. The best set for WB-only is #2 & #3. The not-so-good set for WB-only is #4 & #5. At this point many of you with BG filter experience will be able to make an educated guess about what filters are in each set. But I would like to wait until I complete another experiment before revealing which filter is which. I don't want to bias your analyses.

 

---

 

SET 2

What is surprising is how good the Photo Ninja color profiling is for bringing the CC Passport colors to view even in #66 which is the profiled version of the unfiltered photo. Given that #2 & #3 in the first set are "best" after a white-balance-only, do we see that result in the second set? To my eye the differences between profiled photos in the second set are very subtle.

 

---

 

REMARK about COLOR PROFILING

Color profiling is done by professional photographers to provide standardized results for photos made with different gear and under different lights.

 

There are obviously photographic situations for which you do not want to alter the color cast or color renditions of a photo. Sunset photographs easily come to mind.

 

But when using converted cameras for visible light photos, as many of us do, I can't think of any other way but one^(*) to standardize the results other than using color profiling. You could always switch to a stock camera for which white balance only is sufficient to standardize an outcome. But I usually don't want to make that kind of switch and mess up the composition and perspective I have already recorded in UV or IR.

 

^(*)IF, and that is a big "if", you could know exactly what kind of internal filtration was removed from your camera and you could obtain that to use as an external filter, then the visible color would come out OK with WB-only.

 

Maybe someone who has done a teardown could ID the specific kind of filters which were removed? Internal filters might be BG-only, but they might also be coated. And the UV blocking component could be part of the IR-blocking glass or be separate. It is possible the UV-blocker also affects color.

 

---

 

Corrections and comments always welcomed.

I'm off now to photograph some kind of subject which is not a CC Passport using the 5 IR-blockers.

 

 

See Post #31 for the IDs of the filters.

CONCLUSION: Post #38

But wait....don't go there yet....there's more.

 

 

.

[Andrea B.]

Look at these labeled color patches while I'm gone.

They are crops from the second set.

 

Kindly note that while there are tiny differences in brightness and saturation (impossible to make that come out the same for various reasons), you can see that the color wheel values are remarkably close. Minor differences in color wheel values are also partly due to imprecise sampling techniques. With time and trouble, I could find an average value of each color patch, but not today.

 

The only anomalous value I see is in the first strip where the green from the profiled Baader UV/IR-Cut is "off". Of course, other anomalous values might occur on the patches I didn't measure. Note that I took color wheel values from the TIF versions.

 

Don't go to the bank with this data! These are only approximations, OK?

 

[Andrea B.]

Camera: Nikon D610-full spectrum conversion

Lens: UV-Nikkor 105mm f/4.5

Light: Ambient afternoon skylight

 

Set 1: Flower Mix with 5 IR-Blockers and an Unfiltered version.

White-balance and stock color profile from conversion in Capture NX2.

The filter order is the same as in all preceding examples.

I wish you could have the flowers in front of you in order to make some color assessments.

I'll do my best for you on that.

 

From left to right:

Senecio & Bud, Jupiter's Beard, Coreopsis, Prairie Zinna with scattered petals from Blue Flax.

 

 

 

---

 

Set 2: Flower Mix with 5 IR-Blockers and an Unfiltered version.

Custom color profiling in Photo Ninja using CC Passport.

Each custom profile is specific to camera + lens + filter + lighting used to make the photo.

The filter order is the same as in all preceding examples.

 

From left to right:

Senecio & Bud, Jupiter's Beard, Coreopsis, Prairie Zinna with scattered petals from Blue Flax.

 

 

---

 

DISCUSSION

 

The yellow or yellow-orange colors fare well in all the photos except, of course, for the two Unfiltered photos.

So I'm mostly going to omit discussion of the yellows.

 

Unfiltered, last photos, #6

• We can obviously discard the Unfiltered photos from consideration.
  

• Remember that in Set 2 above featuring the CC Passport, the color profiling did wonders for the color patches in the Unfiltered photo. But here, not so much. Why? I think that perhaps the flowers are much more IR-reflective than the color patches so they remain washed out looking even after color profiling has been applied. This should not be a surprise.
  

Baader UV/IR Cut, #1

• WB-only: the pink, lavender-blue and dark red are not quite right.
  

• Custom profile: the pink is better, the dark red is slightly better,
  but the lavender-blue is still not quite right.
  

• Color profiling is definitely needed for the Baader UV/IR Cut.
  But even then there might be red or purple/blue problems.
  

IR-Blocker #2

• WB-only: the pink, lavender blue and dark red are not quite right,
  but better than in the first Baader UV/IR-Cut photo.
  

• Custom profile: The posted colors are very close to the flower colors in front of me.
  I'm not 100% happy with that lavender blue, but it's not bad.
  (That seems to be one tough color to photograph accurately.)
  

• Color profiling improves results for IR-Blocker #2.
  

• Note here that custom profiled #2 looks almost the same as custom profiled #3. Interesting.
  

IR-Blocker #3

• WB-only: The pink goes slightly off again with this filter.
  The dark red is ever so slightly too dark.
  And someone might point out that maybe the yellow orange of the coreopsis is a bit dull.
  But this is the best lavender blue of the WB-only set.
  

• Custom profile: The posted colors are very close to the flower colors in front of me.
  The lavender blue doesn't please me 100% but it's not bad.
  

• Color profiling improves some results for IR-Blocker #3.
  

