Infrared photos with game camera

[pdpa...@gmail.com]

Does anyone have any experience taking NDVI equivalent pictures with an IR capable game camera? Is this possible?

I assume these cameras already lack the IR filter, but they must have two modes that probably depend on a light sensor so that they can take daytime visible spectrum photos or nighttime NIR-only photos. So I guess the question is can this mode setting be hacked or tricked in some way so that it takes a hybrid photo?

Game cameras have the advantage for my research that they're already designed to be mounted and left outdoors, but they're more affordable than some security cameras that have been suggested for this use.

To be most specific, I have four Moultrie M-880i Gen 2 cameras that I was planning to use as time lapse phenology cameras, but would love to be able to use them as time lapse NDVI cameras.

Any ideas would be appreciated. Thanks.

[Chris Fastie]

This might be very easy, but some testing is required to know if it will work, and you might need two cameras for each NDVI installation.

The typical trail camera has a near-infrared mode so that it can take photos or video at night using NIR LEDs as illumination. (It could also do this with visible light LEDs, but they might scare away the wildlife.) The camera might have two separate cameras for visible and NIR capture, or maybe a filter is moved behind the lens.

The camera probably decides to use visible or NIR mode based on a light sensor. This would make it easy to hack with a piece of black tape over the sensor which could put the camera into permanent NIR mode. That would allow you to take timelapse NIR photos at will. Another unmodified trail camera could take matching RGB photos and the pairs could be aligned and used to produce NDVI images.

An important issue is whether the NIR mode is pure NIR or just full spectrum. In other words, is it a camera with no IR cut filter or a camera with the IR cut filter replaced with a visible blocking filter. If it is a full spectrum camera, you should add a visible blocking filter like a Wratten 87. If you add this in front of the lens it will compromise the tidy weatherproofness of the camera, so some engineering might be required.

Another issue is whether the NIR mode can handle full sunlight. It is designed for night time photos with weak IR LEDs, so it might automatically crank the ISO way high. This adds crazy noise to the little sensor and the photos are grainy. A quick test can determine whether an NIR photo taken in daylight is acceptable.

Another issue is where the daylight sensor is. If this is a separate sensor, it is easy to cover up. But they might use the camera as the light sensor, and you can’t cover that up and also take photos. So you need to find the light sensor.

If the camera has a separate light sensor, it should be possible to make NDVI images with only one of these cameras. The sensor could be covered with a housing with an LED in it. When a normal RGB photos was desired, the LED could be lit by a microcontroller, and when an NIR photo was desired, the LED could be turned off. In this way, pairs of RGB and NIR photos could be taken sequentially. I’m not sure how you would synchronize the microcontroller and the trail camera intervalometer. This might require a clever hack, but there is plenty of clever around.

Another desirable hack would be to disable the NIR LEDs which waste battery power when you are taking NIR photos in daylight. This could be done either with a strategically directed soldering iron or a more generally aimed hammer.

 

Chris

[Jeffrey Warren]

Hi all- this is a great question, and we did try it a long time ago, when trail cams were not very good resolution:

https://publiclab.org/notes/warren/6-27-2011/slideshow-modifying-plantcam-timelapse-infrared

--
Post to this group at plots-infrared@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-infrared+unsubscribe@googlegroups.com.

[Chris Fastie]

Wow, that research note is so old it won't load on my phone. It looks like back when Jeff was just a youngster he was removing the IR cut filter from the lens. That's an approach I failed to mentioned previously. If the NIR mode on the trail cam is not usable (doesn't work in daylight, can't disable LEDs, too noisy) you might be able to remove the IR cut filter on the regular camera to turn it into a full spectrum camera. I also failed to mention that you could use a red filter (e.g., Wratten 25) to make individual photos which convert directly to NDVI images. A red filter could be added to either the NIR camera or the modified RGB camera.

Chris

[Prahlada Papper]

Thanks for your tips, Chris and Jeff.

As far as I can tell there's only one camera in this. At least there's only one pinhole.

I was also not able to find a light sensor anywhere on the particular model of camera that I have, so I guess it's using the camera sensor itself. That's too bad. It does seem like most other game cameras, including older versions of this model, have a separate light sensor, so I may invest in a different model if I want to continue with this idea.

I did get the camera to take some IR pictures in lowish light. They came out pink and don't give good results when I tried putting them into the InfraGram online tool. Do I understand correctly that this is probably because they're full spectrum and NIR is being stored in the red channel along with actual red light? So I'd need to put in a red filter to get single shots that could be analyzed in a way that approximates NDVI?

Thanks again.

[Chris Fastie]

It's possible that the ambient light level is being sensed with the motion detector which is the rectangular thing above the camera lens. It might be possible to cover that whole thing to put it into NIR mode.

