When removing an infrared blocking filter, is a passing filter needed? Why?

<p>So I decided to follow the intructions I found here <a href="http://www.dpreview.com/forums/post/29011053">http://www.dpreview.com/forums/post/29011053</a> instead of spending an arm and a leg with LifePixel because I've always wanted an infrared camera that didn't need long exposures. Removed the infrared blocking filter, assembled the sensor, soldered, and finished it.</p>

<p>I have since been reading that part of what LifePixel does is install a infrared passing filter in place of the blocking filter. I seemed to have missed this somehow. Never the less, the photos I've been taking look by all means IR, but when I place my own Hoya IR filtter, they look wholly different but still infrared. For instance, when taking pictures of my computer monitor with the Hoya filter on, the room appear infrared, and the monitor appeared red. Without the Hoya filter, the room still looks infrared, but the computer monitor seems to retain most of its normal (visible) color.</p>

<p>My question is: what exactly did I do to my camera? What IR wavelength is it recording? Is it now shooting a hybrid of visible and infrared light?</p>

The term "IR filter" is used in two opposite ways: though most filters are named after the colour of light they let through, IR filter is used both for a filter that passes or blocks IR.<br><br>Apart from blocking all visible light (necessary if all you want to record is the IR light), the replacement glass is needed to lengthen the optical path length again, which is shortened when you remove the IR blocking filter (by about a third of the filter's thickness). If you don't, sharpness on the sensor depends on depth of focus being great enough to cover up the error in focus.<br>Focus, however, is also affected by the long infra red wavelengths now also contributing to the image formed on the sensor. Longer wavelengths also need a longer path way to form an image, so depending also on what the lens does, the replacement path lengthener (which you did not out in) may make things worse and the right thing to do is not insert such a thing.<br><br>Your camera is now indeed recording both visible and IR light. Nothing wrong with that, though when you only want to record the IR, you need that opaque Hoya on your lens. Then it would have been better if the opaque filter was in front of the sensor and not in the part of the optical path you need to see and frame the subject.<br>I did the same with a Nikon D70s: remove the IR blocking filter with nothing going in to replace it (no soldering, just a few screws and the filter came out), and have not noticed any focus issues. I found a ND filter worked well enough to balance visible light and IR. No need for an opaque IR-only pass filter, so i could still see something through the viewfinder. I think you should be fine with the Hoya filter on the lens, if and when you want the IR-only effect.

<p>I'm actually really excited with this mistake. I had thought I could remove the filter, and in the future have it record both IR and visible, but I guess that is the very first step!</p>

<p>Are you saying the extra glass from the IR pass through filter would correct the focus problems? That would explain why my Nikon A to Pentax K adapter is correcting this issue. I guess because the Pentax lens I am using with it is calibrated for IR and the extra glass is helping bring that spectrum into focus? With that combo, I can focus more toward infity than the Nikon lenses can. Haven't truly tested it though.</p>

<p>I'm wondering if there are hot mirrors and IR filters I could could attach to the front of the lens to switch between just visible and just IR and still retain normal "fast" shutter speeds...I don't want to keep using the Hoya just for pure IR. For long exposures, it would still be great, but the point was to eliminate those.</p>

<p>There will be a focus shift when photographing in the Infra-red anyway, since with most (all?) lenses the IR focus is different from the visible focus and the FL of the lens is effectively lengthened.</p>

<p>A microscope slide might provide a suitable thickness of "IR passing" glass to bring the visible focus back into alignment. A true IR passband filter over the sensor theoretically does the same thing as a similar filter over the lens; <em>except</em> that reflections from a near-opaque front of lens filter can give rise to a hot-spot effect.</p>

<p>The IR filter needs to be replaced with glass of the same thickness. The glass increases the focal distance by about 1/3rd of its thickness. This is trivial if the filter is in front of the lens, but highly significant in the rear.</p>

