Jump to content
UltravioletPhotography

Crazy White Balance Trick for Converted, Full-spectrum Nikons


Andrea B.

Recommended Posts

AUTHOR'S NOTE

I began a re-write of this topic on 29 August 2018.

Last Update [30 Aug 2018] Adding more info about UniWB.

 

WB = White Balance


 

UV White Balance in Nikon Converted Cameras Is Impossible

 

As we all know, converted Nikon DSLRs cannot measure an in-camera white balance through any UV-pass filter nor through most IR-pass filters. And, selecting a standard WB settings such as Auto or Daylight produces a raw file with screaming red channel overload. While shooting, all that red makes it very difficult to determine the optimal exposure for a reflected UV photograph because details are obscured and you cannot be sure how many stops to push beyond the already saturated right-hand wall in the red channel histogram. Red overload also makes Live View very difficult to use for focus. However, there are in-camera white balance presets we can make which reduce much of the red overload and make it easier to determine optimal UV exposure. One of those is the well-known Unitary White Balance (UniWB) and the other one I'm going to label Nikon Reduced-Red White Balance (RR-WB).

 

Review of Unitary White Balance (UniWB) for Optimal Exposure

 

Some years ago Iliah Borg of Raw Digger fame demonstrated the benefits of using unitary white balance (UniWB)(1) in visible photography. An in-camera UniWB preset is made using a certain algorithm to force R, G, G and B white balance multipliers as close to 1.000 as possible. This results in an in-camera JPG histogram that is much more accurate for determing optimal exposure and much more useful for enabling the technique of exposure to the right(2). UniWB also produces a very green raw visible file, but that is fixed by re-doing the white balance of the raw file during conversion.

 

Using UniWB for UV photography under a UV-pass filter like the BaaderU results in a dark magenta photo, but does provide the more useful exposure histogram. How well this dark magenta works for focusing through a UV-pass filter using a UV-LED torch or strong sunlight may vary by camera. Setting the contrast and saturation to neutral values will help, so be sure to do that in your camera's JPG settings. Contrast and saturation can always be restored during conversion.

 

The details about how to set UniWB are provided in reference (1), so I will not re-write those in this topic. Here is BaaderU UV photograph in UniWB.

baadUniWb.jpg

 

 

.

Nikon Reduced-Red White Balance (RR-WB), not quite UniWB but useful.

 

There is another way to change the R, G and B white balance multipliers and pull the overall colour cast back from the red/magenta bias. For lack of a better name, I'm just calling it Nikon Reduced-Red White Balance. [Author's Note: In a former version of this topic I labeled this Ner-UniWB, but that was not a good choice because the R and B multipliers were not close enough to 1.000.]

 

The benefits of RR-WB:

  • RR-WB is very easy to set.
  • It provides an improved Live View.
  • Used with a broadband filter like the BaaderU (peak 350nm), RR-WB gives a preview of some of the false-yellow and false-blue seen in white-balanced UV photo conversions. Other filters will vary on this point.

The drawback of RR-WB is that its histogram is not as accurate as a UniWB histogram would be although I have been able to use it successfully to obtain better exposed UV photographs. After working with it for a while, I've gotten a feel for how the RR-WB histogram behaves relative to pushing D610 UV exposures for less noise.

 

BaaderU UV photograph in Nikon RR-WB.

baadB6.jpg

 

 

.

Quick Method for Obtaining Nikon RR-WB

 

Briefly, make an overexposed photo, load it into a WB preset slot and adjust the WB color chart.

  • Make a Very Overexposed Photograph
    Open the UNFILTERED lens up to maximum width aperture. Set exposure time to 30 seconds. Aim at a very bright, white light source, at a bright sunny sky or at very bright white clouds. Don't aim at the sun, please. Then make the 30" exposure. You should get a (raw) white, very blown-out photograph with a histogram showing only R, G and B spikes. The spikes should either hit the right-hand histogram wall or be very close to it. If a 30 second exposure does not produce this set of right-most RGB spikes, then raise your ISO. In my converted D610, it isn't possible to move the spikes completely to the right.

