IR conversion options (for the 20D)

[Ray]

Since buying a Canon 40D, I now have a redundant 20D. There's not much point in having a 20D which is a back-up to a 40D which is a back-up to a 5D. How many back-ups does anyone need?

I bought a 40D as a by-product of going shopping for an EF-S 17-55/2.8 lens for my 20D.

Now that I've got a 40D, I will never use the 20D, except if I make it useful by removing the infra-red filter.

Searching the internet, I find there are lots of interesting effects one can get with infra-red photography. But I'm not particularly interested in those. One can get all sorts of interesting, weird and even ugly effects simply by manipulating normal images in Photoshop. Perhaps the infra-red effect is difficult to emulate in PS. I don't know.

My main concern is the possible benefit for normal photography in low light conditions without resorting to flash. I can't find any information on this use.

With a clear glass filter in place of the IR filter, the sensor is exposed to a greater amount of electromagnetic information for any given exposure. The frequencies are extended in both directions; more infra-red light as well as more ultra-violet light.

What effect does this have on a shot of any scene in low light conditions, apart from the obvious screwing up of color balance? Is signal-to-noise improved? Do we get less shadow noise? Is chromatic aberration a major problem because lenses are generally not optimised for such frequencies?

How would such images, converted to B&W, differ from normal B&W images? Do images become generally less sharp because of smearing? Do skin textures disappear? Do results generally, from a normal perspective, appear inferior?

These are the questions I hope someone reading this can answer.

[Tim Gray]

My main concern is the possible benefit for normal photography in low light conditions without resorting to flash. I can't find any information on this use.

[{POST_SNAPBACK}][/a]

I don't think IR is what you are looking for - it's not like the "heat map" kind of image we see on tv on the si fi channel.

In terms of low light performance, it's actually about 2 stops worse than pre IR conversion - I have an xti modified by max max and it consistently requires a +1.66 to 2.00 +ev.  There are some af issues as well.

I'm taking mine to British Columbia, Pacific Rim Nat Park in a couple of weeks, there will be lots of green there,  in the meantime here are some samples:

[a href=\"http://www.timgrayphotography.com/galleries/20080310-deathvalley/index_2.html]http://www.timgrayphotography.com/gallerie...ey/index_2.html[/url]

http://www.timgrayphotography.com/gallerie...ared/index.html

There are some BW presets that do a reasonable job of emulating IR in a BW conversion, but so far I haven't seen any compelling technical (or artistic) justifications.   I use gadgets like this, and the lens baby to add a little variation to my shooting, but only because I'm a bit of a gadget freak.  Eventually they just end up gathering dust.

[Ray]

I don't think IR is what you are looking for - it's not like the "heat map" kind of image we see on tv on the si fi channel.

In terms of low light performance, it's actually about 2 stops worse than pre IR conversion - I have an xti modified by max max and it consistently requires a +1.66 to 2.00 +ev.  There are some af issues as well.

[a href=\"index.php?act=findpost&pid=184106\"][{POST_SNAPBACK}][/a]

Interesting images, Tim. It seems like you're into 'real' infra red photography. If you need to provide up to 2 stops additional exposure, I take it your modification included replacing the infra red filter with another filter that blocks out normal light. Is that right?

The option that I am interested in replaces the infrared filter with a clear glass filter. I understand that in normal lighting conditions, or at least sunny conditions where there's plenty of infra red light present, exposure has to be reduced by about 1.6 stops.

Another advantage (or disadvantage), is the removal of the AA filter which is part of the IR filter, so one gets to experiment with possibly slightly crisper images.

[Misirlou]

The thing about IR is that you really can't predict how much reflectivity you'll get from something by looking at it. Some objects will be wildly reflective, and others nearly black, when they may appear to have the same visual brightness.

It's best not to think of IR as extra energy, but as a completely different color a thing can emit. Not only is it a completely different color, but we're all colorblind to it. So I don't think it will be much of a help under low light conditions, unless you only want to shoot things in low light that have high IR reflectivity, like chlorophyl laden plants.

I happen to really enjoy looking at IR shots myself, but it's not like having a night vision scope that just intensifies brightness.

