Code
Rfr_from_n(angle_deg = 0, n = c(1.51, 1.54))
[1] 0.04128506 0.04519809
Differences in the transmission spectrum
Pedro J. Aphalo
2021-02-01
2023-04-14
filters, near infrared, spectra
I have updated this page when transferring it to Quarto. I changed the plots to be built in R within the R markdown source file. As a result the page now also includes the listing of the R code used to create all plots.
All the near-infrared (NIR) long-pass filters described in this post are absorptive glass filters and sold for special photography effects. All modern photography digital cameras have internal UVIR-cut filters. Use of these filters requires special film, special digital cameras or modified DSLR or DML cameras. Some filters can be used with unmodified digital cameras with very long exposure times. Few normal camera objectives are designed to work well in NIR. Two problems are common: 1) focus shift which needs special attention when focusing is done in visible light (e.g. not using an EVF or live-view from the camera sensor, 2) reflections causing “hot spots”. Lately some anti-reflection coated (AC) and multi coated (MC) NIR long-pass filters have been advertised as helping reduce the problem of hot spots (sold by Kolari Vision). I have not tested these, as I do not even have one of them. However, some other NIR filters seem to have some type of anti reflection coating based on their spectrum. See the post on UV-cut filters for other examples of spectra for filters with different anti-reflection coatings. Taking into account the reflection on both surfaces of the filter, uncoated optical glass will loose about 10% of transmittance on near normal (at 90 degrees to the surface) illumination. The exact value can be computed from the difference in refractive index between air and glass, while the refractive index of the glass depends to some extent on its composition.
The reflection at a glass-air interface can be computed from the difference in refractive index. In filters and glass plates we need to consider two interfaces. Reflectance, the fraction of light reflected, is at its minimum when the light beam is normal (at 90 degrees) to the surface. The refractive index depends on the composition of the glass, especially in the case of glass types used for lenses.
If we consider an ordinary BK7 glass plate (n = 1.51 at 900 nm and n = 1.54 at 350 nm), and a light beam with no deviation from normal, for a single air interface reflectance varies between 4.5% in the UV-A to 4.1% in the near infrared. Or very roughly, between 8% and 9% for a plate.
Anti-reflection coatings based on thin films modify the properties of the glass-air interfaces. While a single anti-reflection (AR) coating on both surfaces can decrease total reflectance to approximately 1% multi coating (MC) can reduce it to less than 0.1%. Based on this we can guess from the spectra in the figures below, which filters have been coated to reduce reflections (are AR coated) and how effective these coatings are at different wavelengths. For this we need to assume that at some wavelengths the filter glass is fully transparent (internal transmittance near the ideal 100%) and the decrease in transmittance is caused by reflections.
The precision of the optical polishing and lapping and anti-reflection coating are the properties that will determine whether the filters will degrade image quality or not. Imperfections in the optical polishing are most visible when panning while recording video. With respect to mechanical protection the hardness of the glass and possibly also of the metal in the filter frame (brass vs. aluminium) may be relevant. The thickness of optical glass filters used in photography is usually between 1 and 3 mm, except for stacks. Some filters have an additional “nano” coating that helps repel dirt making them easier to keep clean.
The most common cut-on wavelength for “IR” filters used for photography is 720 nm, so we start be comparing filters with this characteristic, at least nominally. The Haida IR720 is the only one of the NIR filters in this figure to be advertised as anti-reflection coated. It seems to use a coating optimized for visible light, very similar to that used in Hoya’s 25A HMC (see ). As transmittance is well above 95% for the Hoya R72 we must conclude that this filter has an AR coating, possibly a single coating optimized for NIR as the high transmittance can be observed all the way to the range of measurement at 1100 nm.
Of the filters measured, only the ones from Heliopan have brass rings. The Orange and Yellow filters from Heliopan and Hoya are multi-coated and reflect negligible amounts of visible light with a normal beam. The filters from UQG, Tangsinuo and Purshee are not AR coated.
autoplot(filters.mspct[c("UQG_GG400_3mm",
"Purshee_JB450_2.1mm_30.5mm",
"Tangsinuo_JB450_2.0mm_52mm",
"Heliopan_Yellow_5_SH_PMC_2.3mm_52mm",
"Hoya_Y_(K2)_HMC_2.3mm_52mm",
"Heliopan_Orange_22_SH_PMC_2.2mm_52mm")],
annotations = list(c("+", "wls", "boundaries"), c("-", "peaks")),
pc.out = TRUE,
facets = 1L)
The RG665 from Heliopan is AR coated, possibly even MC, although not advertised as such. The RG695 and RG780 are not AR coated. The coatings of the Orange and Yellow filters seems to be optimized for visible light, while the Heliopan RG665 has a coating optimized for NIR.
A collection of cheaper filters from Zomei, none of them AR coated and with rather large variation in cut-on wavelength between batches, here visible comparing filters of the same specification except for size:
autoplot(filters.mspct[c("Zomei_IR680_2.1mm_30.5mm",
"Zomei_IR680_2.1mm_52mm")],
annotations = list(c("+", "wls", "boundaries"), c("-", "peaks")),
pc.out = TRUE,
facets = 1L)
autoplot(filters.mspct[c("Zomei_IR720_2.1mm_30.5mm",
"Zomei_IR720_2.0mm_72mm")],
annotations = list(c("+", "wls", "boundaries"), c("-", "peaks")),
pc.out = TRUE,
facets = 1L)
autoplot(filters.mspct[c("Zomei_IR760_2.1mm_30.5mm",
"Zomei_IR760_2.0mm_52mm")],
annotations = list(c("+", "wls", "boundaries"), c("-", "peaks")),
pc.out = TRUE,
facets = 1L)
autoplot(filters.mspct[c("Zomei_IR850_2.1mm_52mm", "Zomei_IR950_TTmm_52mm")],
annotations = list(c("+", "wls", "boundaries"), c("-", "peaks")),
pc.out = TRUE,
facets = 1L)
The KolariVision Gen II filters are missing from this article as I do not have any of them. According to the data provided by KolariVision, they have a very effective AR coating optimized for NIR.
As for cost, taking as example the filters with cut-on near 720 nm and a size of 58 mm, the Zomei is the cheapest at less than 15 € (on never-ending sale), the Haida is at 30 €, the Hoya at 65 € and the Heliopan at 75 €, while the Kolari Vision at 85 USD, would cost over 100 € after delivery to Europe and VAT. In AliExpress one can find even cheaper filters, at around 10 €.
All the spectra shown above in figures have been measured by the author with the same Agilent/Hewlett-Packard diode array spectrophotometer (HP 8453). The data are available in R package ‘photobiologyFilters’. The figures were produced with R packages ‘ggspectra’, which is an extension to package ‘ggplot2’. All these packages are open-source and available through CRAN. See also the R for photobiology web site.
The code is included in this page, but “folded”. To display the code used to produce the figures and compute reflectance from the refractive index click on the word Code or the triangle before it.