Do I insert the filter inside the camera?

[danni...@gmail.com]

Hi there

I´m modifying a Canon A 2200 Powershot. The NIR-filter is removed and I just received the red and blue filters. Do you put the red filter inside (replace the NIR-filter) or place/glue it outside on the camera?

- Thanks

[Chris Fastie]

The filter can go either place. If it's in front of the lens, it's easy to switch it for other filters, including an IR block filter so you can get normal photos again. If the filter is inside, it is protected and might degrade the image quality less than a plastic filter in front of the lens (especially if the exposed filter gets scratched). 

Chris

[shi...@rogers.com]

On Sunday, December 13, 2015 at 9:32:17 AM UTC-5, Danny Morck Jensen wrote:

Hi,
         I have removed the IR block filter from two cameras (an EasyShare M380 and an EasyShare Z915), and they both show pink casts and colours. Will this blue/red filter combo remove the colours and the colour cast?
Thanks,
...... john

[Chris Fastie]

John,

Congratulations for successfully converting those cameras. Adding a red (e.g., Rosco #19) or blue (e.g., Rosco #2007) filter will change the way the photos look. The color balance of the photos will also be determined by the white balance setting of the camera. With the new filter added, to take a photo which can be converted directly to resemble NDVI the channel used for NIR must be several times brighter than the channel used for visible light. This artificial color balance can be achieved with an extreme custom white balance setting. Unfortunately, those Kodak cameras do not allow a custom white balance to be set. The white balance presets you can select will probably not produce photos in which plant foliage has the proper color balance. 

With the added filter, one or more of the color channels will be dominated by near infrared light (e.g., with a red filter, the blue channel is mostly NIR). This channel alone will have good information about plants (healthy plant foliage will be very bright). But computing a plant health index like NDVI might be tricky. If you include targets of known spectral quality in your photos, those can be used to calibrate the process (the channel values for every pixel will be adjusted after the photos are taken to allow the computation of good NDVI values). This process is described here: https://publiclab.org/notes/nedhorning/07-22-2015/introducing-the-calibration-plugin-for-imagej-fiji

If you post some test photos we might be able to confirm some of this.

Chris

The KAPtery

[shi...@rogers.com]

Hi Chris,
            Yes, unfortunately, these cameras only have a few presets for white balance ..... auto, open shade, fluorescent, daylight, and tungsten ..... that's pretty well it. I read somewhere that unexposed and developed E6 (colour slide film) makes a good visible light block. Have you heard this? And would it do what we want?
Thanks,
..... john  

On Sunday, December 13, 2015 at 9:32:17 AM UTC-5, Danny Morck Jensen wrote:

[Chris Fastie]

John,

Exposed developed color slide film or unexposed developed color print film both behave as good visible light blockers but transmit lots of NIR. These can turn your EasyShare cameras into pure NIR cameras. Healthy foliage will be bright and few other natural surfaces (except clouds) will be as bright. If you can get some of the processed film for free, this is a good way to block visible light. If you have to pay for film or processing you might as well buy some polyester Wratten 87 filter which has a sharper cutoff between visible and NIR.

Above: Unexposed developed Kodachrome film transmission

Above: Wratten 87 filter transmission. Small pieces available here.

Photos using these filters will have lots of information about plant health because healthy leaves reflect a lot of NIR and will be very bright. But it is hard to extract quantitative information from such photos. The photos can be used in conjunction with normal color photos of the exact same scene to make NDVI-like images. A colored filter (certain visible wavelengths are transmitted) is required to make NDVI-like images from a single photo. 

Chris

The KAPtery

[shi...@rogers.com]

Hi Chris,
        That's good to hear from you, that film will do the trick. Can I assume then that film performs the same function as the red/blue combo filter? Any idea where I might find some developed film. (I'm reluctant to get a roll developed for one or two small pieces.) I've been looking around for some, but haven't been able to find any. If not, there's a store near here that sells Rosco products. They might have the equivalent of the Wratten 87 you mentioned.
Thanks,
...... john 

On Sunday, December 13, 2015 at 9:32:17 AM UTC-5, Danny Morck Jensen wrote:

[Chris Fastie]

 Can I assume then that film performs the same function as the red/blue combo filter? 