  • Load the Overexposed Photo into a WB Preset Slot
    Here I'm arbitrarily using WB preset slot d-4. I usually leave slot d-1 open for any on-the-fly WB presets and make use of the other three slots for permanent WB presets
    • Shooting Menu > White balance > PRE Preset manual > {d-4} > Select > Select image > {select overeexposed white photo} > OK

  • Fine-tune the WB Preset Color Chart
    You will have to experiment with this a bit to decide in which direction to set the color bias because it will depend on the UV-pass filter you are using. I've used both {B6} and {G6} for the BaaderU. It is possible that you will want to leave the color chart cursor in its central, neutral position.
    The chart cursor is moved with the control pad on the back of the camera.
    • Shooting Menu > White balance > PRE Preset manual > {d-4} > Select > Fine-tune > {B6} > OK

  • Protect the WB Preset Slot
    It's a good idea to prevent accidental overwrites.
    • Shooting Menu > White balance > PRE Preset manual > {d-4} > Select > Protect > ON > OK

.

Some Actual White Balance Multipliers

 

I wanted to list some actual WB multipliers so you could get a better feel for this. You can check your own files using any EXIF reader or using a converter app like Dark Table.

 

Remember, white balance is set relative to the green channel being held at 1.000.

 

Nikon D610-full spectrum with camera's Sunlight WB setting.

  • R = 1.906
  • G = 1.000
  • B = 1.391

Nikon D610-full spectrum with camera's Incandescent WB setting.

  • R = 1.227
  • G = 1.000
  • B = 2.215

Nikon D610 BaaderU photograph white balanced in Dark Table.

In an ideal world, we would be able to set these multipliers in-camera and have a much easier time of it when making UV photos with the BaaderU. :D

  • R = 0.316
  • G = 1.000
  • B = 1.344

Nikon D610 Overexposed Photo with no WB fine-tuning. (As described above.)

  • R = 0.602
  • G = 1.000
  • B = 0.813

Nikon D610 Overexposed Photo with {B6} green adjustment. (Shown above.)

  • R = 0.547
  • G = 1.000
  • B = 0.648


Suggestions for Further Experiments

  • Instead of leaving your lens filterless, create your overexposed photo using the BaaderU -- or any other filter -- to see if you get an improvement in the outcome.

  • Write a program to re-set the R, G and B multipliers in a raw NEF to in such a way that the NEF may be reloaded into the camera and used as a WB preset photo. (One thousand years ago this was possible with the ancient Capture software which was written on papyrus in Fortran.)

  • Experiment with altering RGB multipliers by attempting to measure WB against a monitor filled with a somewhat desaturated colour like (255, 190, 60). That gave me these multipliers: 0.711, 1.000, 2.820. While not useful for UV, it was rather interesting for unfiltered photos. [Note to self: re-check these numbers

  • Look at the techniques in this linked topic to find other multiplier trickery. Given that UV is recorded mostly in the Red channel in Nikons under a broadband UV-pass filter, why not just shoot in the Red channel only? Shoot in Nikon Red using 7.996, 1.000, 0.648.

.

Examples Using StraightEdgeU-Gen2

This new UV-pass filter has a very blue bias when using converted Bayer-filtered cameras. I thought it would be interesting to see what a photo would look like in UniWB and in Nikon RR-WB.

 

SEU-Gen2 UV photograph in UniWB.

seug2Uniwb.jpg

 

SEU-Gen2 UV photograph in RR-WB with {G3} adjustment.

Some fine-tuning may be possible to further reduce the purple while shooting.

Or use a Monochrome setting while shooting if color obscures fine details.

seug2RRWB.jpg

 

SEU-Gen2 after WB in converter.

No other edits.

seug2final.jpg

 

 


Footnotes

(1) http://www.guillermo...wb/index_en.htm

In 2008 Guillermo Luijk wrote the goto page for UniWB. If you scroll down to the last section, you'll see the quick white balance method I just described. Luijk observes that this method does not work perfectly for all cameras -- Nikons, for example. But it works well enough for our reflected UV needs. :D

 

(2) Nikon DSLRs are known for having between 1-3 stops of highlight 'headroom'. This means that if you push the JPG histogram as far to the right as possible while shooting, then you can gather more light and make a raw file having much less noise. During conversion you pull the exposure back to normal. There is some trial and error involved in figuring out just how far you can expose to the right with a given camera and a given scene.