We still don't know if the NIR mode is pure NIR or full spectrum. You can get some hints by doing a few tests. Take some NIR-mode photos of a source of pure NIR light such as a remote control for a TV. Take some photos in the dark and some in room light. Also take some NIR photos outdoors of green plants. Then see which channels in the jpegs captured which light. Post some photos here and we can help. Also listen to the camera when it switches from RGB to NIR mode. Does it sound like a filter is being moved behind the lens? 

To make photos which can be directly converted to NDVI two things are required. Any IR cut filter must be removed, and another filter which blocks all light of at least one color (R,G,B) but not NIR must be installed. Typically a red filter which passes no blue light is used so the blue channel of photos captures only NIR light. Then that channel is used for NIR data and the red channel is used for red data to compute NDVI. Without the red filter all three channels will have a mix of visible and NIR light in proportions that are a functions of several things that are poorly quantified.

Chris

[Chris Fastie]

Funny story. A year ago I was scrounging around the local hunting store for scraps of carbon fiber tubes cut when they make arrow shafts. They had just cleaned up the shaft cutting area and thrown the scraps in the dumpster, so I dove in and found lots of little pieces to use for making parts for Picavets and pendula. Also in the dumpster was a trail camera which I assumed was broken but I took it anyway (it has 32 infrared LEDs, an infrared sensor, a camera, and lots of other things that could come in handy, right?). I never put batteries in it because if it worked they would not have thrown it away, right?  

Wrong. I put batteries in it today and it seems to work fine. So I have a trail camera now. It is a Bushnell Trophy Cam (Model# 119435). The copyright date on the manual (found online) is 2010, and it has a maximum of 8 megapixel photo resolution and 720x480 video. 

Above: The Bushnell Trophy Cam. The light sensor which determines whether to use normal or NIR mode is labelled.

It has one camera, and you can hear a click when it switches between normal and infrared mode. So a filter is being swapped when the mode changes. The normal RGB photos look good, so they are not full spectrum photos, and the infrared photos look good if taken under indoor lighting, so they must be full spectrum photos. So an IR cut filter must be moved into place for normal RGB photos and moved away for NIR photos.

Above: This is a photo from the Trophy Cam in NIR mode with normal room lighting and the NIR LEDs covered up. It looks like a normal grayscale photo, so I assume it is a full spectrum photo (RGB with NIR mixed in). A pure NIR photo would be very dark and grainy (or require a long exposure) because the room lighting does not have a very large proportion of NIR.

My previous suggestion about determining which color channels were capturing NIR can be ignored (at least for this camera). The NIR photos taken by the camera are grayscale photos with only one channel -- they are not color jpegs. When the light sensor detects that NIR mode should be used, a filter is removed from the optical path and a grayscale image is saved. So the NIR photos are not useful for direct transfer to NDVI (that requires that visible light and NIR light are captured in separate channels). The photos are also not useful for producing NDVI with a pair of photos (one visible and one NIR) because they are full spectrum photos with visible and NIR mixed. However, a visible block filter (e.g., Wratten 87) could be added so only NIR light reaches the sensor. That would produce a pure NIR photo which could be used with a normal photo of the same scene to produce an NDVI image. Tests would be needed to determine if enough pure NIR light reached the sensor to take an acceptable photo.

It was easy to find the light sensor that is used to decide whether to take a normal or NIR photo (see photo above). It is in the array of infrared LEDs. If that little sensor is covered up, infrared photos are taken and the infrared LEDs turn on to illuminate the scene. 

It might be that other trail cameras take color photos (with three channels) when taking infrared photos. If so, then they are capable of more tricks than mine is. 

Above: a color photo from the Trophy Cam with an 850 nm NIR LED in the scene. The color histogram is for the marquee on the purplish glow of the LED. The camera is apparently slightly more sensitive to 850 nm light in its red channel compared to the blue channel. Because this is a color photo, the IR cut filter was in place in front of the sensor, but some NIR obviously passes through it. This photo was taken by exposing the light sensor to lots of light but blocking most light from the scene captured by the camera.

This camera has a wildlife detection system. It apparently uses both motion and heat to detect animals. The only operating mode is to start taking photos or video when wildlife is detected. There is no intervalometer mode. It might be that most current models have an intervalometer mode, but it would be good to check that if you want to take photos at regular intervals.

Every YouTube video I watched about trail cameras pointed out the highly desirable feature that they have a camouflage pattern. Mine is just solid black. Maybe that's why it was in the dumpster.

Chris

[Chris Fastie]

The old game camera works pretty well. I let it run last night in my kitchen, pointed at the compost bin on the counter. The detection of activity does a good job, although I don't know if there were any critters in the scene that did not get detected. The room was completely dark, so all the clips are recorded in infrared illuminated by the IR LEDs. At this close range there is a terrible hot spot of overexposure, but maybe that will be less terrible for subjects farther away. Here is the video:  https://youtu.be/n7DgK7mTKh0

I assume the iframe below will not work in a Google Groups post?

<iframe width="560" height="315" src="https://www.youtube.com/embed/n7DgK7mTKh0" frameborder="0" allowfullscreen></iframe>