That would not be a good assumption. The first part of this note explains why we use a red or blue filter instead of one that blocks all visible light: https://publiclab.org/notes/cfastie/11-24-2015/dual-band-pass-filters

Chris

The KAPtery

[shi...@rogers.com]

Hi Chris,
        Thanks again for your help. The shortcut you sent to me has made it clear that my understanding of how filters work, and how NIR is captured, is seriously flawed. I don't know if you want to go into this with me, but let me start with the two sections of the article that left me really confused.
" Photos from cameras without their IR block filters are not useful for plant health analysis because visible and NIR light is mixed in each channel. By adding a new filter, it is possible to allow only one type of light to be captured by a channel. For example, if a red filter replaces the IR block filter, no blue or green light can reach the sensor and only NIR light will be captured in the blue and green channels. The red channel will capture NIR, but also red light, and the two will be mixed in an unknown proportion. So although this modified camera can capture a mostly pure NIR image (e.g., in the blue channel) it cannot also capture a mostly pure visible image."
     I thought that isolating NIR radiation (with these cameras) was simply a matter of cutting out the visible light, with a suitable filter. The clause that left me scratching my head is this one " ... only NIR light will be captured in the blue and green channels." Perhaps you could elaborate on NIR with respect to channels. I assumed that blue and green channels would be synonymous with blue and green light.
      The other thing that confused me was the description under one of the pictures, "Normal color photo of a test scene for comparing filters. Taken with a PowerShot S110 with its internal IR block filter removed, but with another IR block filter screwed onto a filter tube in front of the lens." Wouldn't adding an IR block filter to the front of the lens simply reverse the effect of removing the internal IR block?
     Thanks,
...... john

On Sunday, December 13, 2015 at 9:32:17 AM UTC-5, Danny Morck Jensen wrote:

[Chris Fastie]

    I thought that isolating NIR radiation (with these cameras) was simply a matter of cutting out the visible light, with a suitable filter.

Yes, if all you want is an NIR photo, you can block all the other light (i.e., the visible light).

 

 Perhaps you could elaborate on NIR with respect to channels. I assumed that blue and green channels would be synonymous with blue and green light.

Try this wiki page: https://publiclab.org/wiki/ndvi-plots-ir-kit. It explains how the three color channels in digital cameras allow us to do this stuff.

 

  "... Taken with a PowerShot S110 with its internal IR block filter removed, but with another IR block filter screwed onto a filter tube in front of the lens." Wouldn't adding an IR block filter to the front of the lens simply reverse the effect of removing the internal IR block?

That is exactly correct. It worked as expected.

Chris

The KAPtery

 

[shi...@rogers.com]

Hi Chris,
            OK. I think I'm a little clearer on the channel thingy. My mistake was in how I imagined the different wavelengths of light were sensed by the sensor. I'm a little fuzzy on the explanation regarding the difference between a red block filter vs a blue block filter, but another reading of the article should clarify it.
I'm going to see if I can get the Wratten 87 (or equivalent).

Thanks again for your help.

...... john

On Sunday, December 13, 2015 at 9:32:17 AM UTC-5, Danny Morck Jensen wrote:

[shi...@rogers.com]

Double polarizing filter
           Hi Chris,
         Do you remember my claim that a double polarizer can be used to remove colour? Well, I tried posting pictures to the forum but was never able to figure out how to do it; but ..... I can post them to Postimage and paste the shortcuts into my post, if that works for you. If so, the question is .... what images would you like to see?
        Just to refresh your memory, I have two cameras of interest ..... an Easyshare Z915 and an Easyshare M380, both have their IR block filters removed. If I put the double polarizing filter in front of these, and adjust it to the 'light extinuishing point' the images I get have a strong bluish red cast and virtually no other colour. I can then remove the cast by adjusting the hue, leaving a pretty much black and white image. 

...... john

On Sunday, December 13, 2015 at 9:32:17 AM UTC-5, Danny Morck Jensen wrote:

[Chris Fastie]

Hi John,

Yes I recall that the challenge was to know whether the polarizing filters were blocking all visible light. I suggested taking a photo of a glass of wine or cola. Those liquids transmit most NIR light, so they look like clear water in pure NIR photos. If the liquids do not appear to be clear in the photos, then some visible light is getting through the filters.

Your test also sounds good. Adjust the filters until you see no light being transmitted (use your eye, not the camera). If an image is recorded, then the light making the image was probably NIR. If lots of NIR is getting through the filters, the sensor will see it and there will be an image on the LCD. If the light is all NIR, then all three color channels will be all NIR and any one of them is a grayscale NIR photo. The color of the photo is an artifact of the white balance algorithm. If some visible light gets through the filters, but it is only one color (e.g., blue) then one channel (e.g., red) might be a more pure NIR photo.

You can drag and drop photos into research notes or comments on notes. You can click the image icon at this forum and drag and drop an image.

Chris

The KAPtery

[Dan Beavers]

Another test is to use a TV remote control. They use IR LEDs. If you
see the light when you look at the camera screen while pressing any
button then you are seeing NIR.

