Beyond Visible
Dealing with the Infrared Contamination of Ultraviolet Images

Shane Elen

What is Infrared Contamination of Ultraviolet Images
A reflected ultraviolet image should exhibit contrast resulting only from ultraviolet light and structural effects. Contrast resulting from other sources is undesirable as it can attenuate or mask the ultraviolet contrast. The presence of infrared light in an ultraviolet image acts in this manner and therefore needs to be eliminated or significantly reduced. The cause is often two fold, the camera CCD and the ultraviolet transmission/visible blocking filter attached to the lens.

Infrared free ultraviolet image
Infrared free, ultraviolet image
Infrared contaminated, ultraviolet image
Infrared contaminated, ultraviolet image
Infrared image
Infrared image

The Dandelion - A Common Reference for Reflected Ultraviolet Photography
The common dandelion is often used as a reference for photographers experimenting with reflected ultraviolet photography. This flower provides the UV photographer with a quick and easy reference with which to test cameras, lenses, filters and techniques for ultraviolet imaging. The following reflectance spectra of a dandelion flower exhibits a distinct difference in ultraviolet reflectivity between the central and outer regions of the flower. The central region of the flower strongly absorbs ultraviolet and blue, and would therefore appear black in an ultraviolet reflectance image. In contrast, the outer region strongly absorbs blue but reflects some ultraviolet. It is also important to note that these two regions, while exhibiting differing UV reflectance, display similar infrared reflectance. This is beneficial for the purpose of demonstrating the issue of infrared contamination of ultraviolet images with various filter combinations.

Reflectance Spectra of the Common Dandelion
Reflectance spectra of the outer and central regions of the common dandelion. This shows that these two regions exhibit similar visible and infrared reflectance, but dissimilar ultraviolet reflectance.

The Inherent Problem of Ultraviolet Digital Photography
Unlike the B&W T-Max film often used for ultraviolet photography, the CCD of a digital camera is inherently susceptible to infrared contamination, often despite the manufactures attempt to reduce the problem through the use of an infrared blocking filter on top of the CCD. As these filters do not have a sharp cut-off at infrared wavelengths there has to be a compromise in order not to exclude visible red wavelengths, this results in some infrared being allowed to pass. Although minor infrared contamination is generally not too much of a problem for visible images, it is a significant problem for reflected ultraviolet images because the CCD is far more sensitive to infrared light than it is to ultraviolet. It can be double trouble for some digital cameras because a significant amount of ultraviolet light may actually recorded in the red channel e.g.Nikon D70. It might appear that fitting an ultraviolet transmission/visible blocking filter onto the front of the lens would prevent infrared from reaching the camera CCD, however this is often not the case. The second image (fig.2), using an infrared flash to demonstrate the point, shows the ineffectiveness in preventing infrared from reaching the camera CCD when a UV lens is fitted with a Tiffen 18A ultraviolet pass filter. More importantly, this is an infrared image captured through an ultraviolet pass filter! A similar infrared image can also be obtained using most long-wave ultaviolet lamps (LWUV) as these also generate and pass some infrared, however much longer exposures are required.

Visible image of a dandelion taken inside by natural window lighting
Fig.1. Visible image of a dandelion taken by natural window lighting
Infrared image of a dandelion taken through a Tiffen 18A ultraviolet filter
Fig.2. Infrared image of a dandelion taken through a Tiffen 18A ultraviolet filter using a UV modified Nikon SB-14 but fitted with a custom wavelength SW5-IR infrared adapter (700nm cut-off) i.e. infrared light only

In order to control the lighting environment, the dandelion images portrayed here were shot indoors in the dark using a Nikon D70, 105 UV-Nikkor and a UV enhanced Nikon SB-14. The focus was set once for ultraviolet (figure 12) and not adjusted for subsequent shots, regardless of filter stack thickness etc.. All images were captured at 800ASA, f5.6 except figure 12 which was set to f8 to prevent overexposure. These images are not meant to be artistic but were shot only for demonstration purposes. Filters attached to the lens or the flash are indicated on each image. Other than noise correction and a little sharpening, they are basically as shot.

How does Infrared get past the Ultraviolet Transmission/Visible Blocking Filter
The ultraviolet transmission filters commonly used on a lens for reflected ultraviolet often have two regions of light transmission, one in the ultraviolet and one in the infrared.

Kodak 18A Ultraviolet Transmission Filter
Fig.3. Kodak 18A Ultraviolet Transmission Filter (14% IR leakage)
Hoya U-330 Ultraviolet Transmission Filter
Fig.4. Hoya U-330 Ultraviolet Transmission Filter (50% IR leakage but it does transmits SWUV)

When an infrared emitting source, such as the common xenon flashlamp, is fitted with a standard UV transmission/visible blocking filter it will still transmit some infrared, e.g. Nikon SB-140 fitted with an SW5-UV flash adapter. Although the ultraviolet output now exceeds the infrared output almost two to one, the "ultraviolet" light still contains a significant amount of infrared, some of which is transmitted by the filter and may prevent subtle ultraviolet patterns from being recorded.