 


Illustrations

In the next three posts I've attempted to show how using a Nikon RR-WB can improve exposures in both Visible and UV photographs. And I have an experiment showing bad results when using a standard white balance setting for UV photography.


As always, comments, corrections and links are welcomed. Let us know of your experiences using a Nikon RR-WB setting for UV photography. B)

Link to comment

Standard White Balance Settings Are Not Useful for UV Photographs

 

Equipment: Nikon D610-broadband + UV-Nikkor 105/4.5 + Baader UV-Pass Filter + Sunlight

Exposure: f/11 @ ISO-200 with various speeds

 

I made two UV photos using the Incandescent white balance setting, one at the metered exposure and one for which I tried to expose to the right. Back in the day, Bjørn and I used to shoot Incandescent with our old D200 conversions because it reduced the red overload by a little bit. He soon wised up and began to experiment with custom white balance settings. It took me a while longer but I got there eventually. I still have folders full of horrible unprocessed magenta-red D200 UV photos. Anyway, the Incandescent WB setting got in the way of my attempt to ETTR.

 

 

Unedited photo made at metered exposure using Incandescent white balance.

In the wrong white balance, it is difficult to determine how much underexposure the metering gives you.

Here the red channel has already blown. But the photo is basically underexposed.

incandes_2sec_orig01.jpg

 

Unedited metered photo -- brightness only.

Look how far to the left that histogram sits.

incandes_2sec_brightness01.jpg

 

Metered photo, click-white balance applied, no other edits.

When the exposure is brought up by a stop or so, a lot of noise will be revealed. Not good.

incandes_2sec_wb01.jpg

 

 

Unedited ETTR photo made at X stops over metered exposure using Incandescent white balance.

In the wrong white balance, it is difficult to determine whether you have pushed the exposure far enough to the right to prevent noise.

The red channel blows out and detail is obscured.

incandes_5sec_orig01.jpg

 

Unedited ETTR photo -- brightness only.

I did manage to push the brightness histogram towards the middle. But it could have been pushed further.

incandes_5sec_brightness01.jpg

 

ETTR photo, click-white balance applied, no other edits.

Better than the metered version, but there will still be noise produced when this exposure is brought up.

incandes_5sec_wb01.jpg

Link to comment

UniWB or RRWB Helps Make Better UV Exposures in Nikon Cameras

 

Equipment: Nikon D610-broadband + UV-Nikkor 105/4.5 + Baader UV-Pass Filter + Sunlight

Exposure: f/11 @ ISO-200 with various speeds

 

Surprisingly the converted Nikon camera can sometimes still meter through the dark UV-pass filter. And unsurprisingly that metering will not be accurate. But it does give you a good starting place for obtaining a proper, light-filled UV exposure when using either UniWB or the Nikon Reduced-Red White Balance (described in the first post) together with an exposure to the right.

 

The sunlight I had today measured about 2.5 mW/cm2 using my Solartech Solarmeter. That's a much better reading than we would get mid-winter, but it was not quite up to the 4.0 - 5.0 mW/cm2 that I get during a midsummer's mid-day. So I expected exposures to be long at f/11 using the rather low ISO-200 setting. And they were.

 

For these photos, I made use of the Nikon Reduced-Red WB, described in the first post above, to make an in-camera WB preset for UV in my converted Nikon D610. When reviewing the photo histograms made under this WB, I could see more details on the LCD and also quickly see whether the position of the brightness histogram was far enough to the right. [Note: The RR-WB histogram is not as accurate as the UniWB, but it is still useful with experience.] Using RR-WB made Live View brighter at a wide-open aperture without additional UV illumination from my small UV-LED torch. So I could focus wide open in UV and then stop down.

 

 

Unedited photo made at metered exposure using a Nikon RR-WB preset.

Start at the metered exposure, then add between 1-2 stops more for a good UV exposure.