> The KAPtery <http://kaptery.com/>
>
> --
> Post to this group at plots-i...@googlegroups.com
>
> Public Lab mailing lists (http://publiclab.org/lists) are great for
> discussion, but to get attribution, open source your work, and make it
> easy for others to find and cite your contributions, please publish your
> work at http://publiclab.org
> ---
> You received this message because you are subscribed to the Google
> Groups "plots-infrared" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to plots-infrare...@googlegroups.com
> <mailto:plots-infrare...@googlegroups.com>.

[Chris Fastie]

Dan,

The challenge is to identify visible light, not NIR. We know NIR is entering the camera and being captured by the sensor. What we don't know is how much visible light is mixed in with it. We need a test to distinguish pure NIR from NIR mixed with visible.

Chris

[shi...@rogers.com]

Hi Chris,
        I've taken four pictures of diet pepsi on my counter top  ...... two using the the F717, one in colour, and another in night mode (IR) with the double polarizer at the extinguishing point; the other two with the Easyshare M380 converted to IR, one without the double polarizer, and other with it at the extinguishing point. (I lightened the second one because it was quite dark, and I have made NO effort to remove the magenta cast.) As you can see, none of them are water clear.

..... john

On Sunday, December 13, 2015 at 9:32:17 AM UTC-5, Danny Morck Jensen wrote:

[Chris Fastie]

John,

Good photos. I am assuming these photos are labeled correctly:

Above: This is from the Sony Cybershot DSC F717 in normal mode.

Above: This is from the Sony Cybershot DSC F717 in Nightmode with two polarizing filters adjusted to block most visible light.

Above: This is from the Kodak Easyshare M380 with the IR block filter removed. It is a full spectrum photo.

Above: This is from the Kodak Easyshare M380 with the IR block filter removed and with two polarizing filters adjusted to block most visible light.

The liquid is sort of clear in all of them, so that suggests all of them are capturing lots of NIR light. I think the liquid would be clearer if the light was pure NIR, but don't know for sure. Having an identical glass with pure water next to the Pepsi in each photo would answer that question. 

The comparison on the M380 with and without the polarizing filters is revealing (if I have them labeled correctly). Without the polarizing filters, the M380 is a "full spectrum" camera and the photo should include all visible light and also NIR. The Pepsi looks pretty clear even though all visible light is being recorded. When the polarizing filters are added, the liquid does not get obviously clearer. This suggests that the polarizing filter is not removing much more visible light than NIR.

The individual color channels from tinted NIR photos serve as grayscale NIR photos. 

Above: The red channel from the Kodak Easyshare M380 full spectrum photo.

Above: The red channel from the Kodak Easyshare M380 with the IR block filter removed and with two polarizing filters adjusted to block most visible light.

The red channels from all of your NIR photos look sort of like the two above. The liquid does not look like clear water. This suggests that some visible light, and maybe a lot of visible light, is making it to the sensor.

Chris

The KAPtery

[shi...@rogers.com]

Hi Chris,
          Thanks for looking at these. Yes, you have all the label right.
So ...... it's the red channel that tells the tale, wrt how much visible light is getting through? How do I view the red channel? (I don't have Photoshop.)
I will try what you suggested with a glass of water, although it would seem obvious what the outcome will be.

..... john

On Sunday, December 13, 2015 at 9:32:17 AM UTC-5, Danny Morck Jensen wrote:

[Chris Fastie]

John,

There is not necessarily anything special about the red channel. The blue channel looks very much the same in these photos. The green channel is very dark in the photos, and I'm not sure why that is.

The individual color channels can be viewed in the free programs Irfanview, Fiji, or Gimp.

Chris

The KAPtery

[shi...@rogers.com]

Hi Chris,
          Alright. I'm really confused. What is the significance of the red, blue and green channels? How should they look in a pure NIR image? Is this the right question? Would I be overstating the situation if I were to say that I'm wasting my time with this approach; and if so...... what should I do .... get the red/blue filter pack and install it?
Thanks,

..... john

On Sunday, December 13, 2015 at 9:32:17 AM UTC-5, Danny Morck Jensen wrote:

[Chris Fastie]

John,

I recall that your goal was to capture photos that had information about plant health. I don't know what kind of plants or what kind of health you are interested in learning about, so I can't know what approach is best for you. In general, the cameras that are best for capturing plant health information (single camera NDVI systems) have an IR block filter that is easy to remove without adversely affecting focus, and also allow custom white balance. If a dual camera system is used, only one camera must be converted to capture NIR and custom white balance is not important. Two identical cameras are preferred, or filters can be switched on one camera without moving it between photos. I am not sure you have the cameras which make either of these approaches practical.

In a pure NIR photo (IR block filter replaced with a true visible light block filter) all three color channels will record only NIR light and each will be a grayscale pure NIR photo. The strange color of these photos is an artifact of camera processing and has no meaning.