<p>Secondly, to get true IR effects, the blocking filter needs to be replaced with a low-pass filter. Companies which do this conversion typically substitute a 600 um or 900 um cutoff, The former admits some visible light while the latter only IR. In either case, a filter on the sensor allows full use of the optical viewfinder, whereas a filter in front of the lens (e.g., #72) is virtually opaque, and works only with live view. The conversion may include focus correction.</p>

<p>You can Google for providers, which includes: <br>

http://kolarivision.com/?gclid=CLqturiXicoCFVU8gQod7Y8HWw</p>

<p> </p>

<blockquote>

<p>The IR filter needs to be replaced with glass of the same thickness.</p>

</blockquote>

<p>I would say "should".</p>

<p>For CD players, the thickness of the CD is part of the optical system. The focal length is so short that the CD thickness is a significant part of the path, and it won't focus without it.</p>

<p>As far as I know, that isn't true for SLR lenses. But it will make the optical path distance different from that to the view screen and AF sensor. You will need to change the focus such that it seems to be focusing nearer than the viewfinder and AF show. This is the same direction that you need to change for IR relative to visible focus. </p>

<p>But the prices of DSLRs are low enough that sacrificing an older camera probably makes more sense than sending it out. That is, if you can do it. The camera will be difficult to use for normal photography, so just don't do it. Only use it for IR, where again it won't focus right.</p>

<blockquote>

<p>Without the Hoya filter, the room still looks infrared, but the computer monitor seems to retain most of its normal (visible) color.</p>

</blockquote>

<p>I suspect that the IR sensitivity mostly comes in the red channel. </p>

<p>There is the RGB Bayer filter in front of the sensor, and the transmission spectra of those filters affects what you see. If the CRT gives off R, G, B, but not IR, then it will still look the same.</p>

<p>For other subjects, I would expect the green and blue to be pretty much the same, but those filters might also have some IR transmission at some point. Look at an IR remote control, such as your TV or DVD player. See if it is just red, or maybe some green and blue, too. Incandescent lamps give off a lot of IR, which will be most of what you see in a room.</p>

<p>If you can find a green or blue filter with enough IR transmission, you could use that in front of the lens. Then separate out only the red channel later, as the IR image, in post processing.</p>

Yes, the colour image with IR added will be predominantly red. To turn it into something pleasant to see, it has to be converted to B&W. That conversion provides more opportunity to drop colour and concentrate on the red channel.<br>The Bayer filter does not block IR to a degree that it would prevent IR forming an image, though you would get a much stronger effect without Bayer filter. But unless we can find cheap enough cameras to butcher that do not have a Bayer filter, we'll have to do with those which do have such a filter. It works.<br>The contribution of IR to the 'normal' colour image of course depends largely on the presence of IR. Little or no IR in the light forming the image (the monitor), no visible IR effect. If in such a situation the other colours aren't blocked by a filter such as that Hoya, you will get a normal image.<br>As mentioned before, i found that neutral density filters (at least the ones i tried) are transparent for IR, and with a reasonably dark ND filter on the lens you can still see through the viewfinder and get a good balance between visible and IR light to have a strong enough Wood's effect. When you replace the ND filter with an IR blocking filter, you switch back to normal, unmodified operation, with normal colours. But you will have to check focus, make sure to correct focus or stop down far enough to have DoF cover up faults a bit.<br><br>By the way: the filter needed to get an IR only image is a high pass filter, blocking all wavelengths below a cut off value. The low pass filter is what is removed. ;-)

<p>I used the term "low pass" because the frequency of infrared is lower than that of visible light. That conforms to the usual definition of filters better than wavelength.</p>

<p>The cover glass is normally in the range of 1 to 2 mm, which will affect the focal distance by approximately 1/3rd that thickness, increasing with the angle of incidence. That is highly significant with regards to focusing. This effect is independent of either optical focusing on the ground glass or auto focusing in a DSLR. Simply offsetting the focusing ring is possible, but not very precise. In addition, the focal length of the lens is longer in the infrared region, doubling the effect. To focus precisely, you need to use live view, or modify a camera which relies on an electronic viewfinder.</p>