[Tim Gray]

Interesting images, Tim. It seems like you're into 'real' infra red photography. If you need to provide up to 2 stops additional exposure, I take it your modification included replacing the infra red filter with another filter that blocks out normal light. Is that right?

The option that I am interested in replaces the infrared filter with a clear glass filter. I understand that in normal lighting conditions, or at least sunny conditions where there's plenty of infra red light present, exposure has to be reduced by about 1.6 stops.

Another advantage (or disadvantage), is the removal of the AA filter which is part of the IR filter, so one gets to experiment with possibly slightly crisper images.
[{POST_SNAPBACK}][/a]

Don't know if you've seen the maxmax explanation page, but here's a link:
[a href=\"http://www.maxmax.com/IRCameraConversions.htm]http://www.maxmax.com/IRCameraConversions.htm[/url]  

The aa filter is replaced with an IR filter, which does have the effect blocking light, reqiring a positive EV.  Maxmax also does an AA filter removal on the 5D, but I don't think that effects the image tonality, so I don't know if the IR filter layer is removed as well...

[Panopeeper]

I am in a situation like Ray - my 20D is idle, and I am thinking about IR panoramas.

Following is based on information picked up everywhere, none experience.

1. Replacing the IR filter with an IR-only, like maxmax offers is IMO nonsense: that makes the modified camera useless for anything else.

By replacing the original IR filter with the UV + visible + IR version, one can have everything: one can use an IR cutoff filter and work as before, or an IR filter of the choise. For example the B+W filter 093 is "more infra" than the maxmax "IR only".

UV light should not be an issue at all, for the coating of modern lenses includes UV filtering as well - at least I received this response from Nikon.

2. Focusing appears to be the main issue, and according to what I found, it is not solvable generally. Some other "converter company" I found (but now I don't find them) expected you to send a (one) lens with the camera for calibration of the camera. Then *with that lens* the camera would focus well.

Apparently one needs to try around and find out, which lens is suitable at all, perhaps with a manual correction of the focusing.

[Tim Gray]

By replacing the original IR filter with the UV + visible + IR version, one can have everything: one can use an IR cutoff filter and work as before, or an IR filter of the choise. For example the B+W filter 093 is "more infra" than the maxmax "IR only".
[a href=\"index.php?act=findpost&pid=184221\"][{POST_SNAPBACK}][/a]

Not everything, no free lunch.  Once the filter is on the camera it's like looking at a black cat in a coal mine.

[Panopeeper]

Not everything, no free lunch.  Once the filter is on the camera it's like looking at a black cat in a coal mine.

The difference between the replacement filters offered by maxmax is just this: the IR only version cuts most of the visible light, the IR+visible+UV does not cut anything.

Others I found earlier didn't even offered an IR only filter, only a "clear glass".

[Panopeeper]

I found another company; they are selling the replacement filters if one wants to mount it himself, for $145, or they do the replacement themselves for $450:

http://www.lifepixel.com/index.html

see the Do It Yourself section: very detailed explanation, how to do it, BUT it does not deal with focusing adjustment.

[John Sheehy]

So I don't think it will be much of a help under low light conditions, unless you only want to shoot things in low light that have high IR reflectivity, like chlorophyl laden plants.
[a href=\"index.php?act=findpost&pid=184170\"][{POST_SNAPBACK}][/a]

It would seem to me that a good deal of the inefficiency of incandescent lighting is due to the fact that it emits a lot of IR, and that shooting without the filter will make the camera mose sensitive in IR light.  Of course, colors will be off for color photography.

Personally, I'd like a camera without the IR-cut filter, and without the CFA, but with a mild AA filter.  That would give a B&W camera with about 1.5 stops more sensitivity in cool lighting, and maybe 2 more in wqarm lighting.

That would be a bit more difficult, I understand, as the CFA needs to be chemically dissolved off of the sensor.

A B&W 1Dsmk3!  That's what I want for my birthday!

You could probably get usable results at ISO 200,000 with such a camera.

[Tim Gray]

The difference between the replacement filters offered by maxmax is just this: the IR only version cuts most of the visible light, the IR+visible+UV does not cut anything.