THE WB LABEL IS NOT ACCURATE. My apologies for that. I wasn't paying attention! "-)

helianthus_uvBaadSun_uniWbD3_20170408wf_797.jpg

 

 

Unedited photo made at 1.66 stops over metered exposure using a Nikon RR-WB preset.

I could have gone a full 2 stops over. I shouldn't have been so timid. "-)

THE WB LABEL IS NOT ACCURATE. RR-WB was used.

helianthus_uvBaadSun_uniWbD3_20170408wf_803.jpg

 

 

Finished metered photo using a Nikon RR-WB preset.

Not bad, but it has luminance noise if you look closer.

[Note: Down-sizing a photo often helps to mitigate noise.]

THE WB LABEL IS NOT ACCURATE. RR-WB was used.

helianthus_uvBaadSun_uniWbD3_20170408wf_797pn01.jpg

 

 

Finished ETTR photo using a Nikon RR-WB preset.

THE WB LABEL IS NOT ACCURATE. RR-WB was used.

helianthus_uvBaadSun_uniWbD3_20170408wf_803pn02.jpg

 

 

Crop from the finished metered photo showing luminance noise.

helianthus_uvBaadSun_uniWbD3_20170408wf_797pn03.jpg

 

Same crop from the finished ETTR photo showing less luminance noise.

helianthus_uvBaadSun_uniWbD3_20170408wf_803pn01.jpg

 

 

Detail from the finished ETTR photo. Very low luminance noise.

This crop has not been resized. Might be some motion blur? It was so windy!

An observation: Judging chrominance noise in a UV floral photo can be difficult because there can be so much iridescence at play from surface conical cells. However, the luminance noise is low in the dark areas.

helianthus_uvBaadSun_uniWbD3_20170408wf_803pn03.jpg

Link to comment

Yes. You can use ETTR with any digital camera. Some may have more headroom than others. So experimentation will help you determine the amount you can push the brightness histogram to the right and still recover back to a proper exposure (with less noise). It is a useful technique for reflected UV photography which tends to get noisy when underexposed.

 

I hope I've gotten all the ETTR terminology correctly written. There is a lot about ETTR across the various photography websites and blogs, so you can read more about it. Raw Digger even has a tutorial for using that app to determine the precise amount of headroom for a specific camera. (Although in practice, that ideal may not be reached because I think there is some dependence on the particular subject being photographed and its luminosity.)

 

Anyway, give ETTR a try and see what you think.

Link to comment
  • 1 year later...

Note: 29 August 2018

I've reworked this topic to improve the terminology and to clarify (I hope!) the technique.

 

Please let me know of any typos or errors. I'm not my own best editor! ;) :rolleyes:

Link to comment
UniWB: Not the best solution for reflected UV photography with a converted Nikon

 

Using UniWB for UV photography under a UV-pass filter like the BaaderU results in a dark magenta photo. This is somewhat better than the overload of red obtained with Daylight/Auto WB, but not by much. There is a better way to change the R, G and B white balance multipliers and pull the overall colour cast back from the red/magenta bias.

 

Common WB usage is about producing a colour balanced image. For UV and IR, UniWB is not about making the image look good, it is about making the histogram more accurately reflect the saturation status of the RAW data.

Link to comment

Yes. Agreed.

And I am hoping that I had made the point up there that I was using this red-reduced white balance to improve Live View and to improve exposures.

To that end, the RRWB works better for me than UniWB for UV. The UniWB is too magenta in UV.

 

Here is one UV UniWB example. I've exposed it well, so it isn't too bad. But working in UniWB for a broadband UV-pass filter is still a little difficult due to all the dark red/magenta - especially in a landscape where you have no hope of seeing anything near white like this Spectralon.

 

D610 and BaaderU

baadUniWb.jpg

 


 

Shane, another trick I've used is to set the Nikon to Monochrome. And if you know your camera well (that is, how much ETTR room do you have?), then using the mono Brightness histogram to shoot broadband UV often works as well as anything*. But I have been enjoying the false yellow/blue preview I get with this other RRWB setting.

 

BTW, I think I am unable to push the Nikon R and B multipliers lower than the .547 and .648 shown above. I've tried!