• Again, I really find it very difficult to tell the difference between color profiled #2 and color profiled #3 even though the WB-only versions have some small differences.
  

IR-Blockers #4 and #5

The same remarks work for both of these IR-Blockers.

• WB Only: Easily rejected for its misplaced pink and dark red.
  The cadmium orange around the dark red is also a bit too dull.
  The lavender blue is good though.
  In spite of what I said about the yellows above, they do appear to have picked up a little bit of green from in these photos.
  

• Custom profile: Quite a lot of improvement.
  Making a custom profile for IR-Blockers#4 and #5 would be mandatory to get the best from them in Visible light.
  

.

BEST of The WB-Only SET 1:

I think IR-Blockers #2 and #3 are best followed closely by #1, the Baader UV/IR-Cut.

 

BEST of The Custom Profiled SET 2:

There is a reasonably obvious division. The photos from Baader UV/IR-Cut, IR-Blocker #2 and IR-Blocker #3 are somewhat better than the photos from IR-Blocker #4 or IR-Blocker #5. How much better? Only further tests can tell. There hasn't been any renditions of green or blue yet, for example, except from color patches. It might be interesting to photograph a grass and sky scene through the 5 IR-Blockers.

 

 

See Post #31 for the IDs of the filters.

CONCLUSION: Post #38

But wait....don't go there yet....there's more.

 

 

 

.

[Andy Perrin]

Flowers are making me drift towards numbers 2 and 22. I do not have the actual flowers handy for comparison, but it would be interesting to hear what you (Andrea) think about the fidelity of the different cases to real life.

[Stefano]

If it can be helpful, I remember that when using my Chinese BG39 filter (alone) on my full-spectrum Canon EOS M camera, the reds were a bit darker than what I saw with my eyes. S8612 has the same curve on the red side as my filter (assuming my Chinese filter is close to actual Schott BG39), and as said before S8612 is not ideal for IR suppression in visible photos.

 

The best way to decide which filter is the best is either being there in person or comparing the images with a photo taken with a stock camera which reproduces colors accurately.

[Andrea B.]

As mentioned above, I'm here in person --

-- looking in real life at the actual flowers --

and writing up the Discussion remarks in Post #17. :bee:

 

Given Andy's earlier remark -- that we should use the filters with the best WB-Only results and then profile them -- I'm leaning towards IR-Blockers #2 and #3 as being potentially the ones to choose.

 

But I must say that the differences are subtle between all 5 IR-Blockers after color profiling has been applied. I did not know how this experiment would turn out because I've never compared all 5 of my IR-Blockers in an extensive test like this.^(**) The results are immensely interesting to me because I use my converted camera a lot for Visible photos. I'm hoping this long write-up might be useful to a few other members.

 

....subtle, so subtle, so very subtle....

...on little cat feet....

 

^(**)I should add that over the years I've been told and have known that BG-38 is considered the best way to go. So we shall see how that common wisdom holds up in these experiments.

 

---

If you can stand it, I have one more set to present.

 

---

[Andrea B.]

This a Black Hollyhock. The flower is not really black. Rather, it is a very dark maroon-red. The way sunlight hits it can cause it look black as it wanders around in the breezes.

 

I find this an extremely difficult flower to photograph correctly.

 

Let's only look at the color profiled results for the Black Hollyhock. Because I think everyone has gotten it by now that color profiling improves Visible results from converted cameras.

 

The order of the IR-Blockers used is the same as in preceding sets.

 

Baader UV/IR-Cut

 

 

IR-Blocker #2

 

 

IR-Blocker #3

 

 

IR-Blocker #4

Even though this is a Black Hollyhock, it does not look like this when right in front of my face!

 

 

IR-Blocker #5

Same remark.

 

 

DISCUSSION

No more discussion.....

 

Use either IR-Blocker #2 or #3 and give them color profiles for use with converted cameras in Visible light.

 

 

See Post #31 for the IDs of the filters.

CONCLUSION: Post #38

[Andrea B.]

P.S. I made sure all the little guys inside the flower went to live back outside in another Hollyhock flower.

[Andy Perrin]

I think I am still a number 22 person.

 

Andrea, I don’t know how you are viewing the results, but I found that the screen of my iPhone (same as your iPhone…) has the best color rendition of any of my devices and I verified this with a diffraction grating. The reds in particular are better than my IMac 27” (2020) and my old iPad.

 

Could you please look on your phone for me and verify that none of your conclusions change?

----

 

Update:

Here are the diffraction patterns. Don't read much into the exact colors shown because they were taken with two different cameras and lenses.

 

iMac 27" (2020) screen showing white background, seen through a diffraction grating, pic taken with the iPhone camera...

 

iPhone 12 Pro Max screen showing white background, seen through a diffraction grating, pic taken with Sony A7S with no filter (full spectrum) not well focussed:

 

A VERY crude test, but even here you can see how bumpy that red is in the iMac screen. Interestingly, my iPad screen (not shown) has the same pattern with the bumpy reds.

[Andrea B.]

Will do. I have to take a break first for dinner.

[dabateman]

Filter for number 3 is my favorite.

My guess based on that is the following:

#2 is BG38 2mm

#3 is BG40 2mm Similar to 1mm S8612

#4 is BG39 2mm or 1.75mm S8612

#5 is S8612 2mm

 

I added thickness as I don't know if you have all the different types.