<p>The perceived color depends on the transparency of the Bayer filter to infrared, not necessarily red in hue. Cameras with low efficiency IR filters, like the Nikon D200, D2h and Leica M8 tend to develop a purple cast, seen mainly in dark areas of the image.</p>

The longer IR focus does not double the effect of removing the glass, but negates it (partly or entrirely, depending on the lens used as well). In 'conventional' IR photography, the difference in focus between visible light (used to focus) and IR (used to form the image) is dealt with by increasing the lens to film/sensor distance. Removing the IR blocking glass has the same effect.<br>The IR blocking glass in front of the sensor moves the image back, away from the lens. So removing the glass moves the plane of focus to somewhere just in front of the sensor plane. An image focused using visible light will appear there, but then the longer IR focus is further back, closer to (or even at) the sensor plane again.<br>In my experience with a camera with filter removed with nothing to replace it going in, depth of focus is great enough to cover any remaining faults. Though in the closer and close-up range live view focusing would be advisable, yes.

Here's a test image, through a ND filter, after removing the IR blocking filter from a Nikon D70s. On the left the plain RGB, to the right of that the red, green and blue channels respectively.<br><img src="http://www.photo.net/photo/18152467">

<p>The focal length of a lens is about 2% longer for infrared light. Removing the cover glass shortens the effective focal distance. Consequently the effects are additive. At normal distances (>15x focal length), the DOF is usually large enough to cover the effect.</p>

Removing the cover shortens the distance at which the image is formed, i.e. moves the image towards the lens, in front of the sensor plane.<br>Longer wavelengths have a longer focal length, i.e. using longer wavelengths the image is formed behind the focal plane, behind the sensor.<br>Back and forth. Not additive, but subtractive.

<p>I concede your point. The cover glass makes the sensor appear closer to the lens than without it. As the angle of incidence (departure from vertical) increases, it has the effect of introducing positive field curvature, the nemesis of using RF lenses with mirrorless cameras.</p>

Hate to do this Edward, but if with RF you mean retrofocus, it has to be pointed out that being of a retrofocus design makes them better suited for such a situation than non-retrofocus short lenses.<br>Digital sensors used to (do still) have problems with light hitting wells at an angle. Something unrelated to a plane parallel piece of glass having a different power for light hitting it at different angles. The benefit of retrofocus lenses is the same for both problems. The light projected by a retrofocus lens is 'more parallel' than that of true wide angles, hitting the sensor at less problematic angles.

After a night's sleep, the penny drops. RF stands for rangefinder, i.e. lenses made to sit close to the film/sensor. Sorry!<br>Yes, being typically the 'opposite' of retrofocus lenses, RF wide angles are more prone to create problems when combined with an IR and/or anti-alias filter in front of a sensor.<br>My apologies.

<p>RF = rangefinder</p>

<p>Lenses as long as 50 mm, designed for rangefinder cameras, are physically close to the sensor. Light reach the corners of the sensor at an angle of incidence high enough to dramatically increase the distance traveled through the cover glass, which ranges in thickness from about 0.8 mm for the Leica M8 to 2 mm for the Sony A7 series. This has the effect of positive field curvature, in addition to causing other problems with vignetting and parallax with the Bayer filter. Retro-focus lenses, designed for SLR cameras, have about 1-1/8" more back focus distance, and show little effect due to cover glass thickness, less vignetting and minimal color shifts.</p>

<p>Straying off topic a bit, lenses designed specifically for the A7 seem to include the thick cover glass in the optical design, since it is consistent even though separate from the lens itself. Zeiss Loxia and Batis lenses, for example, are extremely sharp to the extreme corners on the A7, but show evidence of poor corner performance in MTF charts, which are made without any cover glass or sensor in place (virtual image measurements).</p>