 

*Push the mono brightness histo to the center in a scene with both UV-lights and UV-darks and get a better exposure. Typically the red channel blows, but I know with the Nikon D610 that a centered brightness histo pulls back well enough in the Red channel with a good converter.. Obviously if the scene has no brights, you do need to be careful using the mono hist.

And yes you can see the brightness histo when using color, but it is all easier if switched to Monochrome. You don't get distracted by the Red blowout.

Link to comment
And I am hoping that I had made the point up there that I was using this red-reduced white balance to improve Live View and to improve exposures.

To that end, the RRWB works better for me than UniWB for UV. The UniWB is too magenta in UV.

 

Kind of but it might be better to state that you get what you get with UniWB, if you don't like how the image appears, then produce a custom WB and give up the benefits of UWB with its more accurate histograms.

 

Your RRWB doesn't improve exposures, it improves appearance.

 

Formerly I called it "Near-UniWB"

That term was coined by Iliah years ago and differs from what you refer to. It was for a UniWB that had multipliers fractionally different from 1 and it was used for early versions of Photoshop that couldn't handle values of 1:1:1:1 (which might still be the case).

Link to comment

Thanks, Shane! I really appreciate your help with this. :)

 

I will go back through this and make sure my terminology is correct. I did try to remove all references to Near UniWB because I no longer was obtaining values near 1.000.

 

I will also go back and work in more information about UniWB, as suggested. That will be useful.

 

I have put an IN PROGRESS note in the title so that readers will understand that further changes can occur. Also put a Last Update tag at the beginning of the topic.

 

A snag to any UniWB effort is that I cannot show any really good examples of UniWB under UV-pass filter because I do not have a UniWB photo which I can load into the D610. The example I showed above was a screen shot from Dark Table which permits direct alteration of the WB multipliers. But Dark Table cannot save them in a NEF.

 

And I seem to have failed to obtain even a Near UniWB in the converted D610 with the usual efforts. As mentioned above, I can't seem to get the RB multipliers below .547 and .648 in that camera. One of my suggestions above was for someone to try to write a program which would alter the multipliers in a raw file in such a manner that the camera would still accept the photo to use in a WB slot. If I'm remembering correctly, I tried to change multipliers with one of the Exif tool programs but it didn't work out.

 

[Off topic: Shane, what are you photographing these days? Nice to see you here..... B) ]

Link to comment

A UniWB reference must be obtained before converting the camera to UV, IR or UVIR but have long forgotten the reason why. I made this mistake with my D800IR830. It was easy to remedy, I just borrowed an unconverted D800 to build my UWB then transferred the reference image to the converted camera. You might be able to talk someone else into building one for you.

 

While it is possible to upload a Nikon DSLR jpg file back to the camera unfortunately, unlike Canon, Nikon cannot upload a synthetic Photoshop built jpg back to the DSLR without extreme difficulty beyond my capability. However, several years ago I did manage to upload a synthetic jpg (originally built for a Canon)

http://www.brisk.org...og/hiblow1.html

to my Nikon with the help of a jpg expert.

It was a custom built grid of R, G and B squares varying in values from around 230 - 255 and indicated exactly the point the highlight warning blinks turned on. Unfortunately, the jpg expert is otherwise involved in more important projects now. The synthetic jpg must be formatted to Nikon DSLR jpg specifications and fit into the appropriate file location of an existing Nikon DSLR jpg file. The synthetic image can then be uploaded to the DSLR.

 

Perhaps someone here has the skills to do this, it could have some interesting uses.

 

(Off topic.....answered in email)

Link to comment

The RR WB balance technique described above didn't work for me at all using converted Nikon D3200 camera. The preset WB live view looked pretty much the same as when using auto WB. Maybe it just didn't register the image I chose, but it didnt give any error message. Maybe only works for the more expensive full frame models, rather than the DX models?

 

PS - I tried with nikon E 50 mm + UG1 2mm + S6182 2mm outside in sunlight, ISO 800, f/1.8, various different flowers.

Link to comment

Jim, you are supposed to do this without any filter on the lens.

 

FX/DX has nothing to do with it.

