inksandpaper
Offline
Upload & Sell: Off
|
 Re: Official: Voigtlander 35mm f/2 APO-Lanthar announced! | |
realVivek wrote:
Visible starts (by most accepted conventions) at 400nm. 410 is visible violet.
Here is the transmission spectrum I promised earlier.
.. which has a hard cut around 410nm (which by my understanding is within the visible spectrum, I guess you could call it near UV spectrum.)
Thanks for the information @realVivek! Do you happen to know if the UV cut filter for this Sony you measured is based on absorptive or interference coatings?
BTW I have not noticed any magenta shift off-axis on any of my lenses, no matter how I looked, with or without diffuser (similar to expodisc) filter, with or without UV filter in front of the lens, in all kinds of low and high UV sky exposures, even when deliberately testing and trying to look for it, with or without camera "Lens Comp. Shading" on and off, with the same lens that @tsdevine uses. But I can easily notice it in his samples and the new Cosina samples of the new 35mm. This is quite disturbing.
I am trying to understand how it might work. If the UV filter on the sensor is the interference type, the effectiveness is usually tuned to be maximal when the incident ray angle is orthogonal to the filter surface, and the effect is diminished as the ray angle becomes more oblique, which it will off axis, when the exit pupil is nearer to the focal plane, such as with wide angles of non-telecentric design. The BSI sensors of newer Sonys are greatly resistant to lens/sensor cast from oblique exit ray angles but it does not help when the sensor filter stack is transmitting light of a different colour. As the incident angle to the filter becomes more oblique, interference coatings become less effective and the reflected light colour shifts towards the longer wavelengths, i.e. reflected UV becomes reflecting more blue and more cyan light, and thus the transmitted light, having been filtered of that colour, becomes shifted to the opposite colour, which is more reddish or magenta. That's the magenta shift that has been noted.
I suspect this is the true nature of what is happening, though I've not yet seen any other explanation like mine. This is predicated of course, on the assumption that the UV filter is indeed using interference-based coatings. I happen to know, with good authority, that the IR filters (there are two), on the other hand, are using both absorptive and interference-based coatings. The interference-based IR filter should shift to longer wavelengths of reflected light also as the ray angle gets more oblique, which means more red (and far IR) light gets reflected, and thus more blue light is transmitted. Hot filters typically have a bluish tint, I gather.
The effects seem to mostly offset each other in my tests, since I see that the colour distribution from on to off-axis is mostly very neutral, and if anything, the shift has been 1 or two points towards blue/cyan, rather than red/magenta. Interference coatings are difficult to apply consistently, and in all likelihood, the effectiveness of each do not balance each other out perfectly for some batches of filters. It may be that some batches of cameras are more susceptible to colour shift as a result. I honestly don't know if that's really the case. Others more intelligent than I might be able to offer a better guess.
|