Chris

[shi...@rogers.com]

Hi Chris,
            Thanks again for all your help.
........ john

On Sunday, December 13, 2015 at 9:32:17 AM UTC-5, Danny Morck Jensen wrote:

[shi...@rogers.com]

Hi Chris,
          I was wondering if you can tell me if Roscolux 382 Congo Blue, and Roscolux E-colour+ Primary Red would work to block the visible light. I can get these here.
Thanks,

..... john

On Sunday, December 13, 2015 at 9:32:17 AM UTC-5, Danny Morck Jensen wrote:

[Chris Fastie]

John,

Rosco's filters generally do not block all visible light (or else they would be black). The colored filters you named block most colors but transmit others and also transmit NIR (>700nm).

The #106 Primary Red filter would be a good one for a single camera NDVI system. It blocks all blue and green light so the blue and green channels would capture only NIR light (assuming the camera's IR block filter had been removed). This is similar to the Rosco #19 Fire filter that is sold by Public Lab (https://store.publiclab.org/collections/diy-infrared-photography/products/infragram-diy-filter-pack). A more precise filter for this is Wratten 25A (http://kaptery.com/product/wratten-25a-filter).

Rosco's neutral density filters transmit most NIR and block most visible, but a little bit of all colors is transmitted: 

https://www.rosco.com/filters/SED.cfm?titleName=E211:%20.9%20Neutral%20Density&imageName=../images/filters/ecolourplus/211.jpg

To block all visible you can use developed color film or something like Wratten 87 (http://kaptery.com/product/wratten-87-filter)

Chris

[shi...@rogers.com]

Thanks Chris.

.... john

On Sunday, December 13, 2015 at 9:32:17 AM UTC-5, Danny Morck Jensen wrote:

[David Cary]

> We know NIR is entering the camera and being captured by the sensor.
> What we don't know is how much visible light is mixed in with it.
> We need a test to distinguish pure NIR
> from NIR mixed with visible.

If I needed to know how much visible light is mixed into the
photographs -- in other words, how sensitive my camera is to visible
light vs. NIR -- I would take a photograph of a solar spectrum cast
onto a white piece of paper in an otherwise dark box.

Taking a photograph of that spectrum,
then comparing what that photograph looks like with what my eyeball
sees when in roughly the same position,
should make it obvious if the camera completely blocks visible light or not.

The intensity vs. linear position in a photograph of a spectrum gives
the sensitivity of the camera vs. wavelength.

Is there an easy way to calibrate what positions in a photograph of a
solar spectrum correspond to which photon wavelengths (colors)?
Fraunhofer lines maybe?
Is there software for the Public Labs Spectrometer that can
semi-automatically do such spectrum calibration?
https://publiclab.org/wiki/spectrometer

"... in 1800 ... Herschel used a prism to refract light from the sun
and detected the infrared, beyond the red part of the spectrum,
through an increase in the temperature recorded on a thermometer. He
was surprised at the result and called them "Calorific Rays".
-- from
http://en.wikipedia.org/wiki/Infrared#History_of_infrared_science

--
David Cary
+1(918)813-2279
http://OpenCircuits.com/
http://david.carybros.com/

[shi...@rogers.com]

Hi Chris,
         So, basically you're saying that for my camera, which has the IR filter removed, Rosco #19 Fire, Wratten #25A, or Wratten #87, should remove the remainder of the visible light?
Thanks,

[Chris Fastie]

David,

Excellent idea. The concept behind the Public Lab spectrometer (taking a photo of a diffraction pattern) is perfectly suited for this. It can be done with materials present in most households (DVD, CFL bulb, camera). Just set up the camera to photograph the spectral pattern formed when the light from a compact fluorescent lamp passes through a narrow slit and then through the plastic from a DVD. Without moving anything, replace the CFL with a tungsten or halogen bulb and take another photo. The diffraction pattern of a CFL has easily recognized emission peaks so you can know where in the photo the light of 700 nm wavelength is recorded. The photo of the diffraction pattern from the other bulb (which has a continuous spectrum through the visible and into the NIR) will record on both sides of 700 nm (assuming the camera is sensitive to NIR). Now insert the filter which you suspect blocks all visible light. If anything is recorded on both sides of 700 nm, then some visible light is passing through the filter (and you can know which colors are being transmitted).

It's not quite as easy as taking a photo of a glass of wine and a glass of water, but maybe just as effective.

Chris

[Chris Fastie]

John,

I am not sure what you mean by remainder. Only the Wratten 87 will remove all visible light.

[shi...@rogers.com]

Thanks Chris,
           That's what I wanted to know.
....... john