[cdd29]

Recently had a K-01 converted to full spectrum. I set the white balance by pointing at grass in full sun. Used this method for years to set white balance. However when I open the RAWs in Photoshop and channel swap, there is a cyan cast on foliage (slight gray tint as b&w). Even viewing in raw editor or in camera foliage has a tint to it. I've played with the white balance manually in PS but can't quite get it set right. Tried with both 590 and 720 filters. Any suggestions?

---

Last edited by cdd29; 07-19-2021 at 07:22 AM.

[stevebrot]

I don't work with a converted camera, but here are a few questions anyway:

• What camera?
• When you did your manual WB, was the result uniformly gray or were there splashes of color here and there?
• What is ACR using for a profile for its RAW conversion? Did it use your WB or compute its own?
• What brand and model filters are you using?

[Ishpuini]

Setting white balance in camera a great way to get an idea of what the end result of an IR image will be (I use an ExpoDisc for this), but you really need to fine tune it on your computer in post processing for best results. If you're shooting RAW, that's obviously best, but still your software needs to be able to set the white balance sufficiently far from normal to get anything right for infrared. For Lightroom and Photoshop this requires a custom profile for your specific camera to be made with Adobe's DNG profile editor.

Great resource for this are the videos on YouTube by Rob Shea.

Note that Adobe is ending its support for the DNG Profile Editor, so better not delay if you still need to download it.

[ProfessorBuzz]

Have you tried a 720 filter and a good old 18% grey card?

[cdd29]

I don't work with a converted camera, but here are a few questions anyway:

• What camera?
• When you did your manual WB, was the result uniformly gray or were there splashes of color here and there?
• What is ACR using for a profile for its RAW conversion? Did it use your WB or compute its own?
• What brand and model filters are you using?

1. Pentax K-01, 18-135mm & Sigma 35mm art
2. There's a little white but grayish across the board. Both would be in the same tress, so I don't think it's a case of some vegetation reacting differently than others.
3. ACR is using the default color profile (would have to check tonight to be exact). It's using the 'as shot' values from the camera. I've never created or used a special profile in the past with IR work as just setting it in camera has got it close enough to work fine in Photoshop.
4. Filters are just some cheap $20 filters off of Amazon. I've used three different filters (the two 720's are probably made by the same outfit but branded different). Here's the links to the 720's. I've had the 590 too long to find any info:
amazon.com : GREEN.L Infrared Filter 62mm IR 720nm X-Ray Filter Optical Glass Filter : Camera & Photo?tag=pentaxforums-20&
amazon.com : Opteka Infrared Filter ?tag=pentaxforums-20&[X-Ray Effect] ? IR720 Removes Most Visible Light Below & Above 720nm Wavelength for Compatible with All Popular 67mm Camera Lens Models : Camera & Photo

---------- Post added 07-19-21 at 06:15 AM ----------

This is a screen grab of one of the better shots I've got from it. Most of the leaves look correct. Some in the lower right & middle left have a slight cyan cast to it.
1. White balance wasn't tweaked in ACR
2. Standard blue/red channel swap
3. slight color cast removed by tweaking color balance

[cdd29]

This is a screen grab of one of the better shots I've got from it. Most of the leaves look correct. Some in the lower right & middle left have a slight cyan cast to it. Alot of stuff I shot over the weekend are much worse and are cyan throughout (slight gray when viewing with no editing). this was taken last week.
1. White balance was tweaked in ACR but pretty much left as the camera set it.
2. Standard blue/red channel swap
3. slight color cast removed by tweaking color balance

[cdd29]

I may have answered my own question (this is what happens when I post at 11:30 and night and not fully research the problem). Usually don't have to mess with hue & saturation, at least I haven't in the past.

"On occasion you may run into an image that has strange looking cyan highlights, like the bellow example:

You can correct this in Photoshop by going to Image – Adjustments – Hue/Saturation. In the Edit menu select Cyan as the color to work on. Your cursor will change to the color picker tool, find a patch of this cyan highlight in your image and click on it with your picker tool. Now drag the Saturation slider all the way to the left, making the cyan cast gray. The cast is gone but our highlights are now a muddy gray, to make the highlights normal highlights just drag the Lightness slider all the way to the right."

[Ishpuini]

I never managed to get this result through Lightroom without custom profiles. But you are working through ACR straight into Photoshop? Perhaps the K-01 profile is wider than the standard profiles for the DSLRs that I used?


It's hard to tell, but you might have some infrared hotspot going on as well. Modern coating is optimized for visible light but often let's infrared light bounce freely around in a lens, resulting in a brighter and slightly differently coloured centre area of a picture. This can be brought out by sliding some of the dehaze and contrast sliders to the extreme. I had this centralized cyan cast in part of my images as well and found out in my case that it was these hotspots. I'm currently using an old M28/2.8 and the FA43/1.9 limited on my full spectrum converted K-3II and these seem not to show hotspots at all, contrary to most DA lenses. I'm getting a much more uniform result with these lenses.


BTW, if you don't like the cyan cast and would prefer more of an ochre cast with slightly different blue for the sky, try inverting colours (but keeping luminance) instead of channel swapping, like so:

You Tube

You Tube

Depending on the image I will sometime swap channels or use inverted colour.

[cdd29]

It is mildly hot spotting. The 18-135 is listed on the Kolari website as not a great IR lens, but that's what I have for the time being. I get the same cyan cast with a Sigma Art 35mm. I'll have to look into inverted colors.