Others I found earlier didn't even offered an IR only filter, only a "clear glass".
[a href=\"index.php?act=findpost&pid=184232\"][{POST_SNAPBACK}][/a]

Here's a quote from maxmax in the IR + Visible

"Drawbacks on IR+Visible

Using  digital SLR camera means that you look through the lens.  This is great when you can physically see through the lens, but with a true infrared filter on the camera, you will not be able to see anything!  After taking the picture, you can review the shot on the LCD display on the back of the camera. However, the camera image sensor (the CCD) is hidden behind the camera's shutter until you take a picture.  Therefore, you cannot preview an infrared picture prior to taking the shot."

[Panopeeper]

That's right, this is the disadvantage of the filter on the lens vs over directly the sensor - when the IR filter is mounted. On the other hand, the camera still can be used with visible light or IR+visible.

[Ray]

As I understand, the mis-focussing issue will apply when the lens filter is in place to block out normal light. With camera on tripod, one could focus first, then screw on the filter, guess the exposure and take some bracketed shots.

However, the focussing that's correct for normal light will not necessarily be correct for IR. This applies when one is taking IR-only shots.

When taking shots with no lens filter, when the predominant lighting consists of visible rays, focussing should be as normal, except the infra red rays might still be out-of-focus. In other words, one can get a type of chromatic aberration. An expensive, high quality lens starts behaving like a cheap lens with fringing and smearing; simply not sharp at wide apertures.

It's this aspect that worries me. I wondered if anyone had experience with this. The problem might not be as bad as I imagine.

One can always reduce this effect by stopping down, but one has then lost the advantage of the low light capability of removing the in-camera IR filter.

[Misirlou]

It would seem to me that a good deal of the inefficiency of incandescent lighting is due to the fact that it emits a lot of IR, and that shooting without the filter will make the camera mose sensitive in IR light. [a href=\"index.php?act=findpost&pid=184239\"][{POST_SNAPBACK}][/a]

You're missing a crucial consideration. Regardless of how much IR light may fall on an object, it may or may not not actually reflect any of it. Consider a large body of water at high noon. It's being bathed in huge quantities of IR from the sun, but water just doesn't reflect IR very well, so it looks black to an IR sensor.

You'll get similar issues in low light situations indoors. You may be trying to photograph IR black bodies, whether or not they're bathed in IR from lightbulbs and so forth.

Much of this depends on the physical properties of the material in question, and the exact frequency of electromagnetic energy you are sensing. And it's not predicatble, unless you're well-versed in the IR reflectivity of common materials. Again, think of IR is an invisible color.

[Ray]

You're missing a crucial consideration. Regardless of how much IR light may fall on an object, it may or may not not actually reflect any of it. Consider a large body of water at high noon. It's being bathed in huge quantities of IR from the sun, but water just doesn't reflect IR very well, so it looks black to an IR sensor.

[a href=\"index.php?act=findpost&pid=184417\"][{POST_SNAPBACK}][/a]

If you are taking IR-only images, then this is a consideration. Objects that are poor reflectors of IR will obviously appear dark, such as the sea, whereas good reflectors of IR such as foliage will appear lighter than usual

But suppose you remove the IR filter inside the camera for the purpose of allowing more radiation to reach the sensor. The sensor then captures the same reflected light as in a normal exposure, plus reflected IR frequencies.

In that situation, there would be nothing in a captured scene which would be darker than the same scene captured with the unmodified camera at the same exposure and ISO. But there would be objects in the scene that are lighter than the normal capture. In fact any object in the scene that is able to reflect IR frequencies would appear lighter, provided the light source contains such IR frequencies.

The main problem as I see it is that lenses are not designed to focus such a wide range of frequencies. The fact that lenses sometimes have to be modified for precise focussing of IR-only scenes, implies that an umodified lens which is focussing properly in relation to visible light might produce a type of chromatic aberration in relation to the IR part of the spectrum which is unable to be focussed in the same plane as the visible part of the spectrum.

