More IR shots during Christmas BBQ

[DaveB]

I'm currently working with two EOS D60 bodies: each with a different IR filter embedded internally (this is the testing and adjustment phase before I send the cameras back to their owners). This afternoon I had a work Christmas BBQ to attend, so of course I turn up with two weird cameras...


D60/87C, 28-135mm IS
No processing other than sharpening/resizing
Those sunglasses were dark to the eye, but obviously not to long wavelengths...


D60/89B, 17-40mm
Only global adjustments: sharpening, channel swap, auto levels, etc

The focus adjustment in the 89B camera needs more work, as does the metering adjustment in the 87C camera (the above portrait was taken in M mode). But I thought some people here might be interested regardless.

[Sheetshooter]

Fabulous Dave,

Thanks for sharing. The pastey skin tone of the portrait is reminiscent of the old Technical Pan which had quite an extended red sensitivity.

The other shot I find quite breathtaking.

Cheers,

[Matt. K]

Saaay....if IR cuts through dark glasses then maybe it will cut through tinted windows on limmos?

[Nnnnsic]

Interesting.

With the exception of the first shot, the result is similar to what white balancing an IR filter image will give.

Question: What happens if you use an IR filter on the lens on top of this?

And is this embedding the filter directly in the camera AND disabling the IR pass filter or just the embedding?

[DaveB]

I remove the internal IR-blocking filter (which incidentally in the EOS bodies is also the anti-aliasing filter) and replace it with the appropriate piece of IR-pass filter.
Thus when you clean dust off "the sensor" you're actually cleaning it off the IR filter.

An R72 filter (which I believe is what you're referring to as "an IR filter") is about the same as the 89B, so adding one will do almost nothing. The 87C lets in a smaller range of wavelengths than the R72, so again adding an external R72 will do nothing.

[Alpha_7]

Nice work Dave, I really like both shots but number two really has a surreal quality about it, like I'm in some frozen wasteland, yet there are signs of some form of civilization.

[MATT]

Ohh.. to drunk to know, but numevber 2 trippy Love it..


Not sure how the IR atuff works baut Id love one for my d70

MATT

[lotophage]

I've seen some sites with step by step how-to's of how to remove the IR hot filter off the d70s, but I certainly don't have the rocks to mess with my beloved camera! When thinking about this though, one thing really bothered me: if the sensor is receptive to IR frequencies, why cant all the raw info be stored into a file so that bandpass filters can be applied with software? What i really want to try is to separate the IR into a channel, then use that to replace the luminosity channel for the visible spectrum. After all, traditionally a lot of people hand painted their IR shots.

An interesting side note: when I was considering modding my camera, I found a lot of Japanese sites selling IR filters and one of the commonly plugged selling points was that they can see through women's swimsuits. Apparently most bathing suit companies these days use fabrics which allow IR to pass easily as this eliminates tan lines.

[Nnnnsic]

So is the operation you've done essentially the same as screwing on an R72 or RM90?

[DaveB]

So is the operation you've done essentially the same as screwing on an R72 or RM90?

Nope.
Attaching an external filter you have four major components in the system:

• The CCD/CMOS sensor, which has a relatively flat response from visible light down deep into NIR wavelengths.
• The RGBG Bayer filter array, which is fairly transparent to the long NIR wavelengths but attenuate different areas of the visible spectrum (and the shorter NIR wavelengths that are passed by an R72/89B).
• The IR-blocking filter between the sensor and the shutter. This cuts off around the edge of visible light: very little NIR makes it through.
• The IR-pass filter (R72/89B, 87, 87C, or the RM90/87B which has an even-deeper cutoff).

Not only are you left with very long shutter times as very little NIR light gets past the internal IR-blocking filter, but the transmission of that NIR light is not constant across the wavelengths due to the interaction of the IR-blocking and IR-pass filters.

What I do is remove the internal IR-blocking filter so that the spectral response is fairly flat through the interesting NIR wavelengths, and also (primarily, really) so that the shutter speeds are reasonable. That D60/89B image was taken handheld at 1/80s, f/10, ISO 100. In fact the red channel was overexposed by more than a stop and this was adjusted in the RAW conversion (i.e. I should have taken it at around 1/180s instead, but I had forgotten that the camera's histogram is weighted towards the green channel ).

When thinking about this though, one thing really bothered me: if the sensor is receptive to IR frequencies, why cant all the raw info be stored into a file so that bandpass filters can be applied with software? What i really want to try is to separate the IR into a channel, then use that to replace the luminosity channel for the visible spectrum.

Bandpass filters can't magically be applied via software alone. The way "bandpass filters" are applied to distinguish the red/green/blue channels is by having a separate coloured filter covering each pixel on the sensor, and the software is calibrated to know what the colour of that pixel is. Are you proposing that more pixels are sacrificed by manufacturers in order to determine the IR illumination? These coloured filters (the Bayer filter array) happen to be transparent to a lot of the NIR wavelengths, which adds further complication (and is why there's an IR-blocking filter in the camera: otherwise the reconstruction of colours is thrown out!).

I certainly don't have the rocks to mess with my beloved camera!

This is exactly why people pay me to mess with their cameras for them!

[Steffen]

Saaay....if IR cuts through dark glasses then maybe it will cut through tinted windows on limmos?

I wonder what else it will cut through Inquiring minds...

Cheers
Steffen

[Onyx]

As interesting as I find IR shots, the 2nd example Dave posted especially; I can't help but feel it's still a novel way of photographing, akin to fisheeye lenses where occasional use would yield a greater impact. Why anyone would want to permanently convert their cameras for IR stumps me... The presumed target market whom would consider such conversion would be IMO avid amateurs with deep enough pockets to contemplate making one of their bodies into a gimmick™ machine®; given that there's a EOS 20Da that does so from the factory that would satisfy their IR urges.

[DaveB]

I think the wine being drunk in that first picture was red wine BTW: doesn't look very dark here.

As interesting as I find IR shots, the 2nd example Dave posted especially; I can't help but feel it's still a novel way of photographing, akin to fisheeye lenses where occasional use would yield a greater impact.

A reasonable comment. In fact the reason I'm carrying a G3/89B around at the moment is to determine if I would use it enough to justify installing it into a DSLR. The false-colour thing can look weird and interesting, but it might wear thin too quickly. The 87C filter however is one I've been using for a few years in various Coolpix and PowerShot cameras and isn't quite so "freaky".
The 87C photo here shouldn't serve too much as representative of the images one can obtain. Most of the other IR images I've posted here in the past were taken with an 87: I'll post some more soon.

Why anyone would want to permanently convert their cameras for IR stumps me... [...] given that there's a EOS 20Da that does so from the factory that would satisfy their IR urges.

No, the 20Da is not suitable for this. It still has an IR-blocking filter, it's just that the filter lets in a tiny extra bit of NIR at the edge of visible light: the distinctive "Hydrogen-alpha" wavelengths that are very interesting to astrophotographers.
I'm currently working on a 300D for a customer where I'm installing a Baader UV/IR-cut filter which sounds like it does the same thing (I've seen the transmission profile graph for the Baader, but not for the 20Da). Sure he could buy a 20Da, but converting a 300D is a lot cheaper. But again that's NOT a filter suitable for IR photography.