A question about IR modified CMOS cameras

Hi, it's been months since I posted here, I'm affraid I don't have as much time for nature photography as I used to 
Fortunately as I do animal imaging at work also, I do have a question today for the tech specialists here:

I have a biology research project that would need to take footage of bats in flight at night, to measure their movements in space.
I do this on a regular basis for birds, using a GH4 and short tele lenses (100-300mm).
Would it be enough to remove the IR-blocking filter on a GH4 sensor to be able to see bats at night ? From my preliminary research the answer seems to be No, as sensors in consumer cameras would have a sensibility to near IR only, not far IR as thermal imagers (e.g. FLIR cameras).

First, can anybody confirm that recent CMOS sensors do only see near IR with on-sensor filter removed ?

Also, if filming bats at night is not possible with a modified GH4, would it still be capable of capturing video in lower light (at dusk), or with enhanced contrast between animal and background, than a non-modified one ?

Sorry for the slightly off-topic questions, thanks for reading !

Manu.
 

It would depend on the exact sensor and I don't believe anybody has published how far into the IR spectrum sensors can see once the IR cut filter is removed and a visible light cut filter is installed. But they all see up into the 1000nm + range, I just don't know where it starts to roll off.

As for your second question, a lot depends on the filter you chose. If you select something like a SuperColor filter which retains visible spectrum and adds the IR spectrum then most definitely you would be able to use it like any other camera at night although likely with some color shift. I would think even with filters that cut everything below the 650nm range you would still be able to use them although you would likely want to convert everything to B&W or false color (switching the red and blue channels)

You might call the folks at Maxmax or Kolarivision or Lifepixel for more detailed info.

Thanks a lot EJ,

For the record, here is the very quick and pretty clear answer I got from maxmax.com :

"A silicon CMOS/CCD sensor can't see heat until about 500 DegF.  A microbolometer type sensor can see room temperature heat.

For near infrared, a monochrome converted camera won't help you much.  A monochrome sensor can see UV much better than a color sensor, and you gain about 1 stop of exposure in visible light.  In near IR, above 800nm, a color sensor basically functions as a monochrome device because the Color Filter Array (CFA) is open past 800nm."

Yeah I know they are very sensitive in the 800nm range and the 830nm conversions are very popular for high contrast B&W IR work. But it doesn't really answer how far into the IR spectrum the sensor is sensitive to. It's a pretty safe bet though that any of the conversions will see up to 1000nm or even 1100nm. One of the problems you run into is that the lens coatings aren't tested up in that range for normal visible light lenses so some start having all sorts of weird spots. But there are plenty of lenses out there that perform well in the photographable IR spectrum.

I should have linked this before but I did an article a while back in my newsletter that illustrates what different filters see - go to page 6:
http://www.ejphoto.com/Quack%20PDF/Quac ... 202016.pdf

Yes, what remains unclear is wether at dusk, an IR-modified camera would collect more "efficient photons" than a non-modified one.
I mean photons that would help contrast the bat against the background (sky or more probably foliage), and/or reduce global noise level by a significant margin.

I think I have to do a side-by-side test myself ...

Have you thought about using infrared lights to light up the scene? They are similar to what you would use with a CCTV security camera. This would be in conjunction with an IR converted camera. I have one that is about the size of a slide projector, but runs on 110 volts. They are very powerful and can be found in different powers. There are battery operated lights as well. They are hardly visible to the human eye, appearing only as a faint red glow, but when you look through a camera, they light up the scene. This is especially visible on an EVF on certain cameras. I use a Sony DSC-H50 and a Nikon D70 that are both sensitive to IR(using a 900nm cutoff IR filter), although the D70 not as much. A security camera would probably work if you have electric nearby or find a battery operated one. Even flashlights that don't use LEDs give off IR light and are not visible when covered with a piece of unexposed, but developed, slide film like the ends of a processed roll. LED flashlights using standard white LEDs don't give off IR. You can get lights that use IR LEDs, though. Here is a link to one company and a link to general info.


LED IR Lights

Info on IR Filming

General Info Link

and here's a link to IR bat video:

Bats in IR Video

Hopefully, this will be of use to you.

Pretty interesting indeed, thanks, I had forgotten this option because I thought IR lighting was short-distance only (I work within 20-100m approximately), but I might well give it a try. Will have to double-check that bats do not see IR, though !

Back to non-lit "natural light IR", I hope I am not pushing it, but one quick comparison that would be extremely useful at this stage would be :
- one image containing grass, trees and some sky (wide angle or normal FL), shot in late afternoon/early dusk (say f/4, iso3200, 1/60s to 1/250s depending on the hour, should be ok for Ev 5-7) with a regular camera.
- the same image shot a minute later with an IR-modified camera (same model would be best but this is not a necessity), at exactly the same exposure parameters.

This would allow to have a rough idea of how foliage/sky colors change with dim light near-IR, and also if the IR modified camera is more or less sensible to light overall.

If anybody with an IR camera has 10 minutes to perform the test one of these nights, that would be really kind and helpfull.
If there is a bird or bat in flight in the image, I send a bottle of champagne, promised !! ;o)

Here's a link about infrared:

http://www.batcon.org/resources/media-e ... =component

And a general link:

http://www.google.com/search?q=do+bats+ ... ent=safari

Maybe this experience is relevant:

A couple of years ago I set up a system for recording endangered nocturnal possums (3 sets of identical equipment for forestry teams to take into the field for surveys). I used "full-spectrum" modified EOS M cameras with telephoto lenses (mostly f/5.6, but I did also take my Sigma 120-300/2.8 into the field for comparison) and Sony video cameras (with NightShot mode) to record both video and stills.
FWIW, my proof-of-concept testing was done with a modified Panasonic G1.

The cameras are essentially blind at night, unless you add illumination. Having the cameras set up for "full spectrum" allowed us to take advantage of whatever light there was around, but out in the forest even the night sky didn't show up.
The kits I built included battery-powered IR spotlights (invisible to the possums), allowing us to record the animals well over 20m away. But the illumination is quite concentrated and usually it was easiest to use when the beam and camera were co-mounted so they were always aiming at the same thing.
Having something to illuminate and record a wide field (e.g. the entrance to a bat cave) would be a significant challenge.