Nikon SW-5UV Ultraviolet Flash Adapter
Fig.5. Transmission characteristics of the Nikon SW-5UV Ultraviolet flash adapter exhibiting both ultraviolet and infrared
Ultraviolet image captured with Tiffen 18A and Nikon SW-5UV
Fig.6. Ultraviolet image exhibiting a slightly washed out ultraviolet pattern, captured with Tiffen 18A and a UV enhanced Nikon SB-14 fitted with a Nikon SW-5UV.

Some Filter Solutions for Reducing Infrared Contamination
The common solution generally requires the use of two filters, an ultraviolet transmission/visible blocking filter and an ultraviolet transmission/infrared blocking filter. Some combinations are more effective than others, such as the Tiffen 18A in combination with BG40. The Tiffen Hot Mirror is ofen mistakenly used in combination with an 18A (or similar filter) to prevent infrared contamination. Unfortunately this combination isn't very effective at all because it passes infrared while attenuating ultraviolet.

Transmission spectra of the Tiffen 18A and Tiffen Hot Mirror
Fig.7. The transmission spectra tell the tale - the Tiffen Hot Mirror does not prevent the infrared below 750nm from passing through the Tiffen 18A but it does prevent most of the ultraviolet! UNACCEPTABLE SOLUTION but can be salvaged by using Photoshop Channels (see below)
Ultraviolet image captured with Tiffen 18A, Tiffen Hot Mirror and Nikon SW-5UV
Fig.8. Ultraviolet image exhibiting a severely washed out ultraviolet pattern, captured with Tiffen 18A, Tiffen Hot Mirror and a UV enhanced Nikon SB-14 fitted with a Nikon SW-5UV. This is actually worse than just using the Tiffen 18A by itself because the Hot Mirror passes very little ultraviolet while freely transmitting infrared.

In the example below I have substituted a BG38 in place of the BG40 as I have misplaced the latter, however they are very similar in their characteristics.

Tiffen 18A and Schott BG38
Fig.9. Transmission characteristics of the Tiffen 18A and Schott BG 38 exhibit very little possibility of infrared leakage while maintaining high ultraviolet transmission when used in combination
Ultraviolet image captured with a Tiffen 18A, BG38 and Nikon SW-5UV
Fig.10. Ultraviolet image exhibiting a stronger ultraviolet pattern, captured with Tiffen 18A, BG38 and a UV enhanced Nikon SB-14 fitted with a Nikon SW-5UV. Clearly an improvement over the sole use of the Tiffen 18A.

The last filter has a high ultraviolet transmission of 75% at 355nm and a range from 300-400nm, with only minimal infrared transmission around 0.5%. The advantages are high ultraviolet transmission, insignificant infrared leakage, only one filter is required, resulting in improved depth of field/brightness.

Ultraviolet image captured with a Baader Venus Filter and Nikon SW-5UV
Fig.11. Transmission characteristics of the Baader Venus Filter exhibit a high ultraviolet transmission and only 0.5% infrared light leakage.
Baader Venus Filter and Nikon SW-5UV
Fig.12. Ultraviolet image exhibiting a strong ultraviolet pattern, captured with a Baader Venus Filter and a UV enhanced Nikon SB-14 fitted with a Nikon SW-5UV. A smaller aperature was required to prevent over exposure but this still resulted in a brighter image as well as an increased depth of field.

Summary of Filter Combinations
These spectra show the effect on the transmission properties when an ultraviolet transmission and infrared blocking filter are combined.

Tiffen 18A & Tiffen Hot Mirror Combination
Fig.13. Tiffen 18A & Tiffen Hot Mirror Combination
Hoya U-330 & Tiffen Hot Mirror Combination
Fig.14. Hoya U-330 & Tiffen Hot Mirror Combination

The Tiffen 18A and Hot Mirror combination produces three transmission regions, 310-355nm, 370-400nm and 700-765nm. It is important to note that the ultraviolet transmission has changed from single broad band at 60% (fig.7) to two small regions of 15 and 20%, and the infrared transmission from 14% to 4%. Integrating the area under the curves show that the Uv to IR ratio changes from 4.5:1 for the 18A filter to 3.8:1 for the combination of filters, indicating a change for the worse with regards to infrared contamination. The situation for the Hoya U-330 and Tiffen Hot Mirror is obviously even more severe.