 

In the Suggestions for Further Experiments, I did suggest attempting the method with a filter on the lens. But that was just supposed to be a "further experiment" -- which may or may not work.

Link to comment

I was not sure from your post.

 

Let me try a photo with the combination UG1(2.00mm) + S8612(2.0mm) you mentioned and see what I get.

I don't have that 50E but I'll try with some non-dedicated lens.

Link to comment

Both these were made using a Neutral Picture Control [4,0,0,0,0].

The files were converted in Capture to preserve the settings.

 

With this filter combination, UG + S8612, we also see the red overload is reduced and the false-yellow and false-blue begin to appear.

 

610_6511.jpg

 

610_6501.jpg

 

 

 

 

It was so hot & humid out there that the OK button stopped working on the D610 ! The humidity had to be wiped off the lens with a microfiber cloth about every 2 minutes. No more experiments today.

 

Here is an unfiltered photo. Lots of IR in this one. Looks very tropical with the lime and aqua colours. Feels very tropical with 92°F (33°C) and a 79% humidity level.

610_6505.jpg

Link to comment
  • 2 weeks later...

Unfortunately, the D3200 does not have CWB "slots".

 

You can set one CWB by taking a sample shot, or you can use an image on the card. If the D3200 doesn't like the sample shot, it will tell you it's unable to set the white balance; if you select an image, it sometimes works, but you don't get any error message if it doesn't like the image: it just doesn't work.

 

As a result, I usually set my CWB against grey concrete with the filter off. Works great for full spectrum shots and 720nm, but it's useless for UV.

 

I have just 2 filters that I can set the CWB with on the lens: the NKIR B05 and NKIR Y03 - these are blue and yellow in colour, transparent. I would guess the cut-offs are around the 490nm and 560nm mark). Can't do it with my Kolari Vision K550 orange flter, or anything higher than that.

Link to comment

Andy, You might try the method that Bob and Shane call "scene referred white balance", which I tried to outline in the topic below.

It probably will not work for UV, but it might work for 720nm. It will probably work for more filters than not.

So if you have not tried this, you could test it out.

http://www.ultravioletphotography.com/content/index.php/topic/2360-in-camera-white-balance-with-newer-nikons/

Link to comment

Andy, You might try the method that Bob and Shane call "scene referred white balance", which I tried to outline in the topic below.

It probably will not work for UV, but it might work for 720nm. It will probably work for more filters than not.

So if you have not tried this, you could test it out.

http://www.ultraviol...h-newer-nikons/

Bookmarked the link and will try it tomorrow, sunny weather permitting. Will let you know how I get on.

Link to comment

I don't use it, but I tried it, and got more filters to actually white balance in camera with the Nikon cameras I have than ever before,

but since I can't use it for all the filters, I don't use it. Bob says it use to work for everything back with the D200's and such.

The main reason I would like to white balance in the camera is so live view would look closer to what I will get when I optimally white balance out of camera.

I usually hunt around for my preferred white balance anyway.

Now if somehow we could edit the NEF file, and not have the camera reject an edited NEF... then that would be cool, but I haven't found any way to do that.

It is probably possible, if you put Andy P. in a think tank with his MatLab... ;) then who knows.

Link to comment
Now if somehow we could edit the NEF file, and not have the camera reject an edited NEF... then that would be cool, but I haven't found any way to do that.

It was possible to do it for a jpg, it may be possible for a NEF but beyond my available skills and time.

http://www.ultravioletphotography.com/content/index.php/topic/2209-crazy-white-balance-trick-for-converted-full-spectrum-nikons/page__view__findpost__p__23431

Link to comment

This gives me an idea for a Nikon experiment. I'll run through the IR-pass filter stack and determine which ones permit "scene referred white balance" on D610 using a non-dedicated UV lens. Maybe that would be useful to folks?

 

Folks, remember this is basically a Nikon shortcoming. Other cameras can more easily measure white balance through dark UV or IR filters. Although note that these in-camera WB measurements may not always be accurate compared to the results obtained from using a photo of a UV/IR stable white standard and setting WB in the converter. That is always the Gold Standard method.

Link to comment

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
×
×
  • Create New...