[samirkharusi]

I sense a bit of misinformation and confusion on this thread. Since I have played with a modded 20D, perhaps I can shed some light (and IR!) on the topic:

Hutech modifies these DSLRs routinely for astro use. You can have many kinds of modifications, the most flexible is to replace the UV/IR Blocking filter on the sensor with a glass of the correct thickness but that also has anti-reflection coatings. Cost is $400 to $500 but if they supply the brand new camera you also get a one year warranty. Job is done very professionally. The camera arrives as if it had come straight from the factory. I bought a 20D modified like this from them. But note that you will be better off with a newer model that has Live-View. Reasons given below.

The above modded camera is about 3.3x as sensitive in tungsten lighting as an unmodded one. My measured data in tungsten lighting are available here:
http://www.pbase.com/samirkharusi/hutech_mods

The camera still autofocuses correctly, but you can no longer use Auto White Balance and colors cannot be balanced for normal photography. A few lenses are indeed able to focus both IR and visible simultaneously, wide open, eg the Canon 600/4L IS and the Canon 100/2.8macro USM, but the majority cannot. You can compromise by closing down the aperture.

To use the camera for normal photography you need a UV/IR Blocking filter. Hutech  sells such filters that go behind the lens but are in front of the mirror box. Ie one filter suffices for ALL your EF lenses, but it cannot be used with an EFs lens.

Two kinds of UV/IR Blockers are available; a rectangular bandpass filter for astro use, and a Canon-shaped bandpass filter that makes your camera behave as if it were never modified. The rectangular bandpass filter requires that you use Custom White Balance for normal photography. But the colors come out quite well, eg
http://www.pbase.com/samirkharusi/image/91303851
But if you are shooting catalog setups then you are better off using the Canon-shaped-UV/IR Blocker. This filter also enables Auto White Balance. Each filter costs around $300.

For IR photography you use an IR-pass filter that blocks all visible light. Another $300. Or you can cut your own Wratten 89B gel for about $60. Since this is in front of the mirror box, the viewfinder blacks out. Autofocus is fully functional and the IQ is only limited by how well your lens focuses IR. See:
http://www.pbase.com/samirkharusi/image/54146316/original
Exposure speed in IR is about the same as an unmodded camera in daylight. Auto exposure does not work, but the back-of-camera histogram is functional for judging exposure. Since the viewfinder is now black and useless, you will really prefer a DSLR with Live-View! I.e. a 40D or 450D.

I have measured the spectral sensitivity of a modded 20D with and without filters, data displayed here:
http://www.samirkharusi.net/spectrograph.html

After digesting all that above, I'd be happy to answer any questions you may still have    

Other camera modders tend to offer only the option of the IR-pass filter on the sensor. This makes the viewfinder functional but the camera becomes dedicated to IR, useless for normal photography. I am unsure as to how well the autofocus works since the focus sensors see both Visible and IR and the lens may focus these slightly differently. By placing the IR-pass filter in front of the mirror box, the autofocus sensors in the camera can see only IR, so the focus will be as good, or as bad, as the lens is capable of in IR.

[Ray]

The above modded camera is about 3.3x as sensitive in tungsten lighting as an unmodded one. My measured data in tungsten lighting are available here:
http://www.pbase.com/samirkharusi/hutech_mods

The camera still autofocuses correctly, but you can no longer use Auto White Balance and colors cannot be balanced for normal photography. A few lenses are indeed able to focus both IR and visible simultaneously, wide open, eg the Canon 600/4L IS and the Canon 100/2.8macro USM, but the majority cannot. You can compromise by closing down the aperture.

[a href=\"index.php?act=findpost&pid=184639\"][{POST_SNAPBACK}][/a]

This is precisely the modification that I am interested in. There's a company in Australia by the name of Khromagery that provides a number of options, different filters for different purposes. The filter that I would request would be a CLR 280nm such as the Schott WG280. That extra sensitivity in tungsten lighting could be very useful. 3.3x the sensitivity seems quite significant. That's about 1 1/2 stops, isn't it.

I just hope my EF-S 17-55/2.8 happens to be one of the few lenses that can focus the IR and visible light simultaneously.