Tiffen 18A & Schott BG38 Combination
Fig.15. Tiffen 18A & Schott BG38 Combination
Hoya U-330 & Schott BG38 Combination
Fig.16. Hoya U-330 & Schott BG38 Combination

The Tiffen 18A and Schott BG38 combination shows an infrared free spectra however the ultraviolet transmission of this combination has dropped from 60% (fig.7) down to 40%. The U-330 filter used in combination with the Schott BG38 results in a strong ultraviolet transmission around 60% with only minimal infrared leakage (<0.5%), minimal blue leakage around 460nm (0.5%) and a small region from 400-420nm (8% @ 400nm). With a few minor trade offs, these filter combinations compare favorably with the Baader Venus filter with regards to purity of ultraviolet light however neither come close to the 75% transmission exhibited by the Venus filter.

A Software Solution for Reducing Infrared Contamination
This method is dependent upon how the CCD handles ultraviolet light with respect to the RGB colour channels. The following method pertains to the Nikon D70 however not all digital cameras respond the same way. In testing the Nikon D70 for ultraviolet capability it was observed from the histograms that the blue channel was dominant for ultraviolet light between 395nm and 400nm, whereas the red channel became dominant for 390nm and below. In general the green channel was the weakest channel, however between 380nm and 370nm it contained enough signal to be useful. Useful for what? In situations where infrared is contaminating the ultraviolet image it might be possible (Nikon D70) to exclude the red channel and improve the ultraviolet contrast by using either the blue and/or green channel data only. Of course this will only work if the sample actually reflects ultraviolet in this range, otherwise the channel will have little ultraviolet data in it - check the histogram.

The ultraviolet image of the dandelion captured with an 18A filter in combination with a Tiffen hot mirror (fig. 8, 17 and 13) is a good case for using the blue channel data only. In this image, the minimal amount of ultraviolet occuring in the region below 355nm will occur primarily in the red channel along with much of the infrared from the 700-765nm region, so this channel can be considered unusable. The red channel can therefore be excluded (Photoshop) without significant loss of ultraviolet data. The region 370-400nm provides useful data in both the blue and the green channel. However infrared contamination occurs primarily in the red channel, followed by the green channel, and lastly the blue channel. Therefore, in this case, the blue channel is likely to exhibit better ultraviolet contrast than the green channel. Using only the blue channel (fig. 18) and excluding the red channel (infrared) and the green channel (infrared, ultraviolet and noise), permits the ultraviolet contrast to be somewhat salvaged from the original infrared contaminated image.

Ultraviolet image captured with Tiffen 18A, Tiffen Hot Mirror and Nikon SW-5UV
Fig.17. Ultraviolet image exhibiting a severely washed out ultraviolet pattern, captured with Tiffen 18A, Tiffen Hot Mirror and a UV enhanced Nikon SB-14 fitted with a Nikon SW-5UV. This is actually worse than just using the Tiffen 18A by itself because the Hot Mirror passes very little ultraviolet while freely transmitting infrared.
Ultraviolet image using the blue channel data in Photoshop
Fig.18. Ultraviolet image extracted from fig. 17 by using only the blue channel data in Photoshop and excluding the green and red channel. Although some noise and loss of detail has occured the overall result is not bad considering the small amount of UV light actually passed by this filter combination.

Red channel
Fig.19. Red channel
Green channel
Fig.20. Green channel
Blue channel
Fig.21. Blue channel

In the image of the dandelion captured with the 18A in combination with the BG38 filter (fig. 10 & 22), the ultraviolet transmission peak occurs around 370nm and so the green channel is likely to contain a lot of ultraviolet data whereas the blue channel will only contain a minimal amount. However, the filter/camera combination permits ultraviolet transmission down to 340nm so it is likely that the red channel will contain significant ultraviolet data in addition to the few percent of infrared that is passed by this filter combination. These predictions can be confirmed by observing the ultraviolet contrast exhibited by the dandelion in the red channel (fig. 23), blue channel (fig.24) and green channel (fig.25). In this case, excluding the red channel will result in significant ultraviolet contrast loss and would not be recommended.

Ultraviolet image captured with a Tiffen 18A, BG38 and Nikon SW-5UV
Fig.22. Ultraviolet image exhibiting a stronger ultraviolet pattern, captured with Tiffen 18A, BG38 and a UV enhanced Nikon SB-14 fitted with a Nikon SW-5UV.
Ultraviolet image using the red channel data in Photoshop
Fig.23. The red channel data from fig. 22 shows a reasonably good ultraviolet pattern in the flower indicating that it is making a significant contribution to the ultraviolet image and shouldn't be excluded.

Ultraviolet image captured with a Tiffen 18A, BG38 and Nikon SW-5UV
Fig.24. The blue channel data from fig. 22 shows a noisy weak ultraviolet pattern in the flower.
Ultraviolet image using the red channel data in Photoshop
Fig.25. The green channel data from fig. 22 shows a reasonably good ultraviolet pattern in the flower.

Back to Homepage

© Shane Elen 2006. Last updated Jul 27th, 2007.