[stevebrot]

4. Filters are just some cheap $20 filters off of Amazon.

I would place the blame there. I have both a Hoya R72 and a (cheap) Zomei 720 and the Zomei is, to put it kindly, sort of "leaky" while the Hoya lets little, if any visible spectrum through. It is my understanding that cheapo IR filters generally do that.


Steve

[Ishpuini]

Fwiw I got the same discoloured center with Hoya R72 filters. I'm putting it down to a hotspot prone lens.



---------- Post added 19-07-21 at 21:26 ----------

It is mildly hot spotting. The 18-135 is listed on the Kolari website as not a great IR lens, but that's what I have for the time being. I get the same cyan cast with a Sigma Art 35mm. I'll have to look into inverted colors.

I would imagine the sigma art to have a similar modern coating. IMHO it's really worthwhile getting some old glass for infrared photography. It's cheap as well. Anyway that's my current position with the lenses I own personally.

[AstroDave]

a good old 18% grey card?

Not sure how "grey"! Read on:

Perhaps this begs the question: What, really, is “white” balance for an IR-converted camera?

As we all (should) know, our standard color digital cameras use a sensor chip (CCD or CMOS) with an array of R, G, and B filters deposited just on top of the actual sensor pixels.

In order to get a proper-looking image, the camera must adjust the relative proportions of the signals from the various filtered sensors. Selecting a WB setting in the camera sets up the expected ratio, depending on the spectrum of the light source (how much R, G, or B light there is coming from, e.g., the Sun). The actual application of the relative factors can be done internally by the camera for the creation of jpeg files, or be left to PP software in the case of RAW files.

One can do a “custom white balance” by taking a shot of a “gray scale” card and then having the camera adjust the amount of R,G,B from the sensor so that the card looks “gray” (or white - a sheet of white paper works just about as well). When we do this we are relying on the maker of the gray card to guarantee that over some range of light wavelengths, the card is truly gray: it reflects the same fractional amount of R, G, and B at all wavelengths. For normal color photography, the wavelength range will be from around 400 nm (very deep blue/violet) to somewhere between 650 and 700 nm (very red). And, I suspect, the average gray scale maker does not worry about the reflectivity of their product outside this range of wavelengths.

You can readily find the color response of cameras on the web. Look here:

RGB: All About Digital Camera Color

Bin space: Spectral response of Nikon DSLRs (D90 and D300s)

Nikon CFA Spectral Power Distribution | Strolls with my Dog

(Sorry about all the Nikons!)

Other cameras have quite similar responses, including our Pentaxes. Notice that sharp falloff at the red end of the spectrum - this is due to that IR-Cut filter.

What happens when that IR filter is removed? Colors from any standard processing program may well look bonkers!

You can find the intrinsic sensitivity of many camera sensors with a bit of web searching (i.e “Sony CCD sensors spectral response”). Here’s something you might find: This plot shows the non-IR-filtered behavior of a Sony IMX250 chip - note that the Blue and Green filters on the sensor allow a lot of leakage out in the IR and the overall sensitivity is quite high out to almost 1000 nm! The concept of blue or green or red color response is not so clear cut when that IR filter is removed.



Of course, the response of your modified camera will depend on the external filter you use, and the transmissivity of your lenses, along with the idiosyncracies of your sensor. Processing with RAW and choosing the behavior of various color weighting / temperature etc. may well give you some very strange looking “white”balance effects.

I think you will just have to play around to find what you like, and might even consider software that will allow you to pull out, say, just the Red part of the RAW data. This will correspond most closely to what you would image if your were using a true IR camera - i.e. something that responded only to wavelengths longer than ~700 nm.

[Ishpuini]

Thanks for posting this! It was only when I saw a similar graph that I finally understood how infrared light translates to RGB colour in a modded digital camera. It's a revealing curve, that shows why infrared images shot with a 720nm filter look a bit orange (red + a bit of green), and why you would need a 850nm filter to get pure B&W (R=G=B). All straight out-of-camera without WB adjustments that is.


As for in-camera white balance, as I wrote earlier I use an ExpoDisc in front of my lens for this. This is made out of an array of tiny fisheye lenses in front of a neutral white diffuser, so you just need to hold it in front of your lens+filter and point it at the scene to get a homogenous neutral colour for your specific combo. I found this perfect for setting custom white balance in my old *istDS which struggled with AWB. More recently, on my full spectrum modded K-3 II, I found this perfect for setting the white balance with maximum colour separation. It seems to provide neutral colour in the whole spectrum, including the infrared range.

IMHO, maximum colour separation is what you need to estimate the processing potential of an infrared image in-camera, regardless of the WB and other trickery you might apply in post.

[ProfessorBuzz]

Refer to te Kodak R27 Gray Card data - it's good from 400nm to 700nm.

I did find a copy of the PDF here, not on the official site:

https://www.zonephoto.it/images/pdf/kodak-grey-card1903061.pdf

[Ishpuini]

Refer to te Kodak R27 Gray Card data - it's good from 400nm to 700nm.

I did find a copy of the PDF here, not on the official site:

https://www.zonephoto.it/images/pdf/kodak-grey-card1903061.pdf

That wouldn't be helpful for setting white balance on an R72 filter equipped infrared camera that registers light only starting from 720nm upwards. However, I see it's just the graph's X-axis going up to 700nm, so it might be good above as well...