[Misirlou]

If you are taking IR-only images, then this is a consideration. Objects that are poor reflectors of IR will obviously appear dark, such as the sea, whereas good reflectors of IR such as foliage will appear lighter than usual

But suppose you remove the IR filter inside the camera for the purpose of allowing more radiation to reach the sensor. The sensor then captures the same reflected light as in a normal exposure, plus reflected IR frequencies.

In that situation, there would be nothing in a captured scene which would be darker than the same scene captured with the unmodified camera at the same exposure and ISO. But there would be objects in the scene that are lighter than the normal capture. In fact any object in the scene that is able to reflect IR frequencies would appear lighter, provided the light source contains such IR frequencies.

The main problem as I see it is that lenses are not designed to focus such a wide range of frequencies. The fact that lenses sometimes have to be modified for precise focussing of IR-only scenes, implies that an umodified lens which is focussing properly in relation to visible light might produce a type of chromatic aberration in relation to the IR part of the spectrum which is unable to be focussed in the same plane as the visible part of the spectrum.
[a href=\"index.php?act=findpost&pid=184609\"][{POST_SNAPBACK}][/a]

Yes, that's all generally correct. Here's one more thing to think about though. Notice that traditional IR photography demonstrates some odd contrast effects. Some of that has to do with the nature of the IR "color." Many objects either reflect a lot of IR, or very little, so you get very contrasty results, and sometimes areas of totally blown out texture. Then there is also an effect due to sensor sensitivity. Dedicated digital IR sensors (in satellites anyway) have significantly lower radiometric resolution in the IR bands than they do in the visible ones, which would just increases the contrast problems in terrestrial DSLR photography.

So even if you were able to use additional energy in low light conditions, it might actually hurt your overall contrast. You might have higher dynamic range light input to the sensor than if you were only looking at the visible spectrum. And with the sensor reading all of that energy through a regular Bayer filter, you could never pull out the IR compenent in post processing. I suppose a new type of filter matrix (r-g-b-IR) might be a really fun toy...

It took me a while to figure out why I had problems shooting flowers in the color film days. Eventually, I realized some flowers reflect a lot of light in bands that my film could see, but I couldn't. Played all sorts of games with color exposure and rendition.

You're absolutely right about focus. A given lens element will focus every frequency of light at a different plane. That's why we use multiple elements to try and get them all close to the same place on the film or sensor. But the manufacturers compute their lens formulas for optical wavelengths. I have no idea what would happen to the IR frequencies once they've passed through all 17 or so elements of some of our modern zooms. Good argument for using simple, old primes for IR photography.

[Ray]

So even if you were able to use additional energy in low light conditions, it might actually hurt your overall contrast. You might have higher dynamic range light input to the sensor than if you were only looking at the visible spectrum. And with the sensor reading all of that energy through a regular Bayer filter, you could never pull out the IR compenent in post processing. I suppose a new type of filter matrix (r-g-b-IR) might be a really fun toy...

[a href=\"index.php?act=findpost&pid=184706\"][{POST_SNAPBACK}][/a]

Every time one uses ISO 3200 instead of underexposing at ISO 1600, one is taking the risk that the dynamic range of the scene might be so great that the bright parts will be blown. (On the basis that ISO 3200 in Canon cameras is merely ISO 1600 amplified with all noise.)

The fact that the presence of invisible light might increase the risk of this happening is certainly something to bear in mind when using such a modified camera. I would therefore suggest one refrain from using false ISOs such as ISO 3200 and always bracket shots at ISO 1600 just in case there's an unexpected area in the scene that is surprisingly bright.


If color accuracy is considered to be less important than other benefits, such as the possibility of significantly greater sensitivity in tungsten light and perhaps greater sensitivity in most low light conditions, particularly as the sun is setting, then we have a tool in which high ISO performance has been increased.

Even in normal daylight we would presumably have the option of using faster shutter speeds than would be possible using an unmodified camera at the same ISO and aperture.

It seems to me that the downside is not just that such results might generally be more suitable for B&W processing as a result of a yukky color balance, but that one's lenses might no longer be sharp.

This is the reason I started this thread. To find out if anyone has had experience using a camera modified in this way, substituting a clear glass filter for the IR filter in order to improve low light performance.