Digital Infrared and Flare, Sharpness, and Detail

[danielleetaylor]

<p>I recently purchased an OEC IR 720 infrared filter off of ebay. It's cheap, I know, but I wanted to experiment before considering the small fortune for a Hoya R72 or a LifePixel conversion of my oldest DSLR, my 10D.</p>

<p>My fear was that the filter wouldn't work at all, but it does seem to work and show nice IR effects. However, I'm encountering the following issues, and I'm curious to hear from those who have tried them if a more expensive filter, a different lens, and/or LifePixel conversion would solve the problems. Or if I just need to improve my technique.</p>

<p>* I'm noticing pretty significant flare. I don't know if I should blame the filter, or the simple fact that a filtered DSLR needs 15-45s @ f/16 in full sunlight to make an image. I know that long exposures at night also produce flare I never see with normal exposures, so I'm curious to know if anyone with an R72 or converted DSLR experiences significant flare.</p>

<p>* Sharpness and detail really seem to suffer with the filter in place. I have some old Kodak HIE shots and quite frankly I had the same experience then. Back then I blamed the nature of the film. Now that I'm seeing the same thing with digital I'm curious to know if it's due to the nature of IR, or in this case the cheap filter, or my 17-40L not being ideal for IR photography. Another possible problem is my focus point, and I'm wondering what procedure others follow in determining optimum IR focus.</p>

<p>* I should note that corner sharpness suffers the most, almost as if different wavelengths are not registering at the same location. By the extreme corners there's almost a double edge effect. But the center seems OK, at least no worse than Kodak HIE was. I never shot HIE with a 17-40, and I'm wondering if the corner issues are indeed due to lens performance in the IR range. Incidentally, my 17-40 has very good corner performance in normal shots.</p>

<p>Basically what I'm getting at is that if I manually clone or burn out the flare in PS, and print to 8x10, the shots seem fine. But I'm trying to decide what I need in the future for higher quality and larger print sizes. Quality IR filters are so expensive that I'm leaning towards a LifePixel conversion. But I would love input from others who have done IR work as to the best choices and as to what I can expect in terms of IQ.</p>

<p>Thanks!</p>

<p> </p>

[danielleetaylor]

<p>Here's a test shot using the OEC filter and 17-40.</p><div></div>

[phule]

<p>Looks like internal reflections. The bridge is mirrored in the sky above.</p>

[bobatkins]

<p>IR image quality is low due to the fact that once you have the IR filter in place in the worst case only the red pixels contain significant intensity information and that means only 1.5MP operate in a typical 6MP sensor. Resolution drops, noise goes up. This varies a bit from camera to camera. With some you also get some IR response in the blue channel. However the green channel (which has 50% of the pixels) is typically pretty blind to IR<br>

The long exposures also mean that any movement in trees, grass etc. blurs the image.</p>

<p>Some lenses flare due to the AR coating not being very effective at IR wavelengths. See <a href="http://lensplay.com/lenses/lens_infra_red_IR.html">http://lensplay.com/lenses/lens_infra_red_IR.html</a> for a list of known "good" and "bad" lenses for IR work. The 17-40/4L is actually one of the better lenses for IR work.</p>

[danielleetaylor]

<p>Thanks for the info Bob. What about corner sharpness? Is it possible that various wavelengths above 720nm don't focus correctly into the corners, a form of IR chromatic aberration? I would swear that's what I'm seeing. (I can post a corner crop later. I've unplugged my notebook from my external drives.)</p>

[craig_meddaugh]

<p>That is quite likey a correct supposition. Lens are designed to achieve the correct convergence of the visible light spectrum. Also, I am not sure if you are aware of this, but typically the focus point for IR is a bit different than that for visible. I don't recall the 17-40 having the IR indices on the distance scale, but it is a pretty common lens and the internet should shed some more light on the proper offset for IR work. Good luck</p>

[JDMvW]

<p>Some lenses flare, as already pointed out. Some information on which lenses at (<a href="http://folk.uio.no/gisle/photo/ir.html">link</a> ). [this is an old link, but has a list of lenses that I can't find in its new location, so while it lasts, stay at the old site]</p>

<p>ps. here is the new link with the lens list, found it after all (<a href="http://dpanswers.com/irphoto_lenses.html">link</a> )</p>

[steven_f1]

<p>One thing to keep in mind with IR is that it will not focus at the same point as visible light. In other words if you focus the camera (manual or auto) and then put on the filter. the image will be slightly out of focus. A small focus adjustment is typically needed. There may be some information on this in your lens manual. </p>

<p>Other than that it would appear that the image is a little over exposed. Reducing the exposure may reduce the flare and internal reflections. </p>

[eric_boutilier_brown1]

<p>I am currently on my fourth IR DSLR (Sigma SD10, Nikon D70, D80, Canon 5D MKII), with three of the four being converted by LifePixel, so my comments are based on converted cameras, not using a filter.<br>

* I'm noticing pretty significant flare.<br>

I don't get any more flare with the IR DSLRs than with a normal DSLR, but what I do have issues with (especially with apertures like f/16) is hotspots - miscoloured (in colour IR image) areas of lighter density in the centre of the frame. This occurs with some lenses more than others, and more often below f/8 - so with IR images, I don't work below f/8 (and you should keep in mind, with the 10D, f/16 will cause you some significant loss of image quality due to diffraction).<br>

* Sharpness and detail really seem to suffer<br>

My IR images are significantly sharper than images made with my non-converted camera - I've assumed with is the lack of the AA filter, but also, all three colours (RGB) are seeing a narrow wavelength of light, leading to a more even focus distance (unlike normal cameras, where often Red is mis-focused slightly, except in high quality or APO lenes. This could also be a focus error, as IR will NOT AF in most cases with most lenses, so focus testing is required. With the 5D MKII I use LiveView to focus, and it is a godsend. Also, in regard to sharpenss, f/16 would cause loss of sharpeness due to diffraction (see above).<br>

* I should note that corner sharpness suffers the most<br>

Can't comment on this - might be a reflection off the AA filter perhaps? Never seen this.<br>

Hope some of the above helps!<br>

More info about my experience using IR DSLRs can be found on my website http://evolvingbeauty.com/index.php/evolving/techPost/infrared_basics/</p>

[allardk]

<p>Hi Daniel,</p>

<p>I have no experience with infrared photography, but from what I know from lens design the problems you note are not unexpected. Lenses are full of compromises, and usually optimized for green, with red and blue hopefully not too far behind. The same goes for coatings. There are some specialized lenses in the industrial vision area that are also corrected for IR, <a href="http://www.edmundoptics.com/onlinecatalog/displayproduct.cfm?productID=3138">like this one</a> for example, but I don't think there are any for DSLR's. If there are, they'll probably be mighty expensive.</p>

<p>Good point also noted earlier about the RGB sensors only very partially being sensitive to IR. And (least of your worries, but just for completeness) of course the diffraction limit is larger for longer wavelengths.</p>

<p